james.miller@research.ge.com
(518) 387-4005, Dial Comm: 8*833-4005,
Cell: (518) 505-7065, Fax: (518) 387-6981
------_=_NextPart_001_01C31010.5BBD3B18
Content-Type: text/html;
charset="iso-8859-1"
Rational Purify and Quantify
require the linker flags
/incremental:no
/fixed:no
If I put /fixed:no on any of
the CMAKE_*_LINKER_FLAGS, it get ignored by the .NET generator. I grep'ed all
the *.vcproj files and /fixed:no is missing in all the
files.
To manually add the options in
.NET, I go the properties page for the executable, and under "Linker"."Command
line", I added /fixed:no to the Additional Options box. The setting gets
written in the vcproj file
<VisualStudioProject>
<Configurations>
<Configuration Name="Debug">
<Tool Name="VCLinkerTool"
AdditionalOptions="/MACHINE:I386 /fixed:no"/>
</Configuration>
<Configurations>
</VisualStudioProject>
Can the .NET generator be set
up so that one of the CMAKE_*_LINKER_FLAGS settings gets put in the
AdditionalOptions section? Or has this been fixed already?
Jim Miller
_____________________________________
Visualization &
Computer Vision
GE
Research
Bldg. KW, Room C218B
P.O. Box 8, Schenectady NY
12301
millerjv@research.ge.com
james.miller@research.ge.com
(518) 387-4005, Dial
Comm: 8*833-4005,
Cell: (518) 505-7065,
Fax: (518) 387-6981
------_=_NextPart_001_01C31010.5BBD3B18--
From bill.hoffman@kitware.com Thu May 1 20:20:55 2003
From: bill.hoffman@kitware.com (Bill Hoffman)
Date: Thu, 01 May 2003 15:20:55 -0400
Subject: [Cmake] .NET Linker options for Rational tools
In-Reply-To:
Message-ID: <5.2.0.9.0.20030501152010.04e6fe18@pop.biz.rr.com>
It is on the board to be fixed, but it has not yet been done.
One way to do it, would be to use the nmake Makefiles.
-Bill
At 02:34 PM 5/1/2003, Miller, James V (Research) wrote:
>Rational Purify and Quantify require the linker flags
>
>/incremental:no
>/fixed:no
>
>If I put /fixed:no on any of the CMAKE_*_LINKER_FLAGS, it get ignored by the .NET generator. I grep'ed all the *.vcproj files and /fixed:no is missing in all the files.
>
>To manually add the options in .NET, I go the properties page for the executable, and under "Linker"."Command line", I added /fixed:no to the Additional Options box. The setting gets written in the vcproj file
>
>
>
>
>
>
>
>
>
>Can the .NET generator be set up so that one of the CMAKE_*_LINKER_FLAGS settings gets put in the AdditionalOptions section? Or has this been fixed already?
>
>
>
>Jim Miller
>_____________________________________
>Visualization & Computer Vision
>GE Research
>Bldg. KW, Room C218B
>P.O. Box 8, Schenectady NY 12301
>
>millerjv@research.ge.com
>
>james.miller@research.ge.com
>(518) 387-4005, Dial Comm: 8*833-4005,
>Cell: (518) 505-7065, Fax: (518) 387-6981
>
>
>
From rmathews@envoyww.com Thu May 1 21:49:24 2003
From: rmathews@envoyww.com (Robert Mathews)
Date: Thu, 1 May 2003 16:49:24 -0400
Subject: [Cmake] .NET Linker options for Rational tools
References: <5.2.0.9.0.20030501152010.04e6fe18@pop.biz.rr.com>
Message-ID: <000b01c31023$277a92b0$c3d3403f@victory>
Edit your CMakeCache.txt file, find the lines that say:
CMAKE_EXE_LINKER_FLAGS
CMAKE_SHARED_LINKER_FLAGS
and add the "/incremental:no /fixed:no" flags.
Regenerate your projects and that should do it. This may still not appear
in the .NET gui, since the flags won't be in the right part of the .vcproj
file, but they should get take effect when project is built.
BTW, almost all the compile flags and link flags can't be added to .NET. For
example, any compile flag other than "-D" is silently tossed by CMake. You
can however add those flags directly to the appropriate defnitions in
CMakeCache.txt, which is ok provided that you don't mind them being global.
Hope that helps.
Rob.
----- Original Message -----
From: "Bill Hoffman"
To: "Miller, James V (Research)" ; "CMake developers
(E-mail)"
Sent: Thursday, May 01, 2003 3:20 PM
Subject: Re: [Cmake] .NET Linker options for Rational tools
> It is on the board to be fixed, but it has not yet been done.
> One way to do it, would be to use the nmake Makefiles.
>
> -Bill
>
>
> At 02:34 PM 5/1/2003, Miller, James V (Research) wrote:
> >Rational Purify and Quantify require the linker flags
> >
> >/incremental:no
> >/fixed:no
> >
> >If I put /fixed:no on any of the CMAKE_*_LINKER_FLAGS, it get ignored by
the .NET generator. I grep'ed all the *.vcproj files and /fixed:no is
missing in all the files.
> >
> >To manually add the options in .NET, I go the properties page for the
executable, and under "Linker"."Command line", I added /fixed:no to the
Additional Options box. The setting gets written in the vcproj file
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >Can the .NET generator be set up so that one of the CMAKE_*_LINKER_FLAGS
settings gets put in the AdditionalOptions section? Or has this been fixed
already?
> >
> >
> >
> >Jim Miller
> >_____________________________________
> >Visualization & Computer Vision
> >GE Research
> >Bldg. KW, Room C218B
> >P.O. Box 8, Schenectady NY 12301
> >
> >millerjv@research.ge.com
> >
> >james.miller@research.ge.com
> >(518) 387-4005, Dial Comm: 8*833-4005,
> >Cell: (518) 505-7065, Fax: (518) 387-6981
> >
> >
> >
>
>
>
> _______________________________________________
> Cmake mailing list
> Cmake@public.kitware.com
> http://public.kitware.com/mailman/listinfo/cmake
>
From rmathews@envoyww.com Thu May 1 22:04:54 2003
From: rmathews@envoyww.com (Robert Mathews)
Date: Thu, 1 May 2003 17:04:54 -0400
Subject: [Cmake] Handling precompiled headers compiler flag for VC7
References: <5.2.0.9.0.20030501152010.04e6fe18@pop.biz.rr.com> <000b01c31023$277a92b0$c3d3403f@victory>
Message-ID: <001d01c31025$518ceb50$c3d3403f@victory>
I guess every "/Foo" compiler flag for VC7 has to be handled differently
..... ugly.
Here's the diff for handling one of the precompiled header flags && allowing
control of "Detect64BitPortabilityProblems".
Hope that's useful.
Index: cmLocalVisualStudio7Generator.cxx
===================================================================
RCS file: /cvsroot/CMake/CMake/Source/cmLocalVisualStudio7Generator.cxx,v
retrieving revision 1.18
diff -r1.18 cmLocalVisualStudio7Generator.cxx
341a342,362
>
> // Handle the precompiled header flag specially, since VC .NET needs
that.
> std::string aCompilerFlags = m_Makefile->GetDefineFlags();
> std::string::size_type aPrecompiledOptionIndex =
aCompilerFlags.find("/YX");
> if(aPrecompiledOptionIndex != std::string::npos)
> {
> fout << "\t\t\t\tUsePrecompiledHeader=\"2\"\n";
> std::string aHeaderName;
> std::string::size_type aNextIndex = aCompilerFlags.find(" ",
aPrecompiledOptionIndex);
> if(aNextIndex > aPrecompiledOptionIndex+3)
> {
> std::string::size_type aLength = (aNextIndex == std::string::npos) ?
std::string::npos :
> aNextIndex - (aPrecompiledOptionIndex+3);
> aHeaderName =
aCompilerFlags.substr(aPrecompiledOptionIndex+3,aLength);
> if(*aHeaderName.begin()== '"')
> aHeaderName.erase(aHeaderName.begin());
> if(*aHeaderName.rbegin() == '"')
> aHeaderName.erase(aHeaderName.size()-1);
> fout << "\t\t\t\tPrecompileHeaderThrough=\"" << aHeaderName << "\"\n";
> }
> }
348,349c369,375
< fout << "\t\t\t\tDetect64BitPortabilityProblems=\"TRUE\"\n"
< << "\t\t\t\tDebugInformationFormat=\"3\"";
---
> if(m_Makefile->GetDefinition("DETECT64BITPORTABILITYPROBLEMS"))
> fout << "\t\t\t\tDetect64BitPortabilityProblems=\"" <<
> m_Makefile->GetDefinition("DETECT64BITPORTABILITYPROBLEMS") <<
"\"\n";
> else
> fout << "\t\t\t\tDetect64BitPortabilityProblems=\"TRUE\"\n";
>
> fout << "\t\t\t\tDebugInformationFormat=\"3\"";
From jan.fischer22@epost.de Fri May 2 11:17:11 2003
From: jan.fischer22@epost.de (jan.fischer22@epost.de)
Date: Fri, 2 May 2003 12:17:11 +0200
Subject: [Cmake] CMake w/ Visual Studio .Net 2003
Message-ID: <3EA937AC000053D2@PPD27103.x.de>
Hello!
I'm having trouble getting CMake (1.6.6) to run with Visual Studio .NET
2003 (7.1). CMake keeps reporting "CMake Error: EnableLanguage was unable=
to find a CMAKE_MAKE_PROGRAM".
I have already tried to adapt CMakeVS7FindMake.cmake to contain the corre=
ct
registry path:
[HKEY_LOCAL_MACHINE\\SOFTWARE\\Microsoft\\VisualStudio\\7.1\\Setup\\VS;En=
vironmentDirectory]
Unfortunately, this does not resolve the problem. What can be done about
this?
Regards,
Jan Fischer
________________________________________
Mehr Power f=FCr Ihre eMail - mit den neuen Leistungspaketen bei http://w=
ww.epost.de
From millerjv@crd.ge.com Fri May 2 14:14:04 2003
From: millerjv@crd.ge.com (Miller, James V (Research))
Date: Fri, 2 May 2003 09:14:04 -0400
Subject: [Cmake] .NET Linker options for Rational tools
Message-ID:
I had added these flags to these variables but they did not
end up in any of the vcproj files (I grepped the files).
For now, I am adding it via the gui when I need to study
a particular executable.
Jim
> -----Original Message-----
> From: Robert Mathews [mailto:rmathews@envoyww.com]
> Sent: Thursday, May 01, 2003 4:49 PM
> To: Miller, James V (Research); CMake developers (E-mail);
> Bill Hoffman
> Subject: Re: [Cmake] .NET Linker options for Rational tools
>
>
> Edit your CMakeCache.txt file, find the lines that say:
>
> CMAKE_EXE_LINKER_FLAGS
> CMAKE_SHARED_LINKER_FLAGS
>
> and add the "/incremental:no /fixed:no" flags.
>
> Regenerate your projects and that should do it. This may
> still not appear
> in the .NET gui, since the flags won't be in the right part
> of the .vcproj
> file, but they should get take effect when project is built.
>
> BTW, almost all the compile flags and link flags can't be
> added to .NET. For
> example, any compile flag other than "-D" is silently tossed
> by CMake. You
> can however add those flags directly to the appropriate defnitions in
> CMakeCache.txt, which is ok provided that you don't mind them
> being global.
>
> Hope that helps.
>
> Rob.
> ----- Original Message -----
> From: "Bill Hoffman"
> To: "Miller, James V (Research)" ;
> "CMake developers
> (E-mail)"
> Sent: Thursday, May 01, 2003 3:20 PM
> Subject: Re: [Cmake] .NET Linker options for Rational tools
>
>
> > It is on the board to be fixed, but it has not yet been done.
> > One way to do it, would be to use the nmake Makefiles.
> >
> > -Bill
> >
> >
> > At 02:34 PM 5/1/2003, Miller, James V (Research) wrote:
> > >Rational Purify and Quantify require the linker flags
> > >
> > >/incremental:no
> > >/fixed:no
> > >
> > >If I put /fixed:no on any of the CMAKE_*_LINKER_FLAGS, it
> get ignored by
> the .NET generator. I grep'ed all the *.vcproj files and /fixed:no is
> missing in all the files.
> > >
> > >To manually add the options in .NET, I go the properties
> page for the
> executable, and under "Linker"."Command line", I added
> /fixed:no to the
> Additional Options box. The setting gets written in the vcproj file
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >Can the .NET generator be set up so that one of the
> CMAKE_*_LINKER_FLAGS
> settings gets put in the AdditionalOptions section? Or has
> this been fixed
> already?
> > >
> > > "urn:schemas-microsoft-com:office:office" />
> > >
> > >Jim Miller
> > >_____________________________________
> > >Visualization & Computer Vision
> > >GE Research
> > >Bldg. KW, Room C218B
> > >P.O. Box 8, Schenectady NY 12301
> > >
> > >millerjv@research.ge.com
> > >
> > >james.miller@research.ge.com
> > >(518) 387-4005, Dial Comm: 8*833-4005,
> > >Cell: (518) 505-7065, Fax: (518) 387-6981
> > >
> > >
> > >
> >
> >
> >
> > _______________________________________________
> > Cmake mailing list
> > Cmake@public.kitware.com
> > http://public.kitware.com/mailman/listinfo/cmake
> >
>
From rmathews@envoyww.com Fri May 2 15:17:55 2003
From: rmathews@envoyww.com (Robert Mathews)
Date: Fri, 2 May 2003 10:17:55 -0400
Subject: [Cmake] .NET Linker options for Rational tools
References:
Message-ID: <005b01c310b5$a1436760$c3d3403f@victory>
Jim:
I had been using this technique for the compiler flags. Didn't actually test
it for the linker flags. I'm sorry to hear that it doesn't work.
Rob.
----- Original Message -----
From: "Miller, James V (Research)"
To: "'Robert Mathews'" ; "CMake developers (E-mail)"
Sent: Friday, May 02, 2003 9:14 AM
Subject: RE: [Cmake] .NET Linker options for Rational tools
> I had added these flags to these variables but they did not
> end up in any of the vcproj files (I grepped the files).
>
> For now, I am adding it via the gui when I need to study
> a particular executable.
>
> Jim
>
> > -----Original Message-----
> > From: Robert Mathews [mailto:rmathews@envoyww.com]
> > Sent: Thursday, May 01, 2003 4:49 PM
> > To: Miller, James V (Research); CMake developers (E-mail);
> > Bill Hoffman
> > Subject: Re: [Cmake] .NET Linker options for Rational tools
> >
> >
> > Edit your CMakeCache.txt file, find the lines that say:
> >
> > CMAKE_EXE_LINKER_FLAGS
> > CMAKE_SHARED_LINKER_FLAGS
> >
> > and add the "/incremental:no /fixed:no" flags.
> >
> > Regenerate your projects and that should do it. This may
> > still not appear
> > in the .NET gui, since the flags won't be in the right part
> > of the .vcproj
> > file, but they should get take effect when project is built.
> >
> > BTW, almost all the compile flags and link flags can't be
> > added to .NET. For
> > example, any compile flag other than "-D" is silently tossed
> > by CMake. You
> > can however add those flags directly to the appropriate defnitions in
> > CMakeCache.txt, which is ok provided that you don't mind them
> > being global.
> >
> > Hope that helps.
> >
> > Rob.
> > ----- Original Message -----
> > From: "Bill Hoffman"
> > To: "Miller, James V (Research)" ;
> > "CMake developers
> > (E-mail)"
> > Sent: Thursday, May 01, 2003 3:20 PM
> > Subject: Re: [Cmake] .NET Linker options for Rational tools
> >
> >
> > > It is on the board to be fixed, but it has not yet been done.
> > > One way to do it, would be to use the nmake Makefiles.
> > >
> > > -Bill
> > >
> > >
> > > At 02:34 PM 5/1/2003, Miller, James V (Research) wrote:
> > > >Rational Purify and Quantify require the linker flags
> > > >
> > > >/incremental:no
> > > >/fixed:no
> > > >
> > > >If I put /fixed:no on any of the CMAKE_*_LINKER_FLAGS, it
> > get ignored by
> > the .NET generator. I grep'ed all the *.vcproj files and /fixed:no is
> > missing in all the files.
> > > >
> > > >To manually add the options in .NET, I go the properties
> > page for the
> > executable, and under "Linker"."Command line", I added
> > /fixed:no to the
> > Additional Options box. The setting gets written in the vcproj file
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >Can the .NET generator be set up so that one of the
> > CMAKE_*_LINKER_FLAGS
> > settings gets put in the AdditionalOptions section? Or has
> > this been fixed
> > already?
> > > >
> > > > > "urn:schemas-microsoft-com:office:office" />
> > > >
> > > >Jim Miller
> > > >_____________________________________
> > > >Visualization & Computer Vision
> > > >GE Research
> > > >Bldg. KW, Room C218B
> > > >P.O. Box 8, Schenectady NY 12301
> > > >
> > > >millerjv@research.ge.com
> > > >
> > > >james.miller@research.ge.com
> > > >(518) 387-4005, Dial Comm: 8*833-4005,
> > > >Cell: (518) 505-7065, Fax: (518) 387-6981
> > > >
> > > >
> > > >
> > >
> > >
> > >
> > > _______________________________________________
> > > Cmake mailing list
> > > Cmake@public.kitware.com
> > > http://public.kitware.com/mailman/listinfo/cmake
> > >
> >
>
From Stefan.Schmidt@sophia.inria.fr Fri May 2 15:15:52 2003
From: Stefan.Schmidt@sophia.inria.fr (Stefan Schmidt)
Date: Fri, 2 May 2003 16:15:52 +0200
Subject: [Cmake] planning KDevelop support
Message-ID: <200305021615.52811.Stefan.Schmidt@sophia.inria.fr>
Hello,
I'd like to achieve some integration of CMake in KDevelop.
My question: In what way do I need to modify CMake to give me a list of a=
ll=20
source files, say, for a given target, or for the whole project (includin=
g=20
subdirs)?=20
Is it better to create a new Makefile generator, derived from the Unix on=
e,=20
which additionally outputs such a list, or might it be better to create a=
new=20
CMake Command (or modify one of the ADD_EXECUTABLE commands) for that?
Some background information:
In the upcoming Kdevelop 3, there's support for "Parts" that handle proje=
ct=20
management (there's already one for the autotools and one for qmake).=20
I imagine creating a Part like these which runs (a modified) CMake as the=
=20
configure step which creates this list. Then this list is used to form th=
e=20
project's file tree view.=20
Additionally, integrating the cache editing GUI in the Part is possible, =
too.
What do you think about this?
Kind regards,
Stefan
From bill.hoffman@kitware.com Fri May 2 15:22:02 2003
From: bill.hoffman@kitware.com (Bill Hoffman)
Date: Fri, 02 May 2003 10:22:02 -0400
Subject: [Cmake] .NET Linker options for Rational tools
In-Reply-To: <005b01c310b5$a1436760$c3d3403f@victory>
References:
Message-ID: <5.2.0.9.0.20030502101648.04b7e390@pop.biz.rr.com>
I just checked, and this is fixed in CVS cmake.
It should be in the next patch release for 1.6.
Jim, can you try cvs cmake and let me know if it works
for you? Thanks.
-Bill
At 10:17 AM 5/2/2003, Robert Mathews wrote:
>Jim:
>
>I had been using this technique for the compiler flags. Didn't actually test
>it for the linker flags. I'm sorry to hear that it doesn't work.
>
>Rob.
>----- Original Message -----
>From: "Miller, James V (Research)"
>To: "'Robert Mathews'" ; "CMake developers (E-mail)"
>
>Sent: Friday, May 02, 2003 9:14 AM
>Subject: RE: [Cmake] .NET Linker options for Rational tools
>
>
>> I had added these flags to these variables but they did not
>> end up in any of the vcproj files (I grepped the files).
>>
>> For now, I am adding it via the gui when I need to study
>> a particular executable.
>>
>> Jim
>>
>> > -----Original Message-----
>> > From: Robert Mathews [mailto:rmathews@envoyww.com]
>> > Sent: Thursday, May 01, 2003 4:49 PM
>> > To: Miller, James V (Research); CMake developers (E-mail);
>> > Bill Hoffman
>> > Subject: Re: [Cmake] .NET Linker options for Rational tools
>> >
>> >
>> > Edit your CMakeCache.txt file, find the lines that say:
>> >
>> > CMAKE_EXE_LINKER_FLAGS
>> > CMAKE_SHARED_LINKER_FLAGS
>> >
>> > and add the "/incremental:no /fixed:no" flags.
>> >
>> > Regenerate your projects and that should do it. This may
>> > still not appear
>> > in the .NET gui, since the flags won't be in the right part
>> > of the .vcproj
>> > file, but they should get take effect when project is built.
>> >
>> > BTW, almost all the compile flags and link flags can't be
>> > added to .NET. For
>> > example, any compile flag other than "-D" is silently tossed
>> > by CMake. You
>> > can however add those flags directly to the appropriate defnitions in
>> > CMakeCache.txt, which is ok provided that you don't mind them
>> > being global.
>> >
>> > Hope that helps.
>> >
>> > Rob.
>> > ----- Original Message -----
>> > From: "Bill Hoffman"
>> > To: "Miller, James V (Research)" ;
>> > "CMake developers
>> > (E-mail)"
>> > Sent: Thursday, May 01, 2003 3:20 PM
>> > Subject: Re: [Cmake] .NET Linker options for Rational tools
>> >
>> >
>> > > It is on the board to be fixed, but it has not yet been done.
>> > > One way to do it, would be to use the nmake Makefiles.
>> > >
>> > > -Bill
>> > >
>> > >
>> > > At 02:34 PM 5/1/2003, Miller, James V (Research) wrote:
>> > > >Rational Purify and Quantify require the linker flags
>> > > >
>> > > >/incremental:no
>> > > >/fixed:no
>> > > >
>> > > >If I put /fixed:no on any of the CMAKE_*_LINKER_FLAGS, it
>> > get ignored by
>> > the .NET generator. I grep'ed all the *.vcproj files and /fixed:no is
>> > missing in all the files.
>> > > >
>> > > >To manually add the options in .NET, I go the properties
>> > page for the
>> > executable, and under "Linker"."Command line", I added
>> > /fixed:no to the
>> > Additional Options box. The setting gets written in the vcproj file
>> > > >
>> > > >
>> > > >
>> > > >
>> > > >
>> > > >
>> > > >
>> > > >
>> > > >
>> > > >Can the .NET generator be set up so that one of the
>> > CMAKE_*_LINKER_FLAGS
>> > settings gets put in the AdditionalOptions section? Or has
>> > this been fixed
>> > already?
>> > > >
>> > > > > > "urn:schemas-microsoft-com:office:office" />
>> > > >
>> > > >Jim Miller
>> > > >_____________________________________
>> > > >Visualization & Computer Vision
>> > > >GE Research
>> > > >Bldg. KW, Room C218B
>> > > >P.O. Box 8, Schenectady NY 12301
>> > > >
>> > > >millerjv@research.ge.com
>> > > >
>> > > >james.miller@research.ge.com
>> > > >(518) 387-4005, Dial Comm: 8*833-4005,
>> > > >Cell: (518) 505-7065, Fax: (518) 387-6981
>> > > >
>> > > >
>> > > >
>> > >
>> > >
>> > >
>> > > _______________________________________________
>> > > Cmake mailing list
>> > > Cmake@public.kitware.com
>> > > http://public.kitware.com/mailman/listinfo/cmake
>> > >
>> >
>>
>
>_______________________________________________
>Cmake mailing list
>Cmake@public.kitware.com
>http://public.kitware.com/mailman/listinfo/cmake
From bill.hoffman@kitware.com Fri May 2 15:27:21 2003
From: bill.hoffman@kitware.com (Bill Hoffman)
Date: Fri, 02 May 2003 10:27:21 -0400
Subject: [Cmake] planning KDevelop support
In-Reply-To: <200305021615.52811.Stefan.Schmidt@sophia.inria.fr>
Message-ID: <5.2.0.9.0.20030502102339.04c142f8@pop.biz.rr.com>
I think you would need to create a new Makefile generator.
How does the autoconf one work?
I assume that KDevelop still uses make to do the work of
building right? If that is the case, you could sub-class
off the unix makefile generator, and just have it output a few
extra files as well.
-Bill
At 10:15 AM 5/2/2003, Stefan Schmidt wrote:
>Hello,
>
>I'd like to achieve some integration of CMake in KDevelop.
>
>My question: In what way do I need to modify CMake to give me a list of all
>source files, say, for a given target, or for the whole project (including
>subdirs)?
>
>Is it better to create a new Makefile generator, derived from the Unix one,
>which additionally outputs such a list, or might it be better to create a new
>CMake Command (or modify one of the ADD_EXECUTABLE commands) for that?
>
>Some background information:
>In the upcoming Kdevelop 3, there's support for "Parts" that handle project
>management (there's already one for the autotools and one for qmake).
>I imagine creating a Part like these which runs (a modified) CMake as the
>configure step which creates this list. Then this list is used to form the
>project's file tree view.
>Additionally, integrating the cache editing GUI in the Part is possible, too.
>
>What do you think about this?
>
>Kind regards,
>Stefan
>_______________________________________________
>Cmake mailing list
>Cmake@public.kitware.com
>http://public.kitware.com/mailman/listinfo/cmake
From Stefan.Schmidt@sophia.inria.fr Fri May 2 16:45:19 2003
From: Stefan.Schmidt@sophia.inria.fr (Stefan Schmidt)
Date: Fri, 2 May 2003 17:45:19 +0200
Subject: [Cmake] strange behavior
Message-ID: <200305021745.19776.Stefan.Schmidt@sophia.inria.fr>
Hello,
I have a strange problem with cmake (1.6.5) that I don't fully understand=
:
I've some CMake code like this (simplified here), that sets some cache en=
tries=20
in a loop:
---
SET (ALLPACKAGES
AuxPkgs/TestData
StdPkgs/ImagePkg
AuxPkgs/Package3
)
FOREACH(PKG ${ALLPACKAGES})
STRING(REGEX REPLACE "/" "_" PKGNAME ${PKG})
SET ("${PKGNAME}_DIR" ${CMAKE_BINARY_DIR}/${PKG} CACHE PATH "Locatio=
n of=20
package ${PKG}" FORCE)
ENDFOREACH(PKG)
SUBDIRS(TestSubDir)
---
Additionally to the intended AuxPkgs_TestData_DIR entries, a _DIR entry i=
s=20
created with the same value as the last correct one.
This isn't the case when there are no SUBDIRS! Appearently, it has someth=
ing=20
to do with the way CMake processes the subdirectories and rereads (?) the=
=20
parent's CMakeList.
I get around this if I use a (obviously superfluous) IF(PKGNAME)...ENDIF=20
around the cache setting.
I think this might either be a bug or the underlying mechanism how CMake=20
traverses the SUBDIRs has to be better documented.=20
Kind regards,
Stefan
From ken.martin@kitware.com Fri May 2 16:56:56 2003
From: ken.martin@kitware.com (Ken Martin)
Date: Fri, 2 May 2003 11:56:56 -0400
Subject: [Cmake] strange behavior
In-Reply-To: <200305021745.19776.Stefan.Schmidt@sophia.inria.fr>
Message-ID: <003f01c310c3$7519fb50$6601a8c0@kitwarein.com>
The foreach loop is being process for each subdirectory as well.
Normally this isn't a problem because all the commands in the foreach
loop are also inherited. But... there was a bug in CMake 1.6.5 where the
STRING command is not inherited. And that is causing your problem. I
know this is fixed in CVS CMake and it should be fixed for CMake 1.6.7
when we release that patch. If you want to know what commands are
inherited (i.e. run for subdirs as well) you can look at the header
files for the command in the source distribution. I believe almost all
commands are inherited except for ADD_EXECUTABLE ADD_LIBRARY
ADD_CUSTOM_COMMAND ADD_CUSTOM_TARGET INSTALL_FILES INSTALL_PROGRAM
INSTALL_TARGET and maybe one or two others.
Thanks
Ken
> -----Original Message-----
> From: cmake-admin@public.kitware.com [mailto:cmake-
> admin@public.kitware.com] On Behalf Of Stefan Schmidt
> Sent: Friday, May 02, 2003 10:45 AM
> To: cmake@public.kitware.com
> Subject: [Cmake] strange behavior
>
> Hello,
> I have a strange problem with cmake (1.6.5) that I don't fully
understand:
>
> I've some CMake code like this (simplified here), that sets some cache
> entries
> in a loop:
> ---
> SET (ALLPACKAGES
> AuxPkgs/TestData
> StdPkgs/ImagePkg
> AuxPkgs/Package3
> )
>
> FOREACH(PKG ${ALLPACKAGES})
> STRING(REGEX REPLACE "/" "_" PKGNAME ${PKG})
> SET ("${PKGNAME}_DIR" ${CMAKE_BINARY_DIR}/${PKG} CACHE PATH
"Location
> of
> package ${PKG}" FORCE)
> ENDFOREACH(PKG)
>
> SUBDIRS(TestSubDir)
> ---
>
> Additionally to the intended AuxPkgs_TestData_DIR entries, a _DIR
entry is
> created with the same value as the last correct one.
>
> This isn't the case when there are no SUBDIRS! Appearently, it has
> something
> to do with the way CMake processes the subdirectories and rereads (?)
the
> parent's CMakeList.
> I get around this if I use a (obviously superfluous)
IF(PKGNAME)...ENDIF
> around the cache setting.
>
> I think this might either be a bug or the underlying mechanism how
CMake
> traverses the SUBDIRs has to be better documented.
>
>
> Kind regards,
> Stefan
> _______________________________________________
> Cmake mailing list
> Cmake@public.kitware.com
> http://public.kitware.com/mailman/listinfo/cmake
From filipe.gr.sousa@alunos.ipb.pt Fri May 2 20:23:54 2003
From: filipe.gr.sousa@alunos.ipb.pt (Filipe Sousa)
Date: Fri, 02 May 2003 20:23:54 +0100
Subject: [Cmake] planning KDevelop support
In-Reply-To: <200305021615.52811.Stefan.Schmidt@sophia.inria.fr>
References: <200305021615.52811.Stefan.Schmidt@sophia.inria.fr>
Message-ID: <3EB2C5CA.4090100@alunos.ipb.pt>
Hi,
You can use cmake with kdevelop. Create your cmake project and import
into kdevelop in Project|Generate Project File menu. You can tell
kdevelop where to run make and where is the binary generated.
Stefan Schmidt wrote:
> Hello,
>
> I'd like to achieve some integration of CMake in KDevelop.
>
> My question: In what way do I need to modify CMake to give me a list of all
> source files, say, for a given target, or for the whole project (including
> subdirs)?
>
> Is it better to create a new Makefile generator, derived from the Unix one,
> which additionally outputs such a list, or might it be better to create a new
> CMake Command (or modify one of the ADD_EXECUTABLE commands) for that?
>
> Some background information:
> In the upcoming Kdevelop 3, there's support for "Parts" that handle project
> management (there's already one for the autotools and one for qmake).
> I imagine creating a Part like these which runs (a modified) CMake as the
> configure step which creates this list. Then this list is used to form the
> project's file tree view.
> Additionally, integrating the cache editing GUI in the Part is possible, too.
>
> What do you think about this?
>
> Kind regards,
> Stefan
> _______________________________________________
> Cmake mailing list
> Cmake@public.kitware.com
> http://public.kitware.com/mailman/listinfo/cmake
>
>
From John Biddiscombe"
Message-ID: <000501c311d4$aa3e62a0$0100a8c0@tigger>
When compiling Java wrappers for vtk, my Borland makefile contains the entry
for Java Archive (see bottom of email)...but all the dependencies are listed
with no line breaks - my editor only displays the first 16,000 characters
before complaining, so I'm not sure how many there are. Can the line be
broken up? The make fails.
[Actually, the make fails before it gets here because it complains about
rule being too long for
VTKJavaJar: \
D:\CMakebuild\vtkDebug\java\vtk\vtkPanel.java \
D:\CMakebuild\vtkDebug\java\vtk\vtkPanel.java \
D:\CMakebuild\vtkDebug\java\vtk\vtkPanel.java \
D:\CMakebuild\vtkDebug\java\vtk\vtkPanel.java \
D:\CMakebuild\vtkDebug\java\vtk\vtkAbstractMapper.class \
D:\CMakebuild\vtkDebug\java\vtk\vtkAbstractTransform.class \
....etc etc for several hundred lines
Note also the duplicated entries for vtkPanel. I'm not sure if this rule is
really too long or if make is being knackered by the later problem of a very
very long line below..
]
#---------------------------------------------------------
# Java Archive
#
D:\cmakebuild\vtkDebug\bin\vtk.jar:
D:\CMakebuild\vtkDebug\java\vtk\vtkObject.class
D:\CMakebuild\vtkDebug\java\vtk\vtk3DSImporter.class
D:\CMakebuild\vtkDebug\java\vtk\vtk3DWidget.class
D:\CMakebuild\vtkDebug\java\vtk\vtkAbstractMapper.class
D:\CMakebuild\vtkDebug\java\vtk\vtkAbstractMapper3D.class
D:\CMakebuild\vtkDebug\java\vtk\vtkAbstractPicker.class
D:\CMakebuild\vtkDebug\java\vtk\vtkAbstractPropPicker.class
....and on for 16,000 more characters
From Mathieu.Malaterre@creatis.insa-lyon.fr Mon May 5 13:46:23 2003
From: Mathieu.Malaterre@creatis.insa-lyon.fr (Mathieu Malaterre)
Date: Mon, 05 May 2003 14:46:23 +0200
Subject: [Cmake] gtk-config / pkg-config
Message-ID: <3EB65D1F.9080601@creatis.insa-lyon.fr>
Hi all,
I needed to test my class wxVTKRenderWindowInteractor:
http://www.creatis.insa-lyon.fr/~malaterre/wxVTK/
with GTK2 so I rewrote FindGTK.cmake (based on FindwxWindow) like this:
#######
IF(WIN32)
...
ELSE(WIN32)
FIND_PROGRAM(CMAKE_PKG_CONFIG pkg-config ../gtk/bin ../../gtk/bin)
IF(CMAKE_PKG_CONFIG)
SET(CMAKE_GTK_CXX_FLAGS "`${CMAKE_PKG_CONFIG} --cflags gtk+-2.0`")
SET(GTK_LIBRARIES "`${CMAKE_PKG_CONFIG} --libs gtk+-2.0`")
ELSE(CMAKE_PKG_CONFIG)
FIND_PROGRAM(CMAKE_GTK_CONFIG gtk-config ../gtk/bin ../../gtk/bin)
SET(CMAKE_GTK_CXX_FLAGS "`${CMAKE_GTK_CONFIG} --cxxflags`")
SET(GTK_LIBRARIES "`${CMAKE_GTK_CONFIG} --libs`")
ENDIF(CMAKE_PKG_CONFIG)
ENDIF(WIN32)
MARK_AS_ADVANCED(
CMAKE_GTK_CXX_FLAGS
GTK_INCLUDE_DIR
)
IF(GTK_LIBRARIES)
IF(CMAKE_GTK_CXX_FLAGS)
SET(GTK_FOUND 1)
ENDIF(CMAKE_GTK_CXX_FLAGS)
ENDIF(GTK_LIBRARIES)
#######
This take advantage of pkg-config which should replace *-config tool. See:
http://www.freedesktop.org/software/pkgconfig/
But I am stuck:
- I would like the user to choose either if he wants GTK 1.x or GTK 2.x
but I don't know how to handle this.
- Furthermore, the user should not be able to use GTK 2.x if pkg-config
can't be found (only gtk-config was found). There will be also some
problems if pkg-config is not in the path and is specified later...
Thanks for your suggestions,
mathieu
--
Mathieu Malaterre
CREATIS
28 Avenue du Doyen LEPINE
B.P. Lyon-Montchat
69394 Lyon Cedex 03
http://www.creatis.insa-lyon.fr/~malaterre/
From andy.cedilnik@kitware.com Mon May 5 14:02:32 2003
From: andy.cedilnik@kitware.com (Andy Cedilnik)
Date: 05 May 2003 09:02:32 -0400
Subject: [Cmake] gtk-config / pkg-config
In-Reply-To: <3EB65D1F.9080601@creatis.insa-lyon.fr>
References: <3EB65D1F.9080601@creatis.insa-lyon.fr>
Message-ID: <1052139752.664.230.camel@andoria>
Hi Mathieu,
I am looking for a generic solution for pkg-config since I will need it
for openssl. I think there should be something like FindPKGConfig.cmake,
which sets some paths and couple of macros. Then you can use these
macros in other Find packages. For example:
IF(HAVE_PKGCONFIG)
PKGCONFIG_GET_LIBRARIES(OPENSSL_LIBRARIES openssl)
PKGCONFIG_GET_FLAGS(OPENSSL_FLAGS openssl)
ENDIF(HAVE_PKGCONFIG)
How does that sound?
Andy
On Mon, 2003-05-05 at 08:46, Mathieu Malaterre wrote:
> Hi all,
> I needed to test my class wxVTKRenderWindowInteractor:
> http://www.creatis.insa-lyon.fr/~malaterre/wxVTK/
>
> with GTK2 so I rewrote FindGTK.cmake (based on FindwxWindow) like this:
>
> #######
> IF(WIN32)
> ...
>
> ELSE(WIN32)
>
> FIND_PROGRAM(CMAKE_PKG_CONFIG pkg-config ../gtk/bin ../../gtk/bin)
> IF(CMAKE_PKG_CONFIG)
> SET(CMAKE_GTK_CXX_FLAGS "`${CMAKE_PKG_CONFIG} --cflags gtk+-2.0`")
> SET(GTK_LIBRARIES "`${CMAKE_PKG_CONFIG} --libs gtk+-2.0`")
> ELSE(CMAKE_PKG_CONFIG)
> FIND_PROGRAM(CMAKE_GTK_CONFIG gtk-config ../gtk/bin ../../gtk/bin)
> SET(CMAKE_GTK_CXX_FLAGS "`${CMAKE_GTK_CONFIG} --cxxflags`")
> SET(GTK_LIBRARIES "`${CMAKE_GTK_CONFIG} --libs`")
> ENDIF(CMAKE_PKG_CONFIG)
> ENDIF(WIN32)
>
> MARK_AS_ADVANCED(
> CMAKE_GTK_CXX_FLAGS
> GTK_INCLUDE_DIR
> )
>
> IF(GTK_LIBRARIES)
> IF(CMAKE_GTK_CXX_FLAGS)
> SET(GTK_FOUND 1)
> ENDIF(CMAKE_GTK_CXX_FLAGS)
> ENDIF(GTK_LIBRARIES)
> #######
>
> This take advantage of pkg-config which should replace *-config tool. See:
> http://www.freedesktop.org/software/pkgconfig/
>
> But I am stuck:
> - I would like the user to choose either if he wants GTK 1.x or GTK 2.x
> but I don't know how to handle this.
>
> - Furthermore, the user should not be able to use GTK 2.x if pkg-config
> can't be found (only gtk-config was found). There will be also some
> problems if pkg-config is not in the path and is specified later...
>
> Thanks for your suggestions,
From Mathieu.Malaterre@creatis.insa-lyon.fr Mon May 5 14:04:21 2003
From: Mathieu.Malaterre@creatis.insa-lyon.fr (Mathieu Malaterre)
Date: Mon, 05 May 2003 15:04:21 +0200
Subject: [Cmake] gtk-config / pkg-config
In-Reply-To: <1052139752.664.230.camel@andoria>
References: <3EB65D1F.9080601@creatis.insa-lyon.fr> <1052139752.664.230.camel@andoria>
Message-ID: <3EB66155.4000708@creatis.insa-lyon.fr>
Andy,
That sounds excellent !
Anyway GTK is still a particular package as it can be accessible either
way though gtk-config or pkg-config. So if you could find a way to hide
this in FindGTK.cmake this would be very nice.
And one feature I would like to see is in FindwxWindow, so that it
determine which GTK was used (I don't know if this is possible without
using your CMakeLists for wxWindow).
BTW, when will FindPKGConfig.cmake be available ?
mathieu
Andy Cedilnik wrote:
> Hi Mathieu,
>
> I am looking for a generic solution for pkg-config since I will need it
> for openssl. I think there should be something like FindPKGConfig.cmake,
> which sets some paths and couple of macros. Then you can use these
> macros in other Find packages. For example:
>
> IF(HAVE_PKGCONFIG)
> PKGCONFIG_GET_LIBRARIES(OPENSSL_LIBRARIES openssl)
> PKGCONFIG_GET_FLAGS(OPENSSL_FLAGS openssl)
> ENDIF(HAVE_PKGCONFIG)
>
> How does that sound?
>
> Andy
>
> On Mon, 2003-05-05 at 08:46, Mathieu Malaterre wrote:
>
>>Hi all,
>> I needed to test my class wxVTKRenderWindowInteractor:
>>http://www.creatis.insa-lyon.fr/~malaterre/wxVTK/
>>
>>with GTK2 so I rewrote FindGTK.cmake (based on FindwxWindow) like this:
>>
>>#######
>>IF(WIN32)
>>...
>>
>>ELSE(WIN32)
>>
>> FIND_PROGRAM(CMAKE_PKG_CONFIG pkg-config ../gtk/bin ../../gtk/bin)
>> IF(CMAKE_PKG_CONFIG)
>> SET(CMAKE_GTK_CXX_FLAGS "`${CMAKE_PKG_CONFIG} --cflags gtk+-2.0`")
>> SET(GTK_LIBRARIES "`${CMAKE_PKG_CONFIG} --libs gtk+-2.0`")
>> ELSE(CMAKE_PKG_CONFIG)
>> FIND_PROGRAM(CMAKE_GTK_CONFIG gtk-config ../gtk/bin ../../gtk/bin)
>> SET(CMAKE_GTK_CXX_FLAGS "`${CMAKE_GTK_CONFIG} --cxxflags`")
>> SET(GTK_LIBRARIES "`${CMAKE_GTK_CONFIG} --libs`")
>> ENDIF(CMAKE_PKG_CONFIG)
>>ENDIF(WIN32)
>>
>>MARK_AS_ADVANCED(
>> CMAKE_GTK_CXX_FLAGS
>> GTK_INCLUDE_DIR
>>)
>>
>>IF(GTK_LIBRARIES)
>> IF(CMAKE_GTK_CXX_FLAGS)
>> SET(GTK_FOUND 1)
>> ENDIF(CMAKE_GTK_CXX_FLAGS)
>>ENDIF(GTK_LIBRARIES)
>>#######
>>
>>This take advantage of pkg-config which should replace *-config tool. See:
>>http://www.freedesktop.org/software/pkgconfig/
>>
>>But I am stuck:
>>- I would like the user to choose either if he wants GTK 1.x or GTK 2.x
>>but I don't know how to handle this.
>>
>>- Furthermore, the user should not be able to use GTK 2.x if pkg-config
>>can't be found (only gtk-config was found). There will be also some
>>problems if pkg-config is not in the path and is specified later...
>>
>>Thanks for your suggestions,
>
>
>
--
Mathieu Malaterre
CREATIS
28 Avenue du Doyen LEPINE
B.P. Lyon-Montchat
69394 Lyon Cedex 03
http://www.creatis.insa-lyon.fr/~malaterre/
From Stefan.Schmidt@sophia.inria.fr Mon May 5 15:25:25 2003
From: Stefan.Schmidt@sophia.inria.fr (Stefan Schmidt)
Date: Mon, 5 May 2003 16:25:25 +0200
Subject: [Cmake] using # in a string
Message-ID: <200305051625.25836.Stefan.Schmidt@sophia.inria.fr>
Hallo,
Can't I use # in a string?=20
Or is there a way to escape its special meaning of starting a comment?
Greetings,
Stefan
From Stefan.Schmidt@sophia.inria.fr Mon May 5 15:58:25 2003
From: Stefan.Schmidt@sophia.inria.fr (Stefan Schmidt)
Date: Mon, 5 May 2003 16:58:25 +0200
Subject: [Cmake] @@ replacement in configure_file
Message-ID: <200305051658.25025.Stefan.Schmidt@sophia.inria.fr>
I've something like this in an .in file:
@INCLUDE =3D @DOXYGEN_INCLUDES@
INCLUDE =3D @DOXYGEN_INCLUDES@
that gets configured by a sequence like this:
SET(DOXYGEN_INCLUDES "${PROJECT_BINARY_DIR}/DoxyTagFiles.inc")
CONFIGURE_FILE (${COMMONCONFIG_DIR}/doxygen.config.in
=09 ${CMAKE_CURRENT_BINARY_DIR}/doxygen.config @ONLY)
this is the outcome:
@INCLUDE =3D
INCLUDE =3D /u/huangdi/0/sschmidt/mips/build/DoxyTagFiles.inc
Appearantly, lines with more than one @ aren't handled the way I want. I =
need=20
an @ sign in front of the INCLUDE in the resulting file. What can I do ab=
out=20
this?
Thanks,
Stefan
From andy.cedilnik@kitware.com Mon May 5 16:29:02 2003
From: andy.cedilnik@kitware.com (Andy Cedilnik)
Date: 05 May 2003 11:29:02 -0400
Subject: [Cmake] using # in a string
In-Reply-To: <200305051625.25836.Stefan.Schmidt@sophia.inria.fr>
References: <200305051625.25836.Stefan.Schmidt@sophia.inria.fr>
Message-ID: <1052148542.663.238.camel@andoria>
Hi Stefan,
You can get any character in the CMake variable by using STRING ASCII
command. Unfortunately there is a bug in CMake which makes all STRING
commands only work in the leaf CMakeLists.txt. So, if you only need the
funny character in the current CMakeLists.txt, then here we go:
STRING(ASCII 35 POUND)
SET(SOME_VAR "The pound is ${POUND}")
Andy
On Mon, 2003-05-05 at 10:25, Stefan Schmidt wrote:
> Can't I use # in a string?
> Or is there a way to escape its special meaning of starting a comment?
From andy.cedilnik@kitware.com Mon May 5 16:46:41 2003
From: andy.cedilnik@kitware.com (Andy Cedilnik)
Date: 05 May 2003 11:46:41 -0400
Subject: [Cmake] @@ replacement in configure_file
In-Reply-To: <200305051658.25025.Stefan.Schmidt@sophia.inria.fr>
References: <200305051658.25025.Stefan.Schmidt@sophia.inria.fr>
Message-ID: <1052149601.666.243.camel@andoria>
Hi Stefan,
That seems to be a bug in parsing. It can be however an ambiguous case.
For example:
FOO = @SOMETHING@FOOBAR@
If you want to fix that, you can always try to use ${} style variables.
I do not think doxygen needs ${} in the doxyfile, so you should be ok
there. Just make sure to remove @ONLY.
Andy
On Mon, 2003-05-05 at 10:58, Stefan Schmidt wrote:
> I've something like this in an .in file:
>
> @INCLUDE = @DOXYGEN_INCLUDES@
> INCLUDE = @DOXYGEN_INCLUDES@
>
> that gets configured by a sequence like this:
>
> SET(DOXYGEN_INCLUDES "${PROJECT_BINARY_DIR}/DoxyTagFiles.inc")
> CONFIGURE_FILE (${COMMONCONFIG_DIR}/doxygen.config.in
> ${CMAKE_CURRENT_BINARY_DIR}/doxygen.config @ONLY)
>
> this is the outcome:
> @INCLUDE =
> INCLUDE = /u/huangdi/0/sschmidt/mips/build/DoxyTagFiles.inc
>
> Appearantly, lines with more than one @ aren't handled the way I want. I need
> an @ sign in front of the INCLUDE in the resulting file. What can I do about
> this?
From Peter.Vanroose@esat.kuleuven.ac.be Mon May 5 17:07:23 2003
From: Peter.Vanroose@esat.kuleuven.ac.be (Peter Vanroose)
Date: Mon, 5 May 2003 18:07:23 +0200 (CEST)
Subject: [Cmake] using # in a string
In-Reply-To: <200305051625.25836.Stefan.Schmidt@sophia.inria.fr>
References: <200305051625.25836.Stefan.Schmidt@sophia.inria.fr>
Message-ID:
> Can't I use # in a string?
> Or is there a way to escape its special meaning of starting a comment?
If I'm not mistaken, "#" is only a comment delimiter at start of line,
so any # which follows a non-space is not considered a comment delimiter.
-- Peter Vanroose.
From Stefan.Schmidt@sophia.inria.fr Mon May 5 17:20:39 2003
From: Stefan.Schmidt@sophia.inria.fr (Stefan Schmidt)
Date: Mon, 5 May 2003 18:20:39 +0200
Subject: [Cmake] using # in a string
In-Reply-To:
References: <200305051625.25836.Stefan.Schmidt@sophia.inria.fr>
Message-ID: <200305051820.39613.Stefan.Schmidt@sophia.inria.fr>
> If I'm not mistaken, "#" is only a comment delimiter at start of line,
> so any # which follows a non-space is not considered a comment delimite=
r.
I think it works everywhere in CMakeList code (but I think at least insid=
e=20
strings, it shouldn't). Tested with CMake 1.6.5 and some older 1.7 cvs=20
version.
But thanks for pointing out the workaround by using STRING(ASCII 35 POUND=
).
Greetings,
Stefan
From Stefan.Schmidt@sophia.inria.fr Mon May 5 17:37:43 2003
From: Stefan.Schmidt@sophia.inria.fr (Stefan Schmidt)
Date: Mon, 5 May 2003 18:37:43 +0200
Subject: [Cmake] recursive custom targets
Message-ID: <200305051837.43465.Stefan.Schmidt@sophia.inria.fr>
Sorry, I've yet another question on which I couldn't find any info:
How can I create a custom target that is made recursively?
My documentation needs to be generated in a 2-pass process, so I'd like t=
o=20
have a "make docprep" and a "make doc" target, which both are done=20
recursively for the whole project.
Is this possible?
Thanks in advance,
Stefan
From John Biddiscombe"
Message-ID: <000701c313c6$f7a0f8c0$0100a8c0@tigger>
After turning on Java wrapping for vtk, I get this list of errors when doing
a build with MSVS7
Microsoft (R) Development Environment Version 7.00.9466.
Copyright (C) Microsoft Corp 1984-2001. All rights reserved.
Project 'D:\CMakebuild\vtkVS7\Common\vtkCommonJava.vcproj' failed to open.
Project 'D:\CMakebuild\vtkVS7\Filtering\vtkFilteringJava.vcproj' failed to
open.
Project 'D:\CMakebuild\vtkVS7\Imaging\vtkImagingJava.vcproj' failed to open.
Project 'D:\CMakebuild\vtkVS7\Graphics\vtkGraphicsJava.vcproj' failed to
open.
Project 'D:\CMakebuild\vtkVS7\IO\vtkIOJava.vcproj' failed to open.
Project 'D:\CMakebuild\vtkVS7\Rendering\vtkRenderingJava.vcproj' failed to
open.
Project 'D:\CMakebuild\vtkVS7\Hybrid\vtkHybridJava.vcproj' failed to open.
Is this a cmake or a vtk issue? (and can I do anything about it?)
the project files in question look fine to the casual observer.
thanks
JB
From d.nuetzel@wearabrain.de Wed May 7 11:54:15 2003
From: d.nuetzel@wearabrain.de (Dieter =?iso-8859-15?q?N=FCtzel?=)
Date: Wed, 7 May 2003 12:54:15 +0200
Subject: [Cmake] CMake CVS build is broken for several days, now.
Message-ID: <200305071254.15805.d.nuetzel@wearabrain.de>
gmake[1]: Entering directory `/opt/VTK/V4.0/CMake/Bootstrap.cmk/Source'
c++ cmake.o cmakewizard.o cmakemain.o cmMakeDepend.o cmMakefile.o
cmDocumentation.o cmGlobalGenerator.o cmLocalGenerator.o
cmRegularExpression.o cmSourceFile.o cmSystemTools.o cmDirectory.o
cmGlobalUnixMakefileGenerator.o cmLocalUnixMakefileGenerator.o cmCommands.o
cmTarget.o cmCustomCommand.o cmCacheManager.o cmListFileCache.o
cmVariableWatch.o cmSourceGroup.o -DCMAKE_ROOT_DIR=3D'"/opt/VTK/V4.0/CMake"=
' -O
=2Dmcpu=3Dk6 -mpreferred-stack-boundary=3D2 -malign-functions=3D4 -fschedul=
e-insns2
=2Dfexpensive-optimizations -DCMAKE_HAS_AUTOCONF
=2DI/opt/VTK/V4.0/CMake/Bootstrap.cmk/Source -o cmake
c++: cmDocumentation.o: No such file or directory
c++: cmSystemTools.o: No such file or directory
gmake[1]: *** [cmake] Error 1
gmake[1]: Leaving directory `/opt/VTK/V4.0/CMake/Bootstrap.cmk/Source'
gmake: *** [all] Error 2
Problem bootstrapping CMake
0.500u 0.210s 0:00.72 98.6% 0+0k 0+0io 16517pf+0w
Thank you very much for great work.
=2D-
Dieter N=FCtzel
Leiter F&E, WEAR-A-BRAIN GmbH, Wiener Str. 5, 28359 Bremen, Germany
Mobil: 0162 673 09 09
From scharver@evl.uic.edu Wed May 7 22:50:22 2003
From: scharver@evl.uic.edu (Chris Scharver)
Date: Wed, 7 May 2003 16:50:22 -0500
Subject: [Cmake] Reconciling MD, MDd, MT, MTd libraries
Message-ID:
Hello,
I am running CMake 1.6.6, and it cannot automatically locate my library file. There are several different libraries available for different code generation styles, and I wonder if that's why things are a little strange.
C:\VRCO\CAVELib_3.0.3\lib32\libcave_ogl_mt_[MD MDd MT MTd].lib
No matter which library I try to have CMake find, the file is never found. The result is always CAVE_cave_LIBRARY-NOTFOUND.
SET (CAVE_HOME "$ENV{CAVE_HOME}")
FIND_LIBRARY(CAVE_cave_LIBRARY cave_ogl_mt_MTd
${CAVE_HOME}/lib32
}
I know the environment variable is being read properly because the header is found without any trouble. Is the extra ending on the library possibly to blame? I tried removing it but still just got CAVE_cave_LIBRARY-NOTFOUND. I can use CMake to set it manually, but then I still have to go change code generation settings to match. I also need to manually set the /MTd switch in order to avoid link errors. Is there an easy way to do that from the CMakeLists.txt file so that it's automatic? Is there an easy way to choose which setting to use and have the appropriate library linked against?
Thanks,
Chris
--
Chris Scharver
Electronic Visualization Laboratory
The University of Illinois at Chicago
Ph: 312-996-3002 FAX: 312-413-7585
From jw@mip.informatik.uni-kiel.de Thu May 8 17:55:52 2003
From: jw@mip.informatik.uni-kiel.de (Jan Woetzel)
Date: Thu, 08 May 2003 18:55:52 +0200
Subject: [Cmake] three questions: CMAKE_BUILD_TYPE, BUILD_SHARED_LIBS, SEND_ERROR
in MESSAGE
Message-ID: <3EBA8C18.7060305@mip.informatik.uni-kiel.de>
Hi,
first of all some flowers to the cmake developers:
I like cmake, it is very useful to me for developing under Linux with
gcc and porting my projects to MS Windows (.Net).
Three questions arised:
(1)
I can switch between static and shared build with BUILD_SHARED_LIBS
or adding STATIC/SHARED to ADD_LIBRARY.
Then one of tehm is built.
But I want to build both, static and shared library.
I tried:
ADD_LIBRARY( MYLIB STATIC MY_SRC)
ADD_LIBRARY( MYLIB SHARED MY_SRC)
But only one one is generated.
--What's the right way to go?
(2)
I want to set my CMAKE_BUILD_TYPE to some default which is not 'None',
but for example "Debug"
I tried:
SET(CMAKE_BUILD_TYPE "Debug")
(with and without quotes around Debug).
But nothing changes running ccmake (version 1.6 patch5 under Linux).
I can 'manually' enter the value Debug, but the *_DEBUG flags are not
used. The *_RELEASE flags are used.
--How do I set the default build in my CMakeLists.txt ?
(3)
I want to echo some info (for my own debugging).which shouldn'f affect
teh build process.
Any MESSAGE, e.g. like
MESSAGE ("hello")
is treated as ERROR - although I didn't add SEND_ERROR.
I expected this behaviour just for
MESSGAE (SEND_ERROR "something is wrong")
--Is this a (known) bug in MESSAGE, exists a workaround?
Best regards,
Jan.
From bill.hoffman@kitware.com Mon May 12 02:39:59 2003
From: bill.hoffman@kitware.com (William A. Hoffman)
Date: Sun, 11 May 2003 21:39:59 -0400
Subject: [Cmake] three questions: CMAKE_BUILD_TYPE,
BUILD_SHARED_LIBS, SEND_ERROR in MESSAGE
In-Reply-To: <3EBA8C18.7060305@mip.informatik.uni-kiel.de>
Message-ID: <5.1.0.14.0.20030511213603.03aa47c8@pop.biz.rr.com>
At 06:55 PM 5/8/2003 +0200, Jan Woetzel wrote:
>Hi,
>
>first of all some flowers to the cmake developers:
>I like cmake, it is very useful to me for developing under Linux with gcc
>and porting my projects to MS Windows (.Net).
>
>
>Three questions arised:
>
>(1)
>I can switch between static and shared build with BUILD_SHARED_LIBS
>or adding STATIC/SHARED to ADD_LIBRARY.
>Then one of tehm is built.
>But I want to build both, static and shared library.
>I tried:
>
>ADD_LIBRARY( MYLIB STATIC MY_SRC)
>ADD_LIBRARY( MYLIB SHARED MY_SRC)
>
>But only one one is generated.
>--What's the right way to go?
>
1. The only way to do this would be to put each library in a separate
SUBDIR directory.
>(2)
>I want to set my CMAKE_BUILD_TYPE to some default which is not 'None',
>but for example "Debug"
>
>I tried:
>SET(CMAKE_BUILD_TYPE "Debug")
>(with and without quotes around Debug).
>
>But nothing changes running ccmake (version 1.6 patch5 under Linux).
>I can 'manually' enter the value Debug, but the *_DEBUG flags are not
>used. The *_RELEASE flags are used.
>
>--How do I set the default build in my CMakeLists.txt ?
The SET command will work, but your cache will still not have the value.
You can use a SET(CACHE FORCE) to change the cache.
>(3)
>I want to echo some info (for my own debugging).which shouldn'f affect teh
>build process.
>Any MESSAGE, e.g. like
> MESSAGE ("hello")
>is treated as ERROR - although I didn't add SEND_ERROR.
>I expected this behaviour just for
>MESSGAE (SEND_ERROR "something is wrong")
>
>--Is this a (known) bug in MESSAGE, exists a workaround?
MESSAGE is not treated as an error. ccmake currently reports it as an error
but does not treat it like an error. SEND_ERROR causes ccmake and CMakeSetup
to not go on to the generate stage.
-Bill
>Best regards,
>Jan.
>
>_______________________________________________
>Cmake mailing list
>Cmake@public.kitware.com
>http://public.kitware.com/mailman/listinfo/cmake
From bill.hoffman@kitware.com Mon May 12 02:46:16 2003
From: bill.hoffman@kitware.com (William A. Hoffman)
Date: Sun, 11 May 2003 21:46:16 -0400
Subject: [Cmake] Reconciling MD, MDd, MT, MTd libraries
In-Reply-To:
Message-ID: <5.1.0.14.0.20030511214322.03a43830@pop.biz.rr.com>
That should work.
You can do something like this so that it will
find any one of them:
FIND_LIBRARY(CAVE_cave_LIBRARY
NAMES cave_ogl_mt_MDd cave_ogl_mt_MD cave_ogl_mt_MT cave_ogl_mt_MTd
PATHS ${CAVE_HOME}/lib32
DOC "Path to cave library")
-Bill
At 04:50 PM 5/7/2003 -0500, Chris Scharver wrote:
>Hello,
>
>I am running CMake 1.6.6, and it cannot automatically locate my library
>file. There are several different libraries available for different code
>generation styles, and I wonder if that's why things are a little strange.
>
>C:\VRCO\CAVELib_3.0.3\lib32\libcave_ogl_mt_[MD MDd MT MTd].lib
>
>No matter which library I try to have CMake find, the file is never found.
>The result is always CAVE_cave_LIBRARY-NOTFOUND.
>
> SET (CAVE_HOME "$ENV{CAVE_HOME}")
> FIND_LIBRARY(CAVE_cave_LIBRARY cave_ogl_mt_MTd
> ${CAVE_HOME}/lib32
> }
>
>I know the environment variable is being read properly because the header
>is found without any trouble. Is the extra ending on the library possibly
>to blame? I tried removing it but still just got
>CAVE_cave_LIBRARY-NOTFOUND. I can use CMake to set it manually, but then I
>still have to go change code generation settings to match. I also need to
>manually set the /MTd switch in order to avoid link errors. Is there an
>easy way to do that from the CMakeLists.txt file so that it's automatic?
>Is there an easy way to choose which setting to use and have the
>appropriate library linked against?
>
>Thanks,
>Chris
>--
>Chris Scharver
>Electronic Visualization Laboratory
>The University of Illinois at Chicago
>Ph: 312-996-3002 FAX: 312-413-7585
>
>_______________________________________________
>Cmake mailing list
>Cmake@public.kitware.com
>http://public.kitware.com/mailman/listinfo/cmake
From filipe.gr.sousa@alunos.ipb.pt Mon May 12 13:40:01 2003
From: filipe.gr.sousa@alunos.ipb.pt (Filipe Sousa)
Date: Mon, 12 May 2003 13:40:01 +0100
Subject: [Cmake] three questions: CMAKE_BUILD_TYPE, BUILD_SHARED_LIBS, SEND_ERROR in MESSAGE
In-Reply-To: <3EBA8C18.7060305@mip.informatik.uni-kiel.de>
References: <3EBA8C18.7060305@mip.informatik.uni-kiel.de>
Message-ID: <200305121340.02051.filipe.gr.sousa@alunos.ipb.pt>
On Thursday 08 May 2003 17:55, Jan Woetzel wrote:
> Hi,
>
> first of all some flowers to the cmake developers:
> I like cmake, it is very useful to me for developing under Linux with
> gcc and porting my projects to MS Windows (.Net).
>
>
> Three questions arised:
>
> (1)
> I can switch between static and shared build with BUILD_SHARED_LIBS
> or adding STATIC/SHARED to ADD_LIBRARY.
> Then one of tehm is built.
> But I want to build both, static and shared library.
> I tried:
>
> ADD_LIBRARY( MYLIB STATIC MY_SRC)
> ADD_LIBRARY( MYLIB SHARED MY_SRC)
>
> But only one one is generated.
> --What's the right way to go?
>
>
>
>
> (2)
> I want to set my CMAKE_BUILD_TYPE to some default which is not 'None',
> but for example "Debug"
>
> I tried:
> SET(CMAKE_BUILD_TYPE "Debug")
> (with and without quotes around Debug).
>
> But nothing changes running ccmake (version 1.6 patch5 under Linux).
> I can 'manually' enter the value Debug, but the *_DEBUG flags are not
> used. The *_RELEASE flags are used.
>
> --How do I set the default build in my CMakeLists.txt ?
>
>
>
> (3)
> I want to echo some info (for my own debugging).which shouldn'f affect
> teh build process.
> Any MESSAGE, e.g. like
> MESSAGE ("hello")
> is treated as ERROR - although I didn't add SEND_ERROR.
> I expected this behaviour just for
> MESSGAE (SEND_ERROR "something is wrong")
>
> --Is this a (known) bug in MESSAGE, exists a workaround?
MESSAGE(STATUS "hello")
>
>
> Best regards,
> Jan.
>
> _______________________________________________
> Cmake mailing list
> Cmake@public.kitware.com
> http://public.kitware.com/mailman/listinfo/cmake
--
Filipe Sousa
From cody@accessdata.com Mon May 12 16:02:28 2003
From: cody@accessdata.com (Cody Batt)
Date: Mon, 12 May 2003 09:02:28 -0600
Subject: [Cmake] Object File Location
Message-ID: <3EBFB784.4000301@accessdata.com>
Hello,
We would like to generate all of our object files into a separate
directory. As it is, the object files are generated in the same
directory as the source files. I've been looking in the CMAKE docs, but
I can't find a way to specify the object file location. Has anyone else
done this with CMAKE?
Cody
From jw@mip.informatik.uni-kiel.de Mon May 12 16:13:12 2003
From: jw@mip.informatik.uni-kiel.de (Jan Woetzel)
Date: Mon, 12 May 2003 17:13:12 +0200
Subject: [Cmake] Object File Location
In-Reply-To: <3EBFB784.4000301@accessdata.com>
References: <3EBFB784.4000301@accessdata.com>
Message-ID: <3EBFBA08.20402@mip.informatik.uni-kiel.de>
Cody Batt wrote:
> We would like to generate all of our object files into a separate
> directory. As it is, the object files are generated in the same
> directory as the source files. I've been looking in the CMAKE docs,
> but I can't find a way to specify the object file location. Has
> anyone else done this with CMAKE?
There are at least two solutions:
(1)
Use one directory for source (e.g. ~/src) and another for builds ,
go to the build directory (e.g. ~/build) and run cmake with the source
directory as argument.
cd ~/build
cmake ~/src
compile in ~/build as usual
(2)
You can set a subdir for objects, libs and binaries with:
LIBRARY_OUTPUT_PATH
and
EXECUTABLE_OUTPUT_PATH
Regards,
Jan.
--
Dipl.-Ing. Jan Woetzel
--------------------------------------------------
University of Kiel
From scharver@evl.uic.edu Tue May 13 03:12:40 2003
From: scharver@evl.uic.edu (Chris Scharver)
Date: Mon, 12 May 2003 21:12:40 -0500
Subject: [Cmake] Reconciling MD, MDd, MT, MTd libraries
In-Reply-To: <5.1.0.14.0.20030511214322.03a43830@pop.biz.rr.com>
References: <5.1.0.14.0.20030511214322.03a43830@pop.biz.rr.com>
Message-ID:
At 9:46 PM -0400 05/11/2003, William A. Hoffman wrote:
>You can do something like this so that it will
>find any one of them:
>
>FIND_LIBRARY(CAVE_cave_LIBRARY
>NAMES cave_ogl_mt_MDd cave_ogl_mt_MD cave_ogl_mt_MT cave_ogl_mt_MTd
>PATHS ${CAVE_HOME}/lib32
>DOC "Path to cave library")
Thanks for the response. In trying again, I noticed that I forgot to add the 'lib' prefix. (The files are named libcave_ogl_mt...) Too many conflicts in library-naming paradigms across platforms... ;)
What about the other question? Is there an easier way to change the Visual Studio code generation type without manually changing each /MD in the cache to another value? Maybe something like a central code-generation OPTION() that would only appear when generating Visual Studio workspaces. Based on that variable, the appropriate /MD, /MT, etc. switch would get added automatically to the CMAKE_CXX_FLAGS variables.
Thanks,
Chris
--
Chris Scharver
Electronic Visualization Laboratory
The University of Illinois at Chicago
Ph: 312-996-3002 FAX: 312-413-7585
From dlacct@hotmail.com Mon May 12 22:03:28 2003
From: dlacct@hotmail.com (DL)
Date: Mon, 12 May 2003 17:03:28 -0400
Subject: [Cmake] Dependency on generated file
Message-ID:
I have a file, called generator.c, which when compiled and run, dumps to std
out a bit of C code that I need to redirect to a file - out.c, for example.
The file out.c is then #included into another file, lib.c, which I want to
compile into a static library.
How do I create a dependency between the generated file out.c and the
generating executable made from generator.c so that they're built first and
lib.c can be properly compiled?
In other words,
generator.c => generator
generator ==> out.c
out.c, lib.c == > libmylib
Thanks,
DL
From ken.martin@kitware.com Tue May 13 14:05:04 2003
From: ken.martin@kitware.com (Ken Martin)
Date: Tue, 13 May 2003 09:05:04 -0400
Subject: [Cmake] Dependency on generated file
In-Reply-To:
Message-ID: <003901c31950$4881cc90$6601a8c0@kitwarein.com>
> I have a file, called generator.c, which when compiled and run, dumps
to
> std
> out a bit of C code that I need to redirect to a file - out.c, for
> example.
> The file out.c is then #included into another file, lib.c, which I
want to
> compile into a static library.
>
> How do I create a dependency between the generated file out.c and the
> generating executable made from generator.c so that they're built
first
> and
> lib.c can be properly compiled?
>
> In other words,
> generator.c => generator
> generator ==> out.c
> out.c, lib.c == > libmylib
Custom commands are being cleaned up right now but the basic idea is the
same. Without testing I think the following is the basic idea:
# add the executable
ADD_EXECUTABLE(generator generator.c)
# mark out.c as generated
SET_SOURCE_FILE_PROPERTIES(${PROJECT_BINARY_DIR}/out.c GENERATED 1)
# add a post build command (confusing...)
ADD_CUSTOM_COMMAND(
SOURCE generator
COMMAND ${EXECUTABLE_OUTPUT_PATH}/generator${EXE_EXTENSION}
ARGS your_args
TARGET generator
OUTPUTS ${PROJECT_BINARY_DIR}/out.c)
# add the library
ADD_LIBRARY(mylib lib.c ${PROJECT_BINARY_DIR}/out.c)
# make sure the generator is built before mylib
ADD_DEPENDENCIES(mylib generator)
- Ken
>
> Thanks,
> DL
> _______________________________________________
> Cmake mailing list
> Cmake@public.kitware.com
> http://public.kitware.com/mailman/listinfo/cmake
From ken.martin@kitware.com Tue May 13 14:18:03 2003
From: ken.martin@kitware.com (Ken Martin)
Date: Tue, 13 May 2003 09:18:03 -0400
Subject: [Cmake] Dependency on generated file
In-Reply-To: <003901c31950$4881cc90$6601a8c0@kitwarein.com>
Message-ID: <003b01c31952$190f4670$6601a8c0@kitwarein.com>
> # add a post build command (confusing...)
> ADD_CUSTOM_COMMAND(
> SOURCE generator
> COMMAND ${EXECUTABLE_OUTPUT_PATH}/generator${EXE_EXTENSION}
> ARGS your_args
> TARGET generator
> OUTPUTS ${PROJECT_BINARY_DIR}/out.c)
Actually to be safe on windows you need to put CMAKE_CFG_INTDIR in the
path to the exec as well. So the command should be closer to..
COMMAND
${EXECUTABLE_OUTPUT_PATH}/${CMAKE_CFG_INTDIR}/generator${EXE_EXTENSION}
From prochon@stelvio.com Tue May 13 16:01:19 2003
From: prochon@stelvio.com (Patrick Rochon)
Date: Tue, 13 May 2003 11:01:19 -0400
Subject: [Cmake] Debug or Release
Message-ID: <002f01c31960$831aeac0$6801a8c0@stelvio.net>
Hi,
I am building a .DLL.
I am using cmake to generate VisualStudio7 project files.
Is it possible to link with a .lib in Debug and a different .lib in Release
?
I think that is possible using something like this :
IF(DEBUG)
LINK_LIBRARIES(rxmd.lib)
ELSE(DEBUG)
LINK_LIBRARIES(rxmr.lib)
ENDIF(DEBUG)
But, I can't find the exact "IF" command.
Thank you!
Patrick Rochon
Code Administrator
Stelvio inc.
From rmathews@envoyww.com Tue May 13 16:36:14 2003
From: rmathews@envoyww.com (Robert Mathews)
Date: Tue, 13 May 2003 11:36:14 -0400
Subject: [Cmake] Debug or Release
References: <002f01c31960$831aeac0$6801a8c0@stelvio.net>
Message-ID: <007101c31965$63261fa0$c3d3403f@victory>
Have you tried something like this:
# These are the default libraries for everyone, then for
# debug and then for release mode
LINK_LIBRARIES(
advapi32.lib
kernel32.lib
ole32.lib
oleaut32.lib
rpcndr.lib
rpcns4.lib
rpcrt4.lib
uuid.lib
adimgr.lib
agm.lib
cta.lib
ics2api.lib
imagehlp.lib
isdnapi.lib
lib44bsd.lib
libtiff.lib
Msacm32.lib
nccapi.lib
netapi32.lib
nfxmgr.lib
swimgr.lib
vcemgr.lib
ws2_32.lib
debug
des.lib
libbind.lib
optimized
des_release.lib
libbind_release.lib
)
----- Original Message -----
From: "Patrick Rochon"
To:
Sent: Tuesday, May 13, 2003 11:01 AM
Subject: [Cmake] Debug or Release
> Hi,
>
> I am building a .DLL.
>
> I am using cmake to generate VisualStudio7 project files.
>
> Is it possible to link with a .lib in Debug and a different .lib in
Release
> ?
>
> I think that is possible using something like this :
>
> IF(DEBUG)
> LINK_LIBRARIES(rxmd.lib)
> ELSE(DEBUG)
> LINK_LIBRARIES(rxmr.lib)
> ENDIF(DEBUG)
>
> But, I can't find the exact "IF" command.
>
> Thank you!
>
> Patrick Rochon
> Code Administrator
> Stelvio inc.
>
>
> _______________________________________________
> Cmake mailing list
> Cmake@public.kitware.com
> http://public.kitware.com/mailman/listinfo/cmake
>
From bill.hoffman@kitware.com Tue May 13 16:50:43 2003
From: bill.hoffman@kitware.com (Bill Hoffman)
Date: Tue, 13 May 2003 11:50:43 -0400
Subject: [Cmake] Debug or Release
In-Reply-To: <002f01c31960$831aeac0$6801a8c0@stelvio.net>
Message-ID: <5.2.0.9.0.20030513114822.0432cdd8@pop.biz.rr.com>
Try this:
TARGET_LINK_LIBRARIES(mylib debug rxmd)
TARGET_LINK_LIBRARIES(mylib optimized rxmd)
Where mylib is your library or executable that you added
with an ADD_LIBRARY or ADD_EXECUTABLE.
At the time cmake is running there is no Debug or Release since
cmake generates both at the same time.
-Bill
At 11:01 AM 5/13/2003, Patrick Rochon wrote:
>Hi,
>
>I am building a .DLL.
>
>I am using cmake to generate VisualStudio7 project files.
>
>Is it possible to link with a .lib in Debug and a different .lib in Release
>?
>
>I think that is possible using something like this :
>
>IF(DEBUG)
> LINK_LIBRARIES(rxmd.lib)
>ELSE(DEBUG)
> LINK_LIBRARIES(rxmr.lib)
>ENDIF(DEBUG)
>
>But, I can't find the exact "IF" command.
>
>Thank you!
>
>Patrick Rochon
>Code Administrator
>Stelvio inc.
>
>
>_______________________________________________
>Cmake mailing list
>Cmake@public.kitware.com
>http://public.kitware.com/mailman/listinfo/cmake
From john.biddiscombe@mirada-solutions.com Thu May 15 08:50:20 2003
From: john.biddiscombe@mirada-solutions.com (John Biddiscombe)
Date: Thu, 15 May 2003 08:50:20 +0100
Subject: [Cmake] how to detect a macro
Message-ID: <4926A5BE4AFE7C4A83D5CF5CDA7B775407128F@oxcore01.mirada-solutions.com>
If one does
MACRO(MyMacro p1 p2 p3)
do stuff here
ENDMACRO(MyMacro)
...
later I want to do
IF (MyMacro)
MyMAcro("1" "2" "3")
ENDIF (MyMacro)
But it never executes because IF(macroname) always returns false. I need =
to put a macro in an INCLUDE(OPTIONAL) file, and conditionally execute =
it in another.
Is there a way to detect it's presence (apart from the slightly Klunky
SET(MYMACROPRESENT 1)
in the optional file
and=20
IF (MYMACROPRESENT)
MyMAcro("1" "2" "3")
ENDIF (MYMACROPRESENT)
thanks
JB
From john.biddiscombe@mirada-solutions.com Thu May 15 09:16:06 2003
From: john.biddiscombe@mirada-solutions.com (John Biddiscombe)
Date: Thu, 15 May 2003 09:16:06 +0100
Subject: [Cmake] how to detect a macro
Message-ID: <4926A5BE4AFE7C4A83D5CF5CDA7B7754071290@oxcore01.mirada-solutions.com>
OK ignore this question, I like the (Klunky) version after all. It =
allows better control when doing this in several subdirs (I think)
> IF (MYMACROPRESENT)
> MyMacro("1" "2" "3")
> ENDIF (MYMACROPRESENT)
JB
> -----Original Message-----
> From: John Biddiscombe=20
> Sent: 15 May 2003 08:50
> To: cmake@public.kitware.com
> Subject: [Cmake] how to detect a macro
>=20
>=20
> If one does
>=20
> MACRO(MyMacro p1 p2 p3)
> do stuff here
> ENDMACRO(MyMacro)
>=20
> ...
>=20
> later I want to do
>=20
> IF (MyMacro)
> MyMAcro("1" "2" "3")
> ENDIF (MyMacro)
>=20
> But it never executes because IF(macroname) always returns=20
> false. I need to put a macro in an INCLUDE(OPTIONAL) file,=20
> and conditionally execute it in another.
>=20
> Is there a way to detect it's presence (apart from the slightly Klunky
> SET(MYMACROPRESENT 1)
> in the optional file
>=20
> and=20
> IF (MYMACROPRESENT)
> MyMAcro("1" "2" "3")
> ENDIF (MYMACROPRESENT)
>=20
> thanks
>=20
> JB
> _______________________________________________
> Cmake mailing list
> Cmake@public.kitware.com
> http://public.kitware.com/mailman/listinfo/cmake
>=20
From wheeler@crd.ge.com Thu May 15 14:42:08 2003
From: wheeler@crd.ge.com (Wheeler, Frederick W (Research))
Date: Thu, 15 May 2003 09:42:08 -0400
Subject: [Cmake] FindXerces.cmake
Message-ID:
What is the criteria/process for putting a new .cmake file in the CMake
Modules directory? I wrote a FindXerces.cmake, below, and think it is
simple enough to just go in. I started with FindZLIB.cmake.
Fred Wheeler
#
# Find the native Xerces includes and library
#
# XERCES_INCLUDE_DIR - where to find dom/dom.hpp, etc.
# XERCES_LIBRARIES - List of fully qualified libraries to link against
when using Xerces.
# XERCES_FOUND - Do not attempt to use Xerces if "no" or undefined.
FIND_PATH(XERCES_INCLUDE_DIR dom/dom.hpp
/usr/local/include
/usr/include
)
# There may be some API changes between Xerces 1.x and 2.x
# I'm not sure how to deal with that in a .cmake file
# Perhaps it should be up to the application to figure out the version and
# API specifics from macros set in the headers?
FIND_LIBRARY(XERCES_LIBRARY
NAMES
xerces-c_2
xerces-c_2D
xerces-c_1
xerces-c_1D
PATHS
/usr/local/lib
/usr/lib
)
IF(XERCES_INCLUDE_DIR)
IF(XERCES_LIBRARY)
SET( XERCES_LIBRARIES ${XERCES_LIBRARY} )
SET( XERCES_FOUND "YES" )
ENDIF(XERCES_LIBRARY)
ENDIF(XERCES_INCLUDE_DIR)
From wheeler@crd.ge.com Thu May 15 14:49:48 2003
From: wheeler@crd.ge.com (Wheeler, Frederick W (Research))
Date: Thu, 15 May 2003 09:49:48 -0400
Subject: [Cmake] directory search order in .cmake files
Message-ID:
Modules/FindZLIB.cmake contains these commands:
FIND_PATH(ZLIB_INCLUDE_DIR zlib.h
/usr/local/include
/usr/include
)
FIND_LIBRARY(ZLIB_LIBRARY z
/usr/lib
/usr/local/lib
)
Shouldn't the FIND_LIBRARY command be changed to this?
FIND_LIBRARY(ZLIB_LIBRARY z
/usr/local/lib
/usr/lib
)
Seems to me that one should first check /usr/local and then /usr so that
libraries in /usr/local override whatever comes with the OS. Also, the
FIND_PATH for the includes and the FIND_LIBRARY should be consistent,
otherwise you risk version mismatch between the headers and libraries.
Fred Wheeler
From wheeler@crd.ge.com Thu May 15 20:16:58 2003
From: wheeler@crd.ge.com (Wheeler, Frederick W (Research))
Date: Thu, 15 May 2003 15:16:58 -0400
Subject: [Cmake] RE: FindXerces.cmake
Message-ID:
I've greatly enhanced the FindXerces.cmake file I sent to this list this
morning. The new version is below.
If there should be a FindXerces.cmake in CMake/Modules, and this could be
it, let me know if there is anything else I can do to it.
-Fred Wheeler
#
# Find the native Xerces includes and library
#
# XERCES_INCLUDE_DIR - where to find dom/dom.hpp, etc.
# XERCES_LIBRARIES - List of fully qualified libraries to link against
when using Xerces.
# XERCES_VERSION_1 - Use this Version
# XERCES_VERSION_2 - Use this Version
# XERCES_FOUND - Do not attempt to use Xerces if "no" or undefined.
OPTION(XERCES_VERSION_1 "Use Xerces version 1.x" 0)
OPTION(XERCES_VERSION_2 "Use Xerces version 2.x" 1)
# Exclusion between the versions
IF(XERCES_VERSION_2)
SET(XERCES_VERSION_1 0)
ENDIF(XERCES_VERSION_2)
FIND_PATH(XERCES_INCLUDE_DIR dom/dom.hpp
/usr/local/include
/usr/include
)
# Make sure that the Xerces include path has been set
# So the XERCES_LIBRARY does not appear the first time
IF(XERCES_INCLUDE_DIR)
IF(XERCES_VERSION_1)
FIND_LIBRARY(XERCES_LIBRARY
NAMES
xerces-c_1
xerces-c_1D
PATHS
/usr/local/lib
/usr/lib
${XERCES_INCLUDE_DIR}/../lib
)
ENDIF(XERCES_VERSION_1)
IF(XERCES_VERSION_2)
FIND_LIBRARY(XERCES_LIBRARY
NAMES
xerces-c_2
xerces-c_2D
PATHS
/usr/local/lib
/usr/lib
${XERCES_INCLUDE_DIR}/../lib
)
ENDIF(XERCES_VERSION_2)
ENDIF(XERCES_INCLUDE_DIR)
IF(XERCES_INCLUDE_DIR)
IF(XERCES_LIBRARY)
SET( XERCES_LIBRARIES ${XERCES_LIBRARY} )
SET( XERCES_FOUND "YES" )
ENDIF(XERCES_LIBRARY)
ENDIF(XERCES_INCLUDE_DIR)
MARK_AS_ADVANCED(
XERCES_VERSION_1
XERCES_VERSION_2
)
From wheeler@crd.ge.com Thu May 15 20:42:07 2003
From: wheeler@crd.ge.com (Wheeler, Frederick W (Research))
Date: Thu, 15 May 2003 15:42:07 -0400
Subject: [Cmake] finding matching includes and libraries
Message-ID:
I have a suggestion on how to find matching includes and libraries in .cmake
files. This is extracted from FindZLIB.cmake:
FIND_PATH(ZLIB_INCLUDE_DIR zlib.h
/usr/local/include
/usr/include
)
FIND_LIBRARY(ZLIB_LIBRARY z
/usr/local/lib
/usr/lib # (I reordered this line)
)
The potential problem with this is that if there is a broken installation of
zlib in /usr/local/include that has headers but no library, then one might
end up using the headers from /usr/local/include and the library from
/usr/lib.
A potential solution is below. With this method, we only look for a library
if the header is found and we only look for the library in the same install
prefix as the header. An added benefit is that there is only one list of
install locations.
FIND_PATH(ZLIB_INCLUDE_DIR zlib.h
/usr/local/include
/usr/include
)
IF(ZLIB_INCLUDE_DIR)
FIND_LIBRARY(ZLIB_LIBRARY z
${ZLIB_INCLUDE_DIR}/../lib
)
ENDIF(ZLIB_INCLUDE_DIR)
I'm not sure about the portability of "${ZLIB_INCLUDE_DIR}/../lib". Perhaps
there is a better way to strip off the "/include" from ZLIB_INCLUDE_DIR in
CMake.
-Fred Wheeler
From brad.king@kitware.com Thu May 15 21:44:53 2003
From: brad.king@kitware.com (Brad King)
Date: Thu, 15 May 2003 16:44:53 -0400 (EDT)
Subject: [Cmake] finding matching includes and libraries
In-Reply-To:
Message-ID:
Hi Fred,
> IF(ZLIB_INCLUDE_DIR)
> FIND_LIBRARY(ZLIB_LIBRARY z
> ${ZLIB_INCLUDE_DIR}/../lib
> )
> ENDIF(ZLIB_INCLUDE_DIR)
>
> I'm not sure about the portability of "${ZLIB_INCLUDE_DIR}/../lib".
> Perhaps there is a better way to strip off the "/include" from
> ZLIB_INCLUDE_DIR in CMake.
This is safe and works well. Several other find modules do this already.
FindZLIB was one of the first modules, and hasn't been updated.
-Brad
From prochon@stelvio.com Fri May 16 14:10:58 2003
From: prochon@stelvio.com (Patrick Rochon)
Date: Fri, 16 May 2003 09:10:58 -0400
Subject: [Cmake] How to preform different "PostBuildEvent" in debug and release
Message-ID: <007d01c31bac$984d1820$6801a8c0@stelvio.net>
Hi,
I am using cmake to generate Visual Studio 7 project files.
I would like to know how to preform a "PostBuildEvent" in debug and a
different one in release. Is it possible ?
For example, after building :
c:\dev\rts\debug\rts.dll
I want to copy it in :
c:\dev\binaries\rtsD.dll
and after building :
c:\dev\rts\release\rts.dll
I want to copy it in :
c:\dev\binaries\rtsR.dll
Thanks!
Patrick
Code Administrator
Stelvio inc.
From bill.hoffman@kitware.com Fri May 16 16:52:36 2003
From: bill.hoffman@kitware.com (Bill Hoffman)
Date: Fri, 16 May 2003 11:52:36 -0400
Subject: [Cmake] How to preform different "PostBuildEvent" in debug
and release
In-Reply-To: <007d01c31bac$984d1820$6801a8c0@stelvio.net>
Message-ID: <5.2.0.9.0.20030516114928.043a65b8@pop.biz.rr.com>
There is no way to do that directly. However, what you could
do is use the ${CMAKE_CFG_INTDIR} variable in the command, which
will evaluate to Release/Debug/ when the custom command is run.
Then use a .bat file to copy the file based on the value of IntDir.
-Bill
At 09:10 AM 5/16/2003, Patrick Rochon wrote:
>Hi,
>
>I am using cmake to generate Visual Studio 7 project files.
>
>I would like to know how to preform a "PostBuildEvent" in debug and a
>different one in release. Is it possible ?
>
>For example, after building :
>c:\dev\rts\debug\rts.dll
>I want to copy it in :
>c:\dev\binaries\rtsD.dll
>
>and after building :
>c:\dev\rts\release\rts.dll
>I want to copy it in :
>c:\dev\binaries\rtsR.dll
>
>Thanks!
>
>Patrick
>Code Administrator
>Stelvio inc.
>
>_______________________________________________
>Cmake mailing list
>Cmake@public.kitware.com
>http://public.kitware.com/mailman/listinfo/cmake
From dcthomp@sandia.gov Fri May 16 23:46:57 2003
From: dcthomp@sandia.gov (David Thompson)
Date: 16 May 2003 15:46:57 -0700
Subject: [Cmake] hardcoded prefix for QT_WRAP_UI inputs
Message-ID: <1053125217.18878.6859.camel@fozzy.ran.sandia.gov>
Hi,
I have a QT ui file, blah.ui, that I've created using CONFIGURE_FILE so
that some example text gets set to the CMAKE_INSTALL_PREFIX path.
Because the file is generated and placed in my out-of-source build
project_BINARY_DIR, the Makefile complains that there is no rule to make
${project_SOURCE_DIR}/blah.ui. If I try to specify the full path,
${project_BINARY_DIR}/blah.ui, to QT_WRAP_UI the Makefile looks for
${project_SOURCE_DIR}/${project_BINARY_DIR}/blah.ui ... is this a
feature?
Thanks,
David
From vanessa.noguesruiz@epfl.ch Mon May 19 09:47:52 2003
From: vanessa.noguesruiz@epfl.ch (vanessa.noguesruiz@epfl.ch)
Date: Mon, 19 May 2003 10:47:52 +0200 (MEST)
Subject: [Cmake] (no subject)
Message-ID: <1053334072.3ec89a385e5a8@imapwww.epfl.ch>
Hi everybody.
Does anybody know how may I include vtk libraries in a makefile I've already
got implemented? I think I don't need Cmake because I've already got a makefile
but, then, how may I include VTK libraries in it?
As well as, I am trying to install VTK under cygwin and ccmake gives me the
following error:
CMake Error: cmake version 1.4
Usage: /cmake [srcdir] [options]
Where cmake is run from the directory where you want the object files
written. If srcdir is not specified, the current directory is used for both
source and object files.
Options are:
-i (puts cmake in wizard mode, not available for ccmake)
-DVAR:TYPE=VALUE (create a cache file entry)
-Cpath_to_initial_cache (a cmake list file that is used to pre-load the cache
with values.)
[-GgeneratorName] (where generator name can be one of these: "Unix Makefiles"
)
Something else... I am working on Cygwin, but I was doing it under Windows
before, so I had installed VTK and CMake versions for Windows. Now I guesss I
have to uninstall them and install Linux versions for working under Cygwin and
with g++ compiler (I work with C++). Am I wrong?
Thanks in advance and sorry for this long mail, that's it, I'm a bit lost in
Cygwin's platform. :(
Vane
From ian.m.scott@stud.man.ac.uk Mon May 19 10:28:31 2003
From: ian.m.scott@stud.man.ac.uk (Ian Scott)
Date: Mon, 19 May 2003 10:28:31 +0100
Subject: [Cmake] hardcoded prefix for QT_WRAP_UI inputs
In-Reply-To: <1053125217.18878.6859.camel@fozzy.ran.sandia.gov>
Message-ID: <000001c31de9$03ffb120$22ea5882@smb.man.ac.uk>
> From: David Thompson [mailto:dcthomp@sandia.gov]
...
> I have a QT ui file, blah.ui, that I've created using
> CONFIGURE_FILE so
> that some example text gets set to the CMAKE_INSTALL_PREFIX path.
> Because the file is generated and placed in my out-of-source build
> project_BINARY_DIR, the Makefile complains that there is no
> rule to make
> ${project_SOURCE_DIR}/blah.ui. If I try to specify the full path,
> ${project_BINARY_DIR}/blah.ui, to QT_WRAP_UI the Makefile looks for
> ${project_SOURCE_DIR}/${project_BINARY_DIR}/blah.ui ... is this a
> feature?
Probably not.
The QT wrapping commands in cmake were not written by the main CMake
authors, and since they don't use them, you'll have to put up the QT mods
authors' less than ideal response....
Franck (author of QT_WARP_UI) and I (author of QT_WRAP_CPP) do not currently
have the time to maintain/enhance these commands beyond their basic
operation - with which there hasn't been any problem recently. We suggest
looking at the CMake Source. The QT_WRAP_* commands are really quite simple.
Sorry we can't be more help.
Ian.
From Ken.Weinert@ihs.com Fri May 16 16:47:17 2003
From: Ken.Weinert@ihs.com (Ken Weinert)
Date: Fri, 16 May 2003 09:47:17 -0600
Subject: [Cmake] multiple compile of an object
Message-ID: <20030516154717.GB29033@ihs.com>
--oC1+HKm2/end4ao3
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable
OK, I thought I had this one solved, but it turns out that it isn't
quite right.
My situation is this:=20
I frequently write a little test program into my class
definitions. That is, at the end of the file where all the methods
have been defined you'll find something like this:=20
#if defined(_MAIN_NEEDED)
int main()
{
// test code goes here
}
#endif
And then I'll compile a small standalone test program and also
compile the .o to be included in the main program.
Is there a way to automate this in CMake? If so, a pointer to
where it might be documented (or an idea of how to do it) would be
appreciated.=20
Thank you.
--=20
/~\ The ASCII Ken Weinert Ken.Weinert@ihs.com=20
\ / Ribbon Campaign 303-858-6956 (V) 303-705-4258 (F)
X Against HTML GnuPG: 9274F1CE GnuPG available at http://www.gnupg.o=
rg/
/ \ Email! 1D87 3720 BB77 4489 A928 79D6 F8EC DD76 9274 F1CE
Giving up on assembly language was the apple in our Garden of Eden:
Languages whose use squanders machine cycles are sinful. The LISP
machine now permits LISP programmers to abandon bra and fig-leaf. --
Alan J. Perlis
--oC1+HKm2/end4ao3
Content-Type: application/pgp-signature
Content-Disposition: inline
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.1 (GNU/Linux)
iD8DBQE+xQgF+OzddpJ08c4RArF+AJ49vtVRQokI3jJtCznQmVfrj6D12gCgggdl
JudOlHF4YkoXUURY9moii6o=
=6359
-----END PGP SIGNATURE-----
--oC1+HKm2/end4ao3--
From vanessa.noguesruiz@epfl.ch Mon May 19 14:25:17 2003
From: vanessa.noguesruiz@epfl.ch (vanessa.noguesruiz@epfl.ch)
Date: Mon, 19 May 2003 15:25:17 +0200 (MEST)
Subject: [Cmake] (no subject)
Message-ID: <1053350717.3ec8db3d6e45c@imapwww.epfl.ch>
Does anybody know why Cmake gives me the following error when trying to install
vtk under cygwin?
CMake Error: Error in cmake code at /home/van/LocalSoft/LocalCVS/vtk/vtk-
4.2.2/Common/CMakeLists.txt:223 Unknown CMake command "VTK_MAKE_INSTANTIATOR2",
the same error for Filtering,Imaging and Graphics CMakeLists.txt.
Thanks in advance.
From brad.king@kitware.com Mon May 19 14:31:58 2003
From: brad.king@kitware.com (Brad King)
Date: Mon, 19 May 2003 09:31:58 -0400 (EDT)
Subject: [Cmake] (no subject)
In-Reply-To: <1053334072.3ec89a385e5a8@imapwww.epfl.ch>
Message-ID:
> Does anybody know how may I include vtk libraries in a makefile I've
> already got implemented? I think I don't need Cmake because I've already
> got a makefile but, then, how may I include VTK libraries in it?
Once you run "make install" on VTK from the build tree you can just add
the proper -I... -L... and -l... flags in your makefile.
> As well as, I am trying to install VTK under cygwin and ccmake gives me
> the following error:
>
>
> CMake Error: cmake version 1.4
VTK 4.2 requires CMake version 1.6. Please upgrade your cmake.
-Brad
From andy.cedilnik@kitware.com Mon May 19 15:05:10 2003
From: andy.cedilnik@kitware.com (Andy Cedilnik)
Date: 19 May 2003 10:05:10 -0400
Subject: [Cmake] multiple compile of an object
In-Reply-To: <20030516154717.GB29033@ihs.com>
References: <20030516154717.GB29033@ihs.com>
Message-ID: <1053353110.1196.220.camel@andoria>
Hi Ken,
You can always write a macro that writes a file and creates executable.
For example, let say you have several classes per library. You would do
something like this:
MACRO(CREATE_TEST classname libname)
STRING(ASCII 35 POUND)
WRITE_FILE("${CMAKE_CURRENT_BINARY_DIR}/test_${classname}.cxx"
"${POUND}define ${classname}_MAIN_NEEDED\n${POUND}include
<${CLASSNAME}.h>")
ADD_EXECUTABLE(test_${classname}
"${CMAKE_CURRENT_BINARY_DIR}/test_${classname}.cxx")
TARGET_LINK_LIBRARIES(test_${classname} ${libname})
ENDMACRO(CREATE_TEST)
Sounds good?
Then you just do:
CREATE_TEST(class1 mylib1)
CREATE_TEST(class2 mylib1)
CREATE_TEST(class3 mylib2)
Andy
On Fri, 2003-05-16 at 11:47, Ken Weinert wrote:
> OK, I thought I had this one solved, but it turns out that it isn't
> quite right.
>
> My situation is this:
>
> I frequently write a little test program into my class
> definitions. That is, at the end of the file where all the methods
> have been defined you'll find something like this:
>
> #if defined(_MAIN_NEEDED)
> int main()
> {
> // test code goes here
> }
> #endif
>
> And then I'll compile a small standalone test program and also
> compile the .o to be included in the main program.
>
> Is there a way to automate this in CMake? If so, a pointer to
> where it might be documented (or an idea of how to do it) would be
> appreciated.
From dcthomp@sandia.gov Mon May 19 16:55:39 2003
From: dcthomp@sandia.gov (David Thompson)
Date: 19 May 2003 08:55:39 -0700
Subject: [Cmake] hardcoded prefix for QT_WRAP_UI inputs
In-Reply-To: <000001c31de9$03ffb120$22ea5882@smb.man.ac.uk>
References: <000001c31de9$03ffb120$22ea5882@smb.man.ac.uk>
Message-ID: <1053359738.19390.3.camel@fozzy.ran.sandia.gov>
--=-uoEEaC/dn5IoofSdIqIL
Content-Type: text/plain
Content-Transfer-Encoding: 7bit
On Mon, 2003-05-19 at 02:28, Ian Scott wrote:
> > From: David Thompson [mailto:dcthomp@sandia.gov]
> > ... I have a QT ui file, blah.ui, that I've created using
> > CONFIGURE_FILE so
> > that some example text gets set to the CMAKE_INSTALL_PREFIX path.
> > Because the file is generated and placed in my out-of-source build
> > project_BINARY_DIR, the Makefile complains that there is no
> > rule to make
> > ${project_SOURCE_DIR}/blah.ui. If I try to specify the full path,
> > ${project_BINARY_DIR}/blah.ui, to QT_WRAP_UI the Makefile looks for
> > ${project_SOURCE_DIR}/${project_BINARY_DIR}/blah.ui ... is this a
> > feature?
>
> Probably not.
>
> The QT wrapping commands in cmake were not written by the main CMake
> authors, and since they don't use them, you'll have to put up the QT mods
> authors' less than ideal response....
>
> Franck (author of QT_WARP_UI) and I (author of QT_WRAP_CPP) do not currently
> have the time to maintain/enhance these commands beyond their basic
> operation - with which there hasn't been any problem recently. We suggest
> looking at the CMake Source. The QT_WRAP_* commands are really quite simple.
>
> Sorry we can't be more help.
Would anyone object to the attached patch? It retains original behavior
unless
1. The filename of the UI file begins with a '/', indicating that it's
an absolute filename and should not have the current source directory
prepended, OR
2. The UI file has the GENERATED property set, indicating that it will
be in the binary directory, not the source directory.
David
--=-uoEEaC/dn5IoofSdIqIL
Content-Description:
Content-Disposition: inline; filename=qt_moc.patch
Content-Type: text/x-patch; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
Index: Source/cmQTWrapUICommand.cxx
===================================================================
RCS file: /cvsroot/CMake/CMake/Source/cmQTWrapUICommand.cxx,v
retrieving revision 1.12
diff -r1.12 cmQTWrapUICommand.cxx
84c84,99
< std::string origname = cdir + "/" + *j;
---
> std::string origname;
> if ( (*j)[0] == '/' )
> {
> origname = *j;
> }
> else
> {
> if ( curr->GetPropertyAsBool("GENERATED") )
> {
> origname = std::string( m_Makefile->GetCurrentOutputDirectory() ) + "/" + *j;
> }
> else
> {
> origname = cdir + "/" + *j;
> }
> }
--=-uoEEaC/dn5IoofSdIqIL--
From vanessa.noguesruiz@epfl.ch Tue May 20 09:50:43 2003
From: vanessa.noguesruiz@epfl.ch (vanessa.noguesruiz@epfl.ch)
Date: Tue, 20 May 2003 10:50:43 +0200 (MEST)
Subject: [Cmake] upgrade to cmake >1.6.1
Message-ID: <1053420643.3ec9ec638aa7a@imapwww.epfl.ch>
Does anybody know where can I find a mirror where downloading a later version
of cmake? It is that I am attempting to install vtk-4.2.2 and I get the
following error:
CMake error:
Command not found: "VTK_MAKE_INSTANTIATOR2" in CMakeLists.txt of
Common,Filtering Graphics and Imaging
Thanks in advance
From Mathieu.Malaterre@creatis.insa-lyon.fr Tue May 20 09:46:26 2003
From: Mathieu.Malaterre@creatis.insa-lyon.fr (Mathieu Malaterre)
Date: Tue, 20 May 2003 10:46:26 +0200
Subject: [Cmake] upgrade to cmake >1.6.1
In-Reply-To: <1053420643.3ec9ec638aa7a@imapwww.epfl.ch>
References: <1053420643.3ec9ec638aa7a@imapwww.epfl.ch>
Message-ID: <3EC9EB62.3010105@creatis.insa-lyon.fr>
http://www.cmake.org/HTML/Download.html
vanessa.noguesruiz@epfl.ch wrote:
> Does anybody know where can I find a mirror where downloading a later version
> of cmake? It is that I am attempting to install vtk-4.2.2 and I get the
> following error:
>
> CMake error:
> Command not found: "VTK_MAKE_INSTANTIATOR2" in CMakeLists.txt of
> Common,Filtering Graphics and Imaging
>
>
> Thanks in advance
> _______________________________________________
> Cmake mailing list
> Cmake@public.kitware.com
> http://public.kitware.com/mailman/listinfo/cmake
>
--
Mathieu Malaterre
CREATIS
28 Avenue du Doyen LEPINE
B.P. Lyon-Montchat
69394 Lyon Cedex 03
http://www.creatis.insa-lyon.fr/~malaterre/
From Mathieu.Malaterre@creatis.insa-lyon.fr Tue May 20 17:29:47 2003
From: Mathieu.Malaterre@creatis.insa-lyon.fr (Mathieu Malaterre)
Date: Tue, 20 May 2003 18:29:47 +0200
Subject: [Cmake] vtkRenderWindowInteractor::GetClassName
Message-ID: <3ECA57FB.8080404@creatis.insa-lyon.fr>
Hi all,
I am trying to build my own program:
http://www.creatis.insa-lyon.fr/~malaterre/wxVTK/
with VC++ 6, but I failed to -sig- !
Here is what I get:
Linking...
wxVTKRenderWindowInteractor.obj : error LNK2001:
unresolved external symbol "public: virtual char const
* __thiscall
vtkRenderWindowInteractor::GetClassName(void)const "
(?GetClassName@vtkRenderWindowInteractor@@UBEPBDXZ)
Release/Sample_gtk.exe : fatal error LNK1120: 1
unresolved externals
Error executing link.exe
Does this means something to any of you ?
Thanks,
mathieu
--
Mathieu Malaterre
CREATIS
28 Avenue du Doyen LEPINE
B.P. Lyon-Montchat
69394 Lyon Cedex 03
http://www.creatis.insa-lyon.fr/~malaterre/
From bill.hoffman@kitware.com Tue May 20 18:22:50 2003
From: bill.hoffman@kitware.com (Bill Hoffman)
Date: Tue, 20 May 2003 13:22:50 -0400
Subject: [Cmake] vtkRenderWindowInteractor::GetClassName
In-Reply-To: <3ECA57FB.8080404@creatis.insa-lyon.fr>
Message-ID: <5.2.0.9.0.20030520132036.043513e0@pop.biz.rr.com>
It means you are not linking the library that contains
vtkRenderWindowInteractor::GetClassName into Sample_gtk.exe.
I guess you would have to give more information about
your CMakeLists.txt files. Have you looked the
cmakelist files in VTK/Examples/Tutorial?
-Bill
At 12:29 PM 5/20/2003, Mathieu Malaterre wrote:
>Hi all,
> I am trying to build my own program:
>http://www.creatis.insa-lyon.fr/~malaterre/wxVTK/
>
>with VC++ 6, but I failed to -sig- !
>
>Here is what I get:
>Linking...
>wxVTKRenderWindowInteractor.obj : error LNK2001:
>unresolved external symbol "public: virtual char const
>* __thiscall
>vtkRenderWindowInteractor::GetClassName(void)const "
>(?GetClassName@vtkRenderWindowInteractor@@UBEPBDXZ)
>Release/Sample_gtk.exe : fatal error LNK1120: 1
>unresolved externals
>Error executing link.exe
>
>Does this means something to any of you ?
>
>Thanks,
>mathieu
>
>--
>Mathieu Malaterre
>CREATIS
>28 Avenue du Doyen LEPINE
>B.P. Lyon-Montchat
>69394 Lyon Cedex 03 http://www.creatis.insa-lyon.fr/~malaterre/
>
>
>_______________________________________________
>Cmake mailing list
>Cmake@public.kitware.com
>http://public.kitware.com/mailman/listinfo/cmake
From Mathieu.Malaterre@creatis.insa-lyon.fr Tue May 20 18:46:54 2003
From: Mathieu.Malaterre@creatis.insa-lyon.fr (Mathieu Malaterre)
Date: Tue, 20 May 2003 19:46:54 +0200
Subject: [Cmake] vtkRenderWindowInteractor::GetClassName
In-Reply-To: <5.2.0.9.0.20030520132036.043513e0@pop.biz.rr.com>
References: <5.2.0.9.0.20030520132036.043513e0@pop.biz.rr.com>
Message-ID: <3ECA6A0E.60503@creatis.insa-lyon.fr>
Hum, the weird thing is that it compiles without no problem under Linux...
Anyway here is what I use:
SET(SRCS_Sample
Sample/Sample.cpp
wxVTKRenderWindowInteractor.cxx
)
LINK_LIBRARIES(
vtkCommon
vtkGraphics
vtkRendering
)
ADD_EXECUTABLE(Sample WIN32
${SRCS_Sample})
TARGET_LINK_LIBRARIES(Sample ${WXWINDOWS_LIBRARY})
What am doing wrong ?
Thanks,
mathieu
Bill Hoffman wrote:
> It means you are not linking the library that contains
> vtkRenderWindowInteractor::GetClassName into Sample_gtk.exe.
> I guess you would have to give more information about
> your CMakeLists.txt files. Have you looked the
> cmakelist files in VTK/Examples/Tutorial?
>
> -Bill
>
>
From bill.hoffman@kitware.com Tue May 20 18:57:24 2003
From: bill.hoffman@kitware.com (Bill Hoffman)
Date: Tue, 20 May 2003 13:57:24 -0400
Subject: [Cmake] vtkRenderWindowInteractor::GetClassName
In-Reply-To: <3ECA6A0E.60503@creatis.insa-lyon.fr>
References: <5.2.0.9.0.20030520132036.043513e0@pop.biz.rr.com>
<5.2.0.9.0.20030520132036.043513e0@pop.biz.rr.com>
Message-ID: <5.2.0.9.0.20030520135501.04456ff0@pop.biz.rr.com>
You maybe running into DLL problems with windows.
Is this a sub-class of a vtk class:
wxVTKRenderWindowInteractor
If so, it needs to export/import the right dll stuff.
See this example:
Examples/Build/vtkMy/Common
-Bill
At 01:46 PM 5/20/2003, Mathieu Malaterre wrote:
>Hum, the weird thing is that it compiles without no problem under Linux...
> Anyway here is what I use:
>
>SET(SRCS_Sample
>Sample/Sample.cpp
>wxVTKRenderWindowInteractor.cxx
>)
>
>LINK_LIBRARIES(
> vtkCommon
> vtkGraphics
> vtkRendering
>)
>
>
>ADD_EXECUTABLE(Sample WIN32
> ${SRCS_Sample})
>
>TARGET_LINK_LIBRARIES(Sample ${WXWINDOWS_LIBRARY})
>
>What am doing wrong ?
>
>Thanks,
>mathieu
>
>Bill Hoffman wrote:
>>It means you are not linking the library that contains
>>vtkRenderWindowInteractor::GetClassName into Sample_gtk.exe.
>>I guess you would have to give more information about
>>your CMakeLists.txt files. Have you looked the
>>cmakelist files in VTK/Examples/Tutorial?
>>-Bill
From Stefan.Schmidt@sophia.inria.fr Wed May 21 09:24:08 2003
From: Stefan.Schmidt@sophia.inria.fr (Stefan Schmidt)
Date: Wed, 21 May 2003 10:24:08 +0200
Subject: [Cmake] Re: [Dart] Configuration command, Symlinks in paths
In-Reply-To: <5.2.0.9.0.20030520091332.042ff928@pop.biz.rr.com>
References: <5.2.0.9.0.20030520091332.042ff928@pop.biz.rr.com>
Message-ID: <200305211024.08178.Stefan.Schmidt@sophia.inria.fr>
On Tuesday 20 May 2003 15:19, Bill Hoffman wrote:
> I don't think you can count on this behavior. It would really depend
> on the project. There is no guarantee that all paths in the resulting
> makefiles will preserve symlinks. Many of the commands in cmake will
> find the absolute path. Future versions of cmake are not even
> guaranteed to keep this same behavior, as it was unintended in the fist
> place.
Why is it unintended?=20
I really need this feature! There are symlinks in my filesystem pointing =
to=20
automounted NFS paths, which are not the same on each machine. But as I w=
ant=20
to use the compiled binaries on multiple machines, the compilation needs =
to=20
keep the common symlink in the path (it's an absolut path, not a relative=
=20
one). Otherwise the rpath in the binaries would be wrong, and when compil=
ing=20
ITK, for example, even the configuration files ITKConfig.cmake would be=20
wrong.
Stefan
>
> -Bill
>
> At 05:16 AM 5/20/2003, Stefan Schmidt wrote:
> >Dear Dart maintainers,
> >
> >could we please change the Dart configuration file Utility.conf.in to =
call
> >CMake with the -H and -B options? If CMake is called like this, it
> > preserves the symlinks in the paths:
> >
> >ConfigureCommand: "@CMAKE_COMMAND@" -H@PROJECT_SOURCE_DIR@
> >-B@PROJECT_BINARY_DIR@
> >
> >Thanks!
> >Stefan Schmidt
> >_______________________________________________
> >Dart mailing list
> >Dart@public.kitware.com
> >http://public.kitware.com/mailman/listinfo/dart
From Peter.Vanroose@esat.kuleuven.ac.be Wed May 21 09:57:47 2003
From: Peter.Vanroose@esat.kuleuven.ac.be (Peter Vanroose)
Date: Wed, 21 May 2003 10:57:47 +0200 (CEST)
Subject: [Cmake] Re: [Dart] Configuration command, Symlinks in paths
In-Reply-To: <200305211024.08178.Stefan.Schmidt@sophia.inria.fr>
References: <5.2.0.9.0.20030520091332.042ff928@pop.biz.rr.com>
<200305211024.08178.Stefan.Schmidt@sophia.inria.fr>
Message-ID:
> I really need this feature! There are symlinks in my filesystem pointing to
> automounted NFS paths, which are not the same on each machine.
Completely agree. I had a similar problem, not with my own symlinks, but
with NFS mounting on /tmp_mnt/something and symlinking this to /something
on the CMake build machine. While on other machines there was no /tmp_mnt
Hence lots of Dart errors ...
-- Peter.
From vanessa.noguesruiz@epfl.ch Wed May 21 10:49:42 2003
From: vanessa.noguesruiz@epfl.ch (vanessa.noguesruiz@epfl.ch)
Date: Wed, 21 May 2003 11:49:42 +0200 (MEST)
Subject: [Cmake] where apply commands cmake +make install for upgradint to 1.6.7 in Cygwin
Message-ID: <1053510582.3ecb4bb674600@imapwww.epfl.ch>
Hi everybody. Anyone knows where may I apply the commands to get a succesfull
upgrade under Cygwin?
From bill.hoffman@kitware.com Wed May 21 13:42:03 2003
From: bill.hoffman@kitware.com (Bill Hoffman)
Date: Wed, 21 May 2003 08:42:03 -0400
Subject: [Cmake] Re: [Dart] Configuration command, Symlinks in paths
In-Reply-To: <200305211024.08178.Stefan.Schmidt@sophia.inria.fr>
References: <5.2.0.9.0.20030520091332.042ff928@pop.biz.rr.com>
<5.2.0.9.0.20030520091332.042ff928@pop.biz.rr.com>
Message-ID: <5.2.0.9.0.20030521083051.04435420@pop.biz.rr.com>
I guess what I am saying, is that you are getting lucky.
There has been no attempt in cmake to use relative paths, and
the -B -H is an unexpected behavior. CMake depends on absolute paths,
and the method it uses to find absolute paths will unfortunately
remove symlinks as well. So, if you use any other libraries on
the system, they will be found by cmake in the realpath. As long
as the project is self contained, I guess the -B -H options will
work, but they are not tested.
I have been thinking that perhaps some sort of path translation table might
be a good feature. This would be a file that could tell cmake, when you see
this path, always replace it with this. Then, relative paths could be
done as well.
-Bill
At 04:24 AM 5/21/2003, Stefan Schmidt wrote:
>On Tuesday 20 May 2003 15:19, Bill Hoffman wrote:
>> I don't think you can count on this behavior. It would really depend
>> on the project. There is no guarantee that all paths in the resulting
>> makefiles will preserve symlinks. Many of the commands in cmake will
>> find the absolute path. Future versions of cmake are not even
>> guaranteed to keep this same behavior, as it was unintended in the fist
>> place.
>
>Why is it unintended?
>I really need this feature! There are symlinks in my filesystem pointing to
>automounted NFS paths, which are not the same on each machine. But as I want
>to use the compiled binaries on multiple machines, the compilation needs to
>keep the common symlink in the path (it's an absolut path, not a relative
>one). Otherwise the rpath in the binaries would be wrong, and when compiling
>ITK, for example, even the configuration files ITKConfig.cmake would be
>wrong.
>
>Stefan
>
>>
>> -Bill
>>
>> At 05:16 AM 5/20/2003, Stefan Schmidt wrote:
>> >Dear Dart maintainers,
>> >
>> >could we please change the Dart configuration file Utility.conf.in to call
>> >CMake with the -H and -B options? If CMake is called like this, it
>> > preserves the symlinks in the paths:
>> >
>> >ConfigureCommand: "@CMAKE_COMMAND@" -H@PROJECT_SOURCE_DIR@
>> >-B@PROJECT_BINARY_DIR@
>> >
>> >Thanks!
>> >Stefan Schmidt
>> >_______________________________________________
>> >Dart mailing list
>> >Dart@public.kitware.com
>> >http://public.kitware.com/mailman/listinfo/dart
>
>_______________________________________________
>Dart mailing list
>Dart@public.kitware.com
>http://public.kitware.com/mailman/listinfo/dart
From bill.hoffman@kitware.com Wed May 21 13:42:52 2003
From: bill.hoffman@kitware.com (Bill Hoffman)
Date: Wed, 21 May 2003 08:42:52 -0400
Subject: [Cmake] Re: [Dart] Configuration command, Symlinks in paths
In-Reply-To:
References: <200305211024.08178.Stefan.Schmidt@sophia.inria.fr>
<5.2.0.9.0.20030520091332.042ff928@pop.biz.rr.com>
<200305211024.08178.Stefan.Schmidt@sophia.inria.fr>
Message-ID: <5.2.0.9.0.20030521084209.04455280@pop.biz.rr.com>
We fix the /tmp_mnt problem explicitly in cmake for the sgi, and have
done so since cmake 1.2.
-Bill
At 04:57 AM 5/21/2003, Peter Vanroose wrote:
>> I really need this feature! There are symlinks in my filesystem pointing to
>> automounted NFS paths, which are not the same on each machine.
>
>Completely agree. I had a similar problem, not with my own symlinks, but
>with NFS mounting on /tmp_mnt/something and symlinking this to /something
>on the CMake build machine. While on other machines there was no /tmp_mnt
>Hence lots of Dart errors ...
>
>
>-- Peter.
>_______________________________________________
>Dart mailing list
>Dart@public.kitware.com
>http://public.kitware.com/mailman/listinfo/dart
From vanessa.noguesruiz@epfl.ch Wed May 21 14:42:02 2003
From: vanessa.noguesruiz@epfl.ch (vanessa.noguesruiz@epfl.ch)
Date: Wed, 21 May 2003 15:42:02 +0200 (MEST)
Subject: [Cmake] vtk-4.2.2 installation under cygwin
Message-ID: <1053524521.3ecb822a0280f@imapwww.epfl.ch>
Hi everybody.
When trying to install vtk-4.2.2 (linux version)under cygwin I get the
following error:
CMake error: error in cmake code at
/home/van/localsoft/localcvs/vtk/vtk-4.2.2/Common/CMakeLists.txt:223:
Unknown CMake Command "VTK_MAKE_INSTANTIATOR2()"
The same with the rest of directories: Graphics, Imaging...
Does anybody know how can I solve it??
Thanks a lot
From ken.martin@kitware.com Wed May 21 14:53:27 2003
From: ken.martin@kitware.com (Ken Martin)
Date: Wed, 21 May 2003 09:53:27 -0400
Subject: [Cmake] CMake version 1.6.7 Released
Message-ID: <002e01c31fa0$5ec764a0$6601a8c0@kitwarein.com>
We are pleased to announce the release of CMake version 1.6.7. This
patch includes a number of bug fixes and support for Visual Studio 2003.
It can be downloaded from www.cmake.org A synopsis of the changes
follows:
Added support for Visual Studio 2003. Fixed a bug where LINK_FLAGS were
not getting passed to Visual Studio generators. Added a fix for MipsPro
7.3. Fix for C++ object file rule for nmake. A fix for search paths in
the FindCable and FindFLTK modules. A fix for the TRY_COMPILE command
when make -I is used. A fix to support very long lines in CmakeList
files. Improved the MACRO command to provide warnings whan a MACRO is
not properly closed with a matching END_MACRO. Fixed the REMOVE command
to not ignore the first argument. Fixed the STRING command to be
inherited. Fixed some keyboard issues with ccmake on the SGI. Fix to
ccmake to not report and error if non error message occur. Fixed some
flags for C++ shared libs on SunOS and win32 executables on Borland. A
fix in the implementation of the CheckIncludeFiles.cmake module.
Improved error messages when a bad generator was selected. More robust
CheckSymbolExists.cmake module.
Thanks
Ken
Ken Martin, PhD
Kitware Inc.
518 371 3971 x101
469 Clifton Corporate Pkwy
Clifton Park NY 12065
From wheeler@crd.ge.com Wed May 21 15:24:45 2003
From: wheeler@crd.ge.com (Wheeler, Frederick W (Research))
Date: Wed, 21 May 2003 10:24:45 -0400
Subject: [Cmake] Adjusting compiler optimization for msvc
Message-ID:
Is there something analogous to this for msvc?
IF(CMAKE_COMPILER_IS_GNUCXX)
SET_SOURCE_FILES_PROPERTIES(test_mil.cxx COMPILE_FLAGS -O0)
ENDIF(CMAKE_COMPILER_IS_GNUCXX)
I'd like to disable optimization for one file in VXL when compiling with
msvc 6.0. With optimization, msvc produces bad code for one particular
file (vxl_src/core/vil2/vil2_bicub_interp.txx). It's a txx file. Would I
actually use a SET_SOURCE_FILES_PROPERTIES command on the .cxx files in the
Templates directory?
Thanks,
Fred Wheeler
From bill.hoffman@kitware.com Wed May 21 15:46:54 2003
From: bill.hoffman@kitware.com (Bill Hoffman)
Date: Wed, 21 May 2003 10:46:54 -0400
Subject: [Cmake] Adjusting compiler optimization for msvc
In-Reply-To:
Message-ID: <5.2.0.9.0.20030521103507.04355a20@pop.biz.rr.com>
Yes, I think you would have to set the source file properties on
the .cxx files from the template directory.
It would be something like this:
AUX_SOURCE_DIRECTORY(dir VARIABLE)
IF(CMAKE_SYSTEM_AND_CXX_COMPILER_INFO_FILE MATCHES "Windows-cl")
FOREACH( f ${sources} )
IF(${f} MATCHES vil2_bicub_interp)
SET_SOURCE_FILES_PROPERTIES(${f} COMPILE_FLAGS -Od)
ENDIF(${f} MATCHES vil2_bicub_interp)
ENDFOREACH(f)
ENDIF(CMAKE_SYSTEM_AND_CXX_COMPILER_INFO_FILE MATCHES "Windows-cl")
At 10:24 AM 5/21/2003, Wheeler, Frederick W (Research) wrote:
>Is there something analogous to this for msvc?
>
>IF(CMAKE_COMPILER_IS_GNUCXX)
> SET_SOURCE_FILES_PROPERTIES(test_mil.cxx COMPILE_FLAGS -O0)
>ENDIF(CMAKE_COMPILER_IS_GNUCXX)
>
>I'd like to disable optimization for one file in VXL when compiling with
>msvc 6.0. With optimization, msvc produces bad code for one particular
>file (vxl_src/core/vil2/vil2_bicub_interp.txx). It's a txx file. Would I
>actually use a SET_SOURCE_FILES_PROPERTIES command on the .cxx files in the
>Templates directory?
>
>Thanks,
>Fred Wheeler
>_______________________________________________
>Cmake mailing list
>Cmake@public.kitware.com
>http://public.kitware.com/mailman/listinfo/cmake
From andy.cedilnik@kitware.com Wed May 21 16:23:21 2003
From: andy.cedilnik@kitware.com (Andy Cedilnik)
Date: 21 May 2003 11:23:21 -0400
Subject: [Cmake] Re: [vtkusers] vtk-4.2.2 installation under cygwin
In-Reply-To: <1053524521.3ecb822a0280f@imapwww.epfl.ch>
References: <1053524521.3ecb822a0280f@imapwww.epfl.ch>
Message-ID: <1053530601.1195.406.camel@andoria>
Hi Vanessa,
What version of CMake are you using?
When saying (linux version) under cygwin, what does that mean? Did you
download source code?
Make sure to use the newest CMake. The current version is 1.6.7 and that
one should not have problems that you are seeing.
Andy
On Wed, 2003-05-21 at 09:42, vanessa.noguesruiz@epfl.ch wrote:
> Hi everybody.
> When trying to install vtk-4.2.2 (linux version)under cygwin I get the
> following error:
> CMake error: error in cmake code at
> /home/van/localsoft/localcvs/vtk/vtk-4.2.2/Common/CMakeLists.txt:223:
> Unknown CMake Command "VTK_MAKE_INSTANTIATOR2()"
> The same with the rest of directories: Graphics, Imaging...
> Does anybody know how can I solve it??
> Thanks a lot
From bill.hoffman@kitware.com Wed May 21 18:08:20 2003
From: bill.hoffman@kitware.com (Bill Hoffman)
Date: Wed, 21 May 2003 13:08:20 -0400
Subject: [Cmake] Re: [vtkusers] vtk-4.2.2 installation under cygwin
In-Reply-To: <1053530601.1195.406.camel@andoria>
References: <1053524521.3ecb822a0280f@imapwww.epfl.ch>
<1053524521.3ecb822a0280f@imapwww.epfl.ch>
Message-ID: <5.2.0.9.0.20030521130616.0445dda8@pop.biz.rr.com>
In the CMakeError.log file sent to me by vanessa.noguesruiz@epfl.ch,
there was the following message:
Building of VTK extensions failed with the following output:
Building dependencies. cmake.depends...
/usr/local/bin/cmake: not found
make: *** [cmake.depends] Error 127
The problem seems to be a bad installation of cmake.
How was cmake installed? Do you have a /usr/local/bin/cmake executable?
-Bill
At 11:23 AM 5/21/2003, Andy Cedilnik wrote:
>Hi Vanessa,
>
>What version of CMake are you using?
>When saying (linux version) under cygwin, what does that mean? Did you
>download source code?
>
>Make sure to use the newest CMake. The current version is 1.6.7 and that
>one should not have problems that you are seeing.
>
> Andy
>
>On Wed, 2003-05-21 at 09:42, vanessa.noguesruiz@epfl.ch wrote:
>> Hi everybody.
>> When trying to install vtk-4.2.2 (linux version)under cygwin I get the
>> following error:
>> CMake error: error in cmake code at
>> /home/van/localsoft/localcvs/vtk/vtk-4.2.2/Common/CMakeLists.txt:223:
>> Unknown CMake Command "VTK_MAKE_INSTANTIATOR2()"
>> The same with the rest of directories: Graphics, Imaging...
>> Does anybody know how can I solve it??
>> Thanks a lot
>
>
>_______________________________________________
>Cmake mailing list
>Cmake@public.kitware.com
>http://public.kitware.com/mailman/listinfo/cmake
From Mathieu.Malaterre@creatis.insa-lyon.fr Fri May 23 08:10:51 2003
From: Mathieu.Malaterre@creatis.insa-lyon.fr (Mathieu Malaterre)
Date: Fri, 23 May 2003 09:10:51 +0200
Subject: [Cmake] Improvements of FindwxWindows.cmake
Message-ID: <3ECDC97B.7010004@creatis.insa-lyon.fr>
Hi all,
I have been hacking FindwxWindows.cmake to fit my needs. My hacked
version could be find:
http://www.creatis.insa-lyon.fr/~malaterre/cmake/FindwxWindows.cmake
Changes:
- Now take into account wx debug library
- Add needed preprocessors definition (according to
http://www.wxwindows.org/technote/vc_ide.htm)
BTW I didn't really understand why WXDialog uses:
CMake/Source/WXDialog/wxincludes.h
what was the issue compared to an:
ADD_DEFINITION(-D__WXMSW__ -DWINVER=0x0400 ...)
Now what I want is that the library (between debug and release) is
choosen at compile time (instead of an option in CMakeSetup), is it
possible ? Could you give me some info about how to do that, thanks.
Thanks,
mathieu
From bill.hoffman@kitware.com Fri May 23 14:06:52 2003
From: bill.hoffman@kitware.com (Bill Hoffman)
Date: Fri, 23 May 2003 09:06:52 -0400
Subject: [Cmake] Improvements of FindwxWindows.cmake
In-Reply-To: <3ECDC97B.7010004@creatis.insa-lyon.fr>
Message-ID: <5.2.0.9.0.20030523090140.04625138@pop.biz.rr.com>
At 03:10 AM 5/23/2003, Mathieu Malaterre wrote:
>Hi all,
> I have been hacking FindwxWindows.cmake to fit my needs. My hacked version could be find:
>
>http://www.creatis.insa-lyon.fr/~malaterre/cmake/FindwxWindows.cmake
>
>Changes:
>- Now take into account wx debug library
>- Add needed preprocessors definition (according to
>http://www.wxwindows.org/technote/vc_ide.htm)
>
> BTW I didn't really understand why WXDialog uses:
>CMake/Source/WXDialog/wxincludes.h
>
>what was the issue compared to an:
>ADD_DEFINITION(-D__WXMSW__ -DWINVER=0x0400 ...)
As a general rule, putting definitions in header files is better that on the command line.
It allows other projects to use the software without having to construct complicated
command lines. If wxwindows itself had this approach, you only have to worry about
getting the correct -I options, and that would be it. Often times it is a complicated
process to construct the -D options, and scripts are provided to be used in makefiles.
If a .h file is configured, then it isolates the build system used and avoids users of
the library having to run a script or compute the -D options.
> Now what I want is that the library (between debug and release) is choosen at compile time (instead of an option in CMakeSetup), is it possible ? Could you give me some info about how to do that, thanks.
>
TARGET_LINK_LIBRARIES can take debug and optimized libraries:
TARGET_LINK_LIBRARIES(target library1
library2
...)
Specify a list of libraries to be linked into the specified target
The debug and optimized strings may be used to indicate that
the next library listed is to be used only for that specific
type of build
-Bill
>Thanks,
>mathieu
>
>
>_______________________________________________
>Cmake mailing list
>Cmake@public.kitware.com
>http://public.kitware.com/mailman/listinfo/cmake
From dean.inglis@on.aibn.com Fri May 23 15:35:57 2003
From: dean.inglis@on.aibn.com (dean.inglis@on.aibn.com)
Date: Fri, 23 May 2003 10:35:57 -0400
Subject: [Cmake] (no subject)
Message-ID: <20030523143557.WNDM20954.tomts5-srv.bellnexxia.net@[209.226.175.135]>
I'm having some pb's with building MinSizeRel CMake 1.6.7
using Borland C++ Builder 6 (bcc32). I see that there is
support for many incarnations of VC and am hoping that
I can somehow assist with extending Borland support beyond
Borland free command line tools and BCB 5. First crack
at building CMake 1.6.7:
Building object file cmMakefile.obj...
Borland C++ 5.6 for Win32 Copyright (c) 1993, 2002 Borland
c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx:
Error E2268 c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx 74: Call to undefined function 'atoi' in function cmMakefile::GetCacheMajorVersion()
Error E2268 c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx 83: Call to undefined function 'atoi' in function cmMakefile::GetCacheMinorVersion()
Error E2268 c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx 393: Call to undefined function 'atof' in function cmMakefile::ConfigureFinalPass()
Warning W8004 c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx 407: 'versionValue' is assigned a value that is never used in function cmMakefile::ConfigureFinalPass()
Error E2268 c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx 1060: Call to undefined function 'getenv' in function cmMakefile::ExpandVariablesInString(_STL::string &,bool,bool) const
Error E2268 c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx 1394: Call to undefined function 'atof' in function cmMakefile::ExpandSourceListArguments(const _STL::vector<_STL::string,_STL::allocator<_STL::string> > &,_STL::vector<_STL::string,_STL::allocator<_STL::string> > &,unsigned int)
Warning W8004 c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx 1417: 'versionValue' is assigned a value that is never used in function cmMakefile::ExpandSourceListArguments(const _STL::vector<_STL::string,_STL::allocator<_STL::string> > &,_STL::vector<_STL::string,_STL::allocator<_STL::string> > &,unsigned int)
*** 5 errors in Compile ***
I had similar pbs with ITK and BCB6 (see post thread):
http://www.itk.org/pipermail/insight-users/2003-May/003642.html
Second crack by adding -D_USE_OLD_RW_STL (use old Rogue Wave implementation of STL)
to cache value CMAKE_CXX_FLAGS:STRING:
Borland C++ 5.6 for Win32 Copyright (c) 1993, 2002 Borland
c:\Builder\CMake_1.6.7\Source\cmDumpDocumentation.cxx:
Building executable c:\Builder\CMake_1.6.7\Source\DumpDocumentation.exe...
Borland C++ 5.6 for Win32 Copyright (c) 1993, 2002 Borland
Turbo Incremental Link 5.60 Copyright (c) 1997-2002 Borland
Error: Unresolved external 'cmMakefile::ExpandArguments(const _STL::vector >&, ...' referenced from C:\BUILDER\CMAKE_1.6.7\SOURCE\CMAKELIB.LIB|cmake
Error: Unresolved external 'GetPredefinedCommands(_STL::list >&)' referenced from C:\BUILDER\CMAKE_1.6.7\SOURCE\CMAKELIB.LIB|cmake
Error: Unresolved external 'cmVariableWatch::AddWatch(const _STL::basic_string, _STL::allocator >&, void (*)(const _STL::basic_string, _STL::allocator >&, int, void *), void *)' referenced from C:\BUILDER\CMAKE_1.6.7\SOURCE\CMAKELIB.LIB|cmake
Error: Unresolved external 'cmake::DumpDocumentationToFile(std::basic_ostream >&)' referenced from C:\BUILDER\CMAKE_1.6.7\SOURCE\CMDUMPDOCUMENTATION.OBJ
I can help out as required...
Dean
From bill.hoffman@kitware.com Fri May 23 15:54:31 2003
From: bill.hoffman@kitware.com (Bill Hoffman)
Date: Fri, 23 May 2003 10:54:31 -0400
Subject: [Cmake] (no subject)
In-Reply-To: <20030523143557.WNDM20954.tomts5-srv.bellnexxia.net@[209.22
6.175.135]>
Message-ID: <5.2.0.9.0.20030523104939.046356f8@pop.biz.rr.com>
Did you remove the binary tree for the second try?
Looks like it may still have some object files that are using
the other stl.
You could also look at how to call atoi, atof and getenv in the documentation
for borland.
If you can get it compiling and set up a dashboard to keep it compiling,
I will integrate patches that you send me.
-Bill
At 10:35 AM 5/23/2003, you wrote:
>I'm having some pb's with building MinSizeRel CMake 1.6.7
>using Borland C++ Builder 6 (bcc32). I see that there is
>support for many incarnations of VC and am hoping that
>I can somehow assist with extending Borland support beyond
>Borland free command line tools and BCB 5. First crack
>at building CMake 1.6.7:
>
>Building object file cmMakefile.obj...
>Borland C++ 5.6 for Win32 Copyright (c) 1993, 2002 Borland
>c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx:
>Error E2268 c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx 74: Call to undefined function 'atoi' in function cmMakefile::GetCacheMajorVersion()
>Error E2268 c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx 83: Call to undefined function 'atoi' in function cmMakefile::GetCacheMinorVersion()
>Error E2268 c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx 393: Call to undefined function 'atof' in function cmMakefile::ConfigureFinalPass()
>Warning W8004 c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx 407: 'versionValue' is assigned a value that is never used in function cmMakefile::ConfigureFinalPass()
>Error E2268 c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx 1060: Call to undefined function 'getenv' in function cmMakefile::ExpandVariablesInString(_STL::string &,bool,bool) const
>Error E2268 c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx 1394: Call to undefined function 'atof' in function cmMakefile::ExpandSourceListArguments(const _STL::vector<_STL::string,_STL::allocator<_STL::string> > &,_STL::vector<_STL::string,_STL::allocator<_STL::string> > &,unsigned int)
>Warning W8004 c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx 1417: 'versionValue' is assigned a value that is never used in function cmMakefile::ExpandSourceListArguments(const _STL::vector<_STL::string,_STL::allocator<_STL::string> > &,_STL::vector<_STL::string,_STL::allocator<_STL::string> > &,unsigned int)
>*** 5 errors in Compile ***
>
>I had similar pbs with ITK and BCB6 (see post thread):
>http://www.itk.org/pipermail/insight-users/2003-May/003642.html
>
>Second crack by adding -D_USE_OLD_RW_STL (use old Rogue Wave implementation of STL)
>to cache value CMAKE_CXX_FLAGS:STRING:
>
>Borland C++ 5.6 for Win32 Copyright (c) 1993, 2002 Borland
>c:\Builder\CMake_1.6.7\Source\cmDumpDocumentation.cxx:
>Building executable c:\Builder\CMake_1.6.7\Source\DumpDocumentation.exe...
>Borland C++ 5.6 for Win32 Copyright (c) 1993, 2002 Borland
>Turbo Incremental Link 5.60 Copyright (c) 1997-2002 Borland
>Error: Unresolved external 'cmMakefile::ExpandArguments(const _STL::vector >&, ...' referenced from C:\BUILDER\CMAKE_1.6.7\SOURCE\CMAKELIB.LIB|cmake
>Error: Unresolved external 'GetPredefinedCommands(_STL::list >&)' referenced from C:\BUILDER\CMAKE_1.6.7\SOURCE\CMAKELIB.LIB|cmake
>Error: Unresolved external 'cmVariableWatch::AddWatch(const _STL::basic_string, _STL::allocator >&, void (*)(const _STL::basic_string, _STL::allocator >&, int, void *), void *)' referenced from C:\BUILDER\CMAKE_1.6.7\SOURCE\CMAKELIB.LIB|cmake
>Error: Unresolved external 'cmake::DumpDocumentationToFile(std::basic_ostream >&)' referenced from C:\BUILDER\CMAKE_1.6.7\SOURCE\CMDUMPDOCUMENTATION.OBJ
>
>I can help out as required...
>
>Dean
>
>_______________________________________________
>Cmake mailing list
>Cmake@public.kitware.com
>http://public.kitware.com/mailman/listinfo/cmake
From pchandra@radonc.unc.edu Fri May 23 20:48:28 2003
From: pchandra@radonc.unc.edu (Parag Chandra)
Date: Fri, 23 May 2003 15:48:28 -0400
Subject: [Cmake] INSTALL_TARGETS command on Windows
Message-ID: <004001c32164$485cbfb0$99221398@radonc.radonc.unc.edu>
This is a multi-part message in MIME format.
------=_NextPart_000_0041_01C32142.C14B1FB0
Content-Type: text/plain;
charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
Hi,
=20
I am trying to get the final output of my build targets (i.e. the .lib,
.exe, and .dll files) to be placed into a single directory under =
Windows. I
tried setting LIBRARY_OUTPUT_PATH and EXECUTABLE_OUTPUT_PATH to the =
folder I
want, and this seems to work, but I also seem to get a lot of the
intermediate files thrown in there as well. For example, building =
foo.exe
causes foo.exe, .ilk, .exp, .lib, and .pdb to all be put in that =
directory,
but I just want foo.exe to be there. I tried using the INSTALL_TARGETS
command in one of my CmakeLists files, but this doesn't seem to do =
anything
under Windows. I know this is possible, because ITK does it, but I don't
understand how it's doing it. Does anyone know the answer?
=20
Best regards,
-Parag Chandra
------=_NextPart_000_0041_01C32142.C14B1FB0
Content-Type: text/html;
charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
Hi,
I am trying to get the final output of my build =
targets
(i.e. the .lib, .exe, and .dll files) to be placed into a single =
directory
under Windows. I tried setting LIBRARY_OUTPUT_PATH and =
EXECUTABLE_OUTPUT_PATH
to the folder I want, and this seems to work, but I also seem to get a =
lot of
the intermediate files thrown in there as well. For example, building =
foo.exe
causes foo.exe, .ilk, .exp, .lib, and .pdb to all be put in that =
directory, but
I just want foo.exe to be there. I tried using the INSTALL_TARGETS =
command in
one of my CmakeLists files, but this doesn’t seem to do anything =
under
Windows. I know this is possible, because ITK does it, but I don’t
understand how it’s doing it. Does anyone know the =
answer?
Best regards,
-Parag Chandra
------=_NextPart_000_0041_01C32142.C14B1FB0--
From dean.inglis@on.aibn.com Fri May 23 20:51:41 2003
From: dean.inglis@on.aibn.com (dean.inglis@on.aibn.com)
Date: Fri, 23 May 2003 15:51:41 -0400
Subject: [Cmake] BCB6 CMake 1.6.7 build
Message-ID: <20030523195141.KRWD12935.tomts19-srv.bellnexxia.net@[209.226.175.135]>
Hi Bill,
I deleted the binary tree and the build
proceeds without error, just a couple of
W8004 warnings: var assigned a value never used in...
I will add Hyoid (FYI, hyoid is a laptop and if
you know what the hyoid bone is you'll get the
joke :)) as an experimental BCB6 for CMake & VTK.
Any chance of getting BCB6 + ITK building with
CMake?
Dean
>
> From: Bill Hoffman
> Date: 2003/05/23 Fri AM 10:54:31 GMT-04:00
> To:
> CC: cmake@public.kitware.com
> Subject: Re: [Cmake] (no subject)
>
> Did you remove the binary tree for the second try?
> Looks like it may still have some object files that are using
> the other stl.
>
> You could also look at how to call atoi, atof and getenv in the documentation
> for borland.
>
> If you can get it compiling and set up a dashboard to keep it compiling,
> I will integrate patches that you send me.
>
>
> -Bill
>
>
> At 10:35 AM 5/23/2003, you wrote:
> >I'm having some pb's with building MinSizeRel CMake 1.6.7
> >using Borland C++ Builder 6 (bcc32). I see that there is
> >support for many incarnations of VC and am hoping that
> >I can somehow assist with extending Borland support beyond
> >Borland free command line tools and BCB 5. First crack
> >at building CMake 1.6.7:
> >
> >Building object file cmMakefile.obj...
> >Borland C++ 5.6 for Win32 Copyright (c) 1993, 2002 Borland
> >c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx:
> >Error E2268 c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx 74: Call to undefined function 'atoi' in function cmMakefile::GetCacheMajorVersion()
> >Error E2268 c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx 83: Call to undefined function 'atoi' in function cmMakefile::GetCacheMinorVersion()
> >Error E2268 c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx 393: Call to undefined function 'atof' in function cmMakefile::ConfigureFinalPass()
> >Warning W8004 c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx 407: 'versionValue' is assigned a value that is never used in function cmMakefile::ConfigureFinalPass()
> >Error E2268 c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx 1060: Call to undefined function 'getenv' in function cmMakefile::ExpandVariablesInString(_STL::string &,bool,bool) const
> >Error E2268 c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx 1394: Call to undefined function 'atof' in function cmMakefile::ExpandSourceListArguments(const _STL::vector<_STL::string,_STL::allocator<_STL::string> > &,_STL::vector<_STL::string,_STL::allocator<_STL::string> > &,unsigned int)
> >Warning W8004 c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx 1417: 'versionValue' is assigned a value that is never used in function cmMakefile::ExpandSourceListArguments(const _STL::vector<_STL::string,_STL::allocator<_STL::string> > &,_STL::vector<_STL::string,_STL::allocator<_STL::string> > &,unsigned int)
> >*** 5 errors in Compile ***
> >
> >I had similar pbs with ITK and BCB6 (see post thread):
> >http://www.itk.org/pipermail/insight-users/2003-May/003642.html
> >
> >Second crack by adding -D_USE_OLD_RW_STL (use old Rogue Wave implementation of STL)
> >to cache value CMAKE_CXX_FLAGS:STRING:
> >
> >Borland C++ 5.6 for Win32 Copyright (c) 1993, 2002 Borland
> >c:\Builder\CMake_1.6.7\Source\cmDumpDocumentation.cxx:
> >Building executable c:\Builder\CMake_1.6.7\Source\DumpDocumentation.exe...
> >Borland C++ 5.6 for Win32 Copyright (c) 1993, 2002 Borland
> >Turbo Incremental Link 5.60 Copyright (c) 1997-2002 Borland
> >Error: Unresolved external 'cmMakefile::ExpandArguments(const _STL::vector >&, ...' referenced from C:\BUILDER\CMAKE_1.6.7\SOURCE\CMAKELIB.LIB|cmake
> >Error: Unresolved external 'GetPredefinedCommands(_STL::list >&)' referenced from C:\BUILDER\CMAKE_1.6.7\SOURCE\CMAKELIB.LIB|cmake
> >Error: Unresolved external 'cmVariableWatch::AddWatch(const _STL::basic_string, _STL::allocator >&, void (*)(const _STL::basic_string, _STL::allocator >&, int, void *), void *)' referenced from C:\BUILDER\CMAKE_1.6.7\SOURCE\CMAKELIB.LIB|cmake
> >Error: Unresolved external 'cmake::DumpDocumentationToFile(std::basic_ostream >&)' referenced from C:\BUILDER\CMAKE_1.6.7\SOURCE\CMDUMPDOCUMENTATION.OBJ
> >
> >I can help out as required...
> >
> >Dean
> >
> >_______________________________________________
> >Cmake mailing list
> >Cmake@public.kitware.com
> >http://public.kitware.com/mailman/listinfo/cmake
>
>
>
>
From bill.hoffman@kitware.com Fri May 23 21:41:02 2003
From: bill.hoffman@kitware.com (Bill Hoffman)
Date: Fri, 23 May 2003 16:41:02 -0400
Subject: [Cmake] BCB6 CMake 1.6.7 build
In-Reply-To: <20030523195141.KRWD12935.tomts19-srv.bellnexxia.net@[209.2
26.175.135]>
Message-ID: <5.2.0.9.0.20030523163323.046b9cc8@pop.biz.rr.com>
At 03:51 PM 5/23/2003, dean.inglis@on.aibn.com wrote:
>Hi Bill,
>
>I deleted the binary tree and the build
>proceeds without error, just a couple of
>W8004 warnings: var assigned a value never used in...
>I will add Hyoid (FYI, hyoid is a laptop and if
>you know what the hyoid bone is you'll get the
>joke :)) as an experimental BCB6 for CMake & VTK.
I would also be interested in fixing it to work without
the -D_USE_OLD_RW_STL flag. It looks like it just needs
a few extra #include lines in the right place.
I have added #include in each file that used
atoi or atof, so that should get around many of the errors
in the first try you did. The only one I do not know what include
to add is the getenv call. Where is that?
Can you try a clean build without the ODL_RW and see if it works now?
-Bill
From John Biddiscombe"
Message-ID: <004801c3216e$e4a6c640$0100a8c0@tigger>
If I remember correctly...atoi, atof etc are std::atoi etc in the BCB 6
includes.
JB
----- Original Message -----
From:
To:
Sent: Friday, May 23, 2003 3:35 PM
Subject: [Cmake] (no subject)
> I'm having some pb's with building MinSizeRel CMake 1.6.7
> using Borland C++ Builder 6 (bcc32). I see that there is
> support for many incarnations of VC and am hoping that
> I can somehow assist with extending Borland support beyond
> Borland free command line tools and BCB 5. First crack
> at building CMake 1.6.7:
>
> Building object file cmMakefile.obj...
> Borland C++ 5.6 for Win32 Copyright (c) 1993, 2002 Borland
> c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx:
> Error E2268 c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx 74: Call to
undefined function 'atoi' in function cmMakefile::GetCacheMajorVersion()
> Error E2268 c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx 83: Call to
undefined function 'atoi' in function cmMakefile::GetCacheMinorVersion()
> Error E2268 c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx 393: Call to
undefined function 'atof' in function cmMakefile::ConfigureFinalPass()
> Warning W8004 c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx 407:
'versionValue' is assigned a value that is never used in function
cmMakefile::ConfigureFinalPass()
> Error E2268 c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx 1060: Call to
undefined function 'getenv' in function
cmMakefile::ExpandVariablesInString(_STL::string &,bool,bool) const
> Error E2268 c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx 1394: Call to
undefined function 'atof' in function
cmMakefile::ExpandSourceListArguments(const
_STL::vector<_STL::string,_STL::allocator<_STL::string> >
&,_STL::vector<_STL::string,_STL::allocator<_STL::string> > &,unsigned int)
> Warning W8004 c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx 1417:
'versionValue' is assigned a value that is never used in function
cmMakefile::ExpandSourceListArguments(const
_STL::vector<_STL::string,_STL::allocator<_STL::string> >
&,_STL::vector<_STL::string,_STL::allocator<_STL::string> > &,unsigned int)
> *** 5 errors in Compile ***
>
> I had similar pbs with ITK and BCB6 (see post thread):
> http://www.itk.org/pipermail/insight-users/2003-May/003642.html
>
> Second crack by adding -D_USE_OLD_RW_STL (use old Rogue Wave
implementation of STL)
> to cache value CMAKE_CXX_FLAGS:STRING:
>
> Borland C++ 5.6 for Win32 Copyright (c) 1993, 2002 Borland
> c:\Builder\CMake_1.6.7\Source\cmDumpDocumentation.cxx:
> Building executable c:\Builder\CMake_1.6.7\Source\DumpDocumentation.exe...
> Borland C++ 5.6 for Win32 Copyright (c) 1993, 2002 Borland
> Turbo Incremental Link 5.60 Copyright (c) 1997-2002 Borland
> Error: Unresolved external 'cmMakefile::ExpandArguments(const
_STL::vector >&,
...' referenced from C:\BUILDER\CMAKE_1.6.7\SOURCE\CMAKELIB.LIB|cmake
> Error: Unresolved external 'GetPredefinedCommands(_STL::list >&)' referenced from
C:\BUILDER\CMAKE_1.6.7\SOURCE\CMAKELIB.LIB|cmake
> Error: Unresolved external 'cmVariableWatch::AddWatch(const
_STL::basic_string, _STL::allocator >&,
void (*)(const _STL::basic_string,
_STL::allocator >&, int, void *), void *)' referenced from
C:\BUILDER\CMAKE_1.6.7\SOURCE\CMAKELIB.LIB|cmake
> Error: Unresolved external
'cmake::DumpDocumentationToFile(std::basic_ostream >&)' referenced from
C:\BUILDER\CMAKE_1.6.7\SOURCE\CMDUMPDOCUMENTATION.OBJ
>
> I can help out as required...
>
> Dean
>
> _______________________________________________
> Cmake mailing list
> Cmake@public.kitware.com
> http://public.kitware.com/mailman/listinfo/cmake
>
From dean.inglis@on.aibn.com Fri May 23 22:20:26 2003
From: dean.inglis@on.aibn.com (dean.inglis@on.aibn.com)
Date: Fri, 23 May 2003 17:20:26 -0400
Subject: [Cmake] BCB6 CMake 1.6.7 build
Message-ID: <20030523212026.PBKU29208.tomts6-srv.bellnexxia.net@[209.226.175.135]>
Trying ITK build clean from scratch,
watch for an experimental with no tests/examples
first. Perhaps JB can offer some assistance as
well????
thanks for your support,
Dean
>
> From: Bill Hoffman
> Date: 2003/05/23 Fri PM 04:41:02 GMT-04:00
> To:
> CC:
> Subject: Re: [Cmake] BCB6 CMake 1.6.7 build
>
> At 03:51 PM 5/23/2003, dean.inglis@on.aibn.com wrote:
> >Hi Bill,
> >
> >I deleted the binary tree and the build
> >proceeds without error, just a couple of
> >W8004 warnings: var assigned a value never used in...
> >I will add Hyoid (FYI, hyoid is a laptop and if
> >you know what the hyoid bone is you'll get the
> >joke :)) as an experimental BCB6 for CMake & VTK.
>
> I would also be interested in fixing it to work without
> the -D_USE_OLD_RW_STL flag. It looks like it just needs
> a few extra #include lines in the right place.
> I have added #include in each file that used
> atoi or atof, so that should get around many of the errors
> in the first try you did. The only one I do not know what include
> to add is the getenv call. Where is that?
>
> Can you try a clean build without the ODL_RW and see if it works now?
>
> -Bill
>
>
>
>
>
From dean.inglis@on.aibn.com Fri May 23 22:36:44 2003
From: dean.inglis@on.aibn.com (dean.inglis@on.aibn.com)
Date: Fri, 23 May 2003 17:36:44 -0400
Subject: [Cmake] BCB6 CMake 1.6.7 build
Message-ID: <20030523213644.PRQS29208.tomts6-srv.bellnexxia.net@[209.226.175.135]>
Bill,
according to BCB6 help, getenv is in stdlib.h
and I am doing BUILD_TESTING...just submitted
to ITK
Dean
> I have added #include in each file that used
> atoi or atof, so that should get around many of the errors
> in the first try you did. The only one I do not know what include
> to add is the getenv call. Where is that?
>
> Can you try a clean build without the ODL_RW and see if it works now?
>
> -Bill
>
>
>
>
>
From bill.hoffman@kitware.com Fri May 23 22:37:14 2003
From: bill.hoffman@kitware.com (Bill Hoffman)
Date: Fri, 23 May 2003 17:37:14 -0400
Subject: [Cmake] (no subject)
In-Reply-To: <004801c3216e$e4a6c640$0100a8c0@tigger>
References: <20030523143557.WNDM20954.tomts5-srv.bellnexxia.net@[209.226.175.135]>
Message-ID: <5.2.0.9.0.20030523173524.046952a8@pop.biz.rr.com>
Yes, if you include them from and not .
Since most compilers do not support the std::cfunction stuff, it
is much more portable to use the .h versions of these files, as
both are standards compliant.
-Bill
At 04:55 PM 5/23/2003, John Biddiscombe wrote:
>If I remember correctly...atoi, atof etc are std::atoi etc in the BCB 6
>includes.
>
>JB
>
>----- Original Message -----
>From:
>To:
>Sent: Friday, May 23, 2003 3:35 PM
>Subject: [Cmake] (no subject)
>
>
>> I'm having some pb's with building MinSizeRel CMake 1.6.7
>> using Borland C++ Builder 6 (bcc32). I see that there is
>> support for many incarnations of VC and am hoping that
>> I can somehow assist with extending Borland support beyond
>> Borland free command line tools and BCB 5. First crack
>> at building CMake 1.6.7:
>>
>> Building object file cmMakefile.obj...
>> Borland C++ 5.6 for Win32 Copyright (c) 1993, 2002 Borland
>> c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx:
>> Error E2268 c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx 74: Call to
>undefined function 'atoi' in function cmMakefile::GetCacheMajorVersion()
>> Error E2268 c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx 83: Call to
>undefined function 'atoi' in function cmMakefile::GetCacheMinorVersion()
>> Error E2268 c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx 393: Call to
>undefined function 'atof' in function cmMakefile::ConfigureFinalPass()
>> Warning W8004 c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx 407:
>'versionValue' is assigned a value that is never used in function
>cmMakefile::ConfigureFinalPass()
>> Error E2268 c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx 1060: Call to
>undefined function 'getenv' in function
>cmMakefile::ExpandVariablesInString(_STL::string &,bool,bool) const
>> Error E2268 c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx 1394: Call to
>undefined function 'atof' in function
>cmMakefile::ExpandSourceListArguments(const
>_STL::vector<_STL::string,_STL::allocator<_STL::string> >
>&,_STL::vector<_STL::string,_STL::allocator<_STL::string> > &,unsigned int)
>> Warning W8004 c:\Builder\CMake_1.6.7\Source\cmMakefile.cxx 1417:
>'versionValue' is assigned a value that is never used in function
>cmMakefile::ExpandSourceListArguments(const
>_STL::vector<_STL::string,_STL::allocator<_STL::string> >
>&,_STL::vector<_STL::string,_STL::allocator<_STL::string> > &,unsigned int)
>> *** 5 errors in Compile ***
>>
>> I had similar pbs with ITK and BCB6 (see post thread):
>> http://www.itk.org/pipermail/insight-users/2003-May/003642.html
>>
>> Second crack by adding -D_USE_OLD_RW_STL (use old Rogue Wave
>implementation of STL)
>> to cache value CMAKE_CXX_FLAGS:STRING:
>>
>> Borland C++ 5.6 for Win32 Copyright (c) 1993, 2002 Borland
>> c:\Builder\CMake_1.6.7\Source\cmDumpDocumentation.cxx:
>> Building executable c:\Builder\CMake_1.6.7\Source\DumpDocumentation.exe...
>> Borland C++ 5.6 for Win32 Copyright (c) 1993, 2002 Borland
>> Turbo Incremental Link 5.60 Copyright (c) 1997-2002 Borland
>> Error: Unresolved external 'cmMakefile::ExpandArguments(const
>_STL::vector >&,
>...' referenced from C:\BUILDER\CMAKE_1.6.7\SOURCE\CMAKELIB.LIB|cmake
>> Error: Unresolved external 'GetPredefinedCommands(_STL::list_STL::allocator >&)' referenced from
>C:\BUILDER\CMAKE_1.6.7\SOURCE\CMAKELIB.LIB|cmake
>> Error: Unresolved external 'cmVariableWatch::AddWatch(const
>_STL::basic_string, _STL::allocator >&,
>void (*)(const _STL::basic_string,
>_STL::allocator >&, int, void *), void *)' referenced from
>C:\BUILDER\CMAKE_1.6.7\SOURCE\CMAKELIB.LIB|cmake
>> Error: Unresolved external
>'cmake::DumpDocumentationToFile(std::basic_ostreamstd::char_traits >&)' referenced from
>C:\BUILDER\CMAKE_1.6.7\SOURCE\CMDUMPDOCUMENTATION.OBJ
>>
>> I can help out as required...
>>
>> Dean
>>
>> _______________________________________________
>> Cmake mailing list
>> Cmake@public.kitware.com
>> http://public.kitware.com/mailman/listinfo/cmake
>>
>
>_______________________________________________
>Cmake mailing list
>Cmake@public.kitware.com
>http://public.kitware.com/mailman/listinfo/cmake
From bill.hoffman@kitware.com Fri May 23 22:38:24 2003
From: bill.hoffman@kitware.com (Bill Hoffman)
Date: Fri, 23 May 2003 17:38:24 -0400
Subject: [Cmake] BCB6 CMake 1.6.7 build
In-Reply-To: <20030523212026.PBKU29208.tomts6-srv.bellnexxia.net@[209.22
6.175.135]>
Message-ID: <5.2.0.9.0.20030523173802.04698c80@pop.biz.rr.com>
Can you also try CMake without the old stl?
At 05:20 PM 5/23/2003, dean.inglis@on.aibn.com wrote:
>Trying ITK build clean from scratch,
>watch for an experimental with no tests/examples
>first. Perhaps JB can offer some assistance as
>well????
>
>thanks for your support,
>Dean
>
>
>>
>> From: Bill Hoffman
>> Date: 2003/05/23 Fri PM 04:41:02 GMT-04:00
>> To:
>> CC:
>> Subject: Re: [Cmake] BCB6 CMake 1.6.7 build
>>
>> At 03:51 PM 5/23/2003, dean.inglis@on.aibn.com wrote:
>> >Hi Bill,
>> >
>> >I deleted the binary tree and the build
>> >proceeds without error, just a couple of
>> >W8004 warnings: var assigned a value never used in...
>> >I will add Hyoid (FYI, hyoid is a laptop and if
>> >you know what the hyoid bone is you'll get the
>> >joke :)) as an experimental BCB6 for CMake & VTK.
>>
>> I would also be interested in fixing it to work without
>> the -D_USE_OLD_RW_STL flag. It looks like it just needs
>> a few extra #include lines in the right place.
>> I have added #include in each file that used
>> atoi or atof, so that should get around many of the errors
>> in the first try you did. The only one I do not know what include
>> to add is the getenv call. Where is that?
>>
>> Can you try a clean build without the ODL_RW and see if it works now?
>>
>> -Bill
>>
>>
>>
>>
>>
>
>_______________________________________________
>Cmake mailing list
>Cmake@public.kitware.com
>http://public.kitware.com/mailman/listinfo/cmake
From Peter.Vanroose@esat.kuleuven.ac.be Sat May 24 12:59:38 2003
From: Peter.Vanroose@esat.kuleuven.ac.be (Peter Vanroose)
Date: Sat, 24 May 2003 13:59:38 +0200 (CEST)
Subject: [Cmake] Re: [Dart] Configuration command, Symlinks in paths
In-Reply-To: <5.2.0.9.0.20030521084209.04455280@pop.biz.rr.com>
References: <200305211024.08178.Stefan.Schmidt@sophia.inria.fr>
<5.2.0.9.0.20030520091332.042ff928@pop.biz.rr.com>
<200305211024.08178.Stefan.Schmidt@sophia.inria.fr>
<5.2.0.9.0.20030521084209.04455280@pop.biz.rr.com>
Message-ID:
> We fix the /tmp_mnt problem explicitly in cmake for the sgi, and have
> done so since cmake 1.2.
It was not on SGI this time
From bill.hoffman@kitware.com Sat May 24 15:15:43 2003
From: bill.hoffman@kitware.com (William A. Hoffman)
Date: Sat, 24 May 2003 10:15:43 -0400
Subject: [Cmake] Re: [Dart] Configuration command, Symlinks in paths
In-Reply-To:
References: <5.2.0.9.0.20030521084209.04455280@pop.biz.rr.com>
<200305211024.08178.Stefan.Schmidt@sophia.inria.fr>
<5.2.0.9.0.20030520091332.042ff928@pop.biz.rr.com>
<200305211024.08178.Stefan.Schmidt@sophia.inria.fr>
<5.2.0.9.0.20030521084209.04455280@pop.biz.rr.com>
Message-ID: <5.1.0.14.0.20030524101311.04783ec8@pop.biz.rr.com>
We remove /tmp_mnt on all machines.
-Bill
At 01:59 PM 5/24/2003 +0200, Peter Vanroose wrote:
> > We fix the /tmp_mnt problem explicitly in cmake for the sgi, and have
> > done so since cmake 1.2.
>
>It was not on SGI this time
>_______________________________________________
>Cmake mailing list
>Cmake@public.kitware.com
>http://public.kitware.com/mailman/listinfo/cmake
From dias@biac.duke.edu Mon May 26 16:51:04 2003
From: dias@biac.duke.edu (Jimmy Dias)
Date: 26 May 2003 11:51:04 -0400
Subject: [Cmake] CMake install issues (AIX 5.2L)
Message-ID: <1053964265.15462.216.camel@dirac.biac.duke.edu>
Hey all,
I checked out the latest stable cmake release and I get the following
errors when trying to configure the Makefile. I'm using
g++-2.9.aix51.020209-3 and running AIX 5.2L on a IBM p670. I also tried
the CVS version and while I can get past the configure step the make
bombs with the error pasted below.
I suppose I could use certain compiler flags but I have no idea which
ones to use or whether or not that would fix anything. Any suggestions?
Thanks in advance,
Jimmy
--CUT--
creating Source/cmConfigure.h
Source/cmConfigure.h is unchanged
cd Source; make
g++ -DCMAKE_ROOT_DIR='"/home/dias/install/source/cmake-1.6.7"'
-DCMAKE_HAS_AUTOCONF
-I/home/dias/install/source/cmake-1.6.7/Bootstrap/Source -c
/home/dias/install/source/cmake-1.6.7/Source/cmSystemTools.cxx -o
cmSystemTools.o
/home/dias/install/source/cmake-1.6.7/Source/cmSystemTools.cxx: In
function `double cmSystemTools::GetTime ()':
/home/dias/install/source/cmake-1.6.7/Source/cmSystemTools.cxx:124:
implicit declaration of function `int time (...)'
...
Problem bootstrapping CMake
--CUT--
CVS VERSION
Building object file cmCommands.o...
In file included from
/home/dias/CMake/Source/cmLoadCommandCommand.cxx:19,
from /home/dias/CMake/Source/cmCommands.cxx:107:
/home/dias/CMake/Source/cmCPluginAPI.cxx:549: cannot convert `void *(*)
(...)' to `void *(*) ()' in initialization
make: 1254-004 The error code from the last command is 1.
From bill.hoffman@kitware.com Mon May 26 22:40:20 2003
From: bill.hoffman@kitware.com (William A. Hoffman)
Date: Mon, 26 May 2003 17:40:20 -0400
Subject: [Cmake] CMake install issues (AIX 5.2L)
In-Reply-To: <1053964265.15462.216.camel@dirac.biac.duke.edu>
Message-ID: <5.1.0.14.0.20030526173903.04751a60@pop.biz.rr.com>
Can you send the errors you get during the configure step with the release?
Also, you may be able to use the AIX binaries we provide. We have only
tested with the native AIX c/C++ compiler and not gnu on AIX.
-Bill
At 11:51 AM 5/26/2003 -0400, Jimmy Dias wrote:
>Hey all,
>
>I checked out the latest stable cmake release and I get the following
>errors when trying to configure the Makefile. I'm using
>g++-2.9.aix51.020209-3 and running AIX 5.2L on a IBM p670. I also tried
>the CVS version and while I can get past the configure step the make
>bombs with the error pasted below.
>
>I suppose I could use certain compiler flags but I have no idea which
>ones to use or whether or not that would fix anything. Any suggestions?
>
>Thanks in advance,
>Jimmy
>
>--CUT--
>creating Source/cmConfigure.h
>Source/cmConfigure.h is unchanged
> cd Source; make
> g++ -DCMAKE_ROOT_DIR='"/home/dias/install/source/cmake-1.6.7"'
>-DCMAKE_HAS_AUTOCONF
>-I/home/dias/install/source/cmake-1.6.7/Bootstrap/Source -c
>/home/dias/install/source/cmake-1.6.7/Source/cmSystemTools.cxx -o
>cmSystemTools.o
>/home/dias/install/source/cmake-1.6.7/Source/cmSystemTools.cxx: In
>function `double cmSystemTools::GetTime ()':
>/home/dias/install/source/cmake-1.6.7/Source/cmSystemTools.cxx:124:
>implicit declaration of function `int time (...)'
>...
>Problem bootstrapping CMake
>--CUT--
>
>CVS VERSION
>
>Building object file cmCommands.o...
>In file included from
>/home/dias/CMake/Source/cmLoadCommandCommand.cxx:19,
> from /home/dias/CMake/Source/cmCommands.cxx:107:
>/home/dias/CMake/Source/cmCPluginAPI.cxx:549: cannot convert `void *(*)
>(...)' to `void *(*) ()' in initialization
>make: 1254-004 The error code from the last command is 1.
>
>_______________________________________________
>Cmake mailing list
>Cmake@public.kitware.com
>http://public.kitware.com/mailman/listinfo/cmake
From john.biddiscombe@mirada-solutions.com Tue May 27 11:45:07 2003
From: john.biddiscombe@mirada-solutions.com (John Biddiscombe)
Date: Tue, 27 May 2003 11:45:07 +0100
Subject: [Cmake] BCB6 CMake 1.6.7 build
Message-ID: <4926A5BE4AFE7C4A83D5CF5CDA7B77540712AE@oxcore01.mirada-solutions.com>
>Perhaps JB can offer some assistance as
> well????
Ummm...
well, if you go back several months on the itk list, you'll see I made =
an effort at making it compile under BCB6. I eventually abandoned the =
attempt becuase I decided I was making so many changes, that it'd become =
painful to maintain with cvs conflicts appearing regularly (and it'd be =
easier to just us V5 or MSvc). Perhaps things've settled down at bit =
now.
anyway...I found a folder on my home machine with the changes I made to =
get things (partially) compiling. I didn't compile all of itk, but quite =
a bit. When I get a moment, I'll see if I can make sense of what's in =
there. It might not be what I think it is, but the folder was titled =
"ITK changes" and should be helpful. Only problem is that you'd need to =
check out old versions of itk, apply my changes, then do an update and =
merge.=20
JB
From paul.smyth@vicon.com Tue May 27 11:47:30 2003
From: paul.smyth@vicon.com (Paul Smyth)
Date: Tue, 27 May 2003 11:47:30 +0100
Subject: [Cmake] Generating gnu Makefiles compiling under MSVC
Message-ID: <609562B7721BD41187B400B0D020856001CF636B@FILESERV>
I haven't managed to find anything under Google, so here goes:
Gnu make supports parallel make, while nmake doesn't, so gnu make might be a
preferable choice for batch building on a multiprocessor box. I was
wondering how much work it would be for me (or someone else) to make a
configuration for CMakeSetup under windows that generated gnu Makefiles
which invoked MSVC, given that gnu makefiles can be generated under
Cygwin/Windows, albeit compiling with gcc.
Is this a bonkers thing to want? Has it already been done and I haven't
noticed?
Thanks, Paul Smyth.
This e-mail, and any attachment, is confidential. If you have received it in
error, please delete it from your system, do not use or disclose the
information in any way, and notify me immediately.
From bill.hoffman@kitware.com Tue May 27 14:11:45 2003
From: bill.hoffman@kitware.com (Bill Hoffman)
Date: Tue, 27 May 2003 09:11:45 -0400
Subject: [Cmake] Generating gnu Makefiles compiling under MSVC
In-Reply-To: <609562B7721BD41187B400B0D020856001CF636B@FILESERV>
Message-ID: <5.2.0.9.0.20030527090659.01369410@pop.biz.rr.com>
This has not been done. There are some other issues you may wan to know
about. Since Microsoft uses a single program database for a group of object
files in a directory parallel builds will not work with the microsoft compiler in
debug and possibly other modes.
It should not be that hard to try. You might be able to try it by running
the cygwin version of cmake, and creating a CYGWIN-cl.cmake file in the Modules/Platform
directory of cmake. If you get one that works, let me know and we can
put it back into cmake, and work on generating that type of makefile from the
windows GUI.
-Bill
At 06:47 AM 5/27/2003, Paul Smyth wrote:
>I haven't managed to find anything under Google, so here goes:
>
>Gnu make supports parallel make, while nmake doesn't, so gnu make might be a
>preferable choice for batch building on a multiprocessor box. I was
>wondering how much work it would be for me (or someone else) to make a
>configuration for CMakeSetup under windows that generated gnu Makefiles
>which invoked MSVC, given that gnu makefiles can be generated under
>Cygwin/Windows, albeit compiling with gcc.
>
>Is this a bonkers thing to want? Has it already been done and I haven't
>noticed?
>
>Thanks, Paul Smyth.
>
>This e-mail, and any attachment, is confidential. If you have received it in
>error, please delete it from your system, do not use or disclose the
>information in any way, and notify me immediately.
>_______________________________________________
>Cmake mailing list
>Cmake@public.kitware.com
>http://public.kitware.com/mailman/listinfo/cmake
From james.dias@duke.edu Sun May 25 21:54:48 2003
From: james.dias@duke.edu (Jimmy Dias)
Date: Sun, 25 May 2003 16:54:48 -0400
Subject: [Cmake] AIX 5.2 issues
Message-ID:
Hey All,
Has anyone been able to successfully build CMake on an AIX 5.2 machine? I
installed g++-2.9.aix51.020209-3 and gcc-2.9.aix51.020209-3 instead of using
AIX's native compiler (mostly because I don't know how or even if there IS a
C++ compiler on the machine) but ./configure crashes with the following
error:
g++ -DCMAKE_ROOT_DIR='"/home/dias/install/source/cmake-1.6.7"' -DCMAKE_HAS_
AUTOCONF -I/home/dias/install/source/cmake-1.6.7/Bootstrap/Source -c
/home/dias/install/source/cmake-1.6.7/Source/cmSystemTools.cxx -o
cmSystemTools.o
/home/dias/install/source/cmake-1.6.7/Source/cmSystemTools.cxx: In function
`double cmSystemTools::GetTime ()':
/home/dias/install/source/cmake-1.6.7/Source/cmSystemTools.cxx:124: implicit
declaration of function `int time (...)'
make: The error code from the last command is 1.
Stop.
make: The error code from the last command is 2.
Stop.
Problem bootstrapping CMake
I suppose setting compiler flags might work but I have no idea what flags to
use. Any suggestions?
Thanks in advance,
Jimmy
From andy.cedilnik@kitware.com Tue May 27 20:43:46 2003
From: andy.cedilnik@kitware.com (Andy Cedilnik)
Date: 27 May 2003 15:43:46 -0400
Subject: [Cmake] AIX 5.2 issues
In-Reply-To:
References:
Message-ID: <1054064626.15359.10.camel@andoria>
Hi Jimmy,
We build CMake on AIX every night, however we use xlc/xlC, not gcc.
Could you please try simple example:
#include
#include
#include
int main()
{
time_t secs;
time(&secs);
std::cout << "Time: " << time << std::endl;
return 0;
}
Andy
On Sun, 2003-05-25 at 16:54, Jimmy Dias wrote:
> Hey All,
>
> Has anyone been able to successfully build CMake on an AIX 5.2 machine? I
> installed g++-2.9.aix51.020209-3 and gcc-2.9.aix51.020209-3 instead of using
> AIX's native compiler (mostly because I don't know how or even if there IS a
> C++ compiler on the machine) but ./configure crashes with the following
> error:
>
>
>
>
>
> g++ -DCMAKE_ROOT_DIR='"/home/dias/install/source/cmake-1.6.7"' -DCMAKE_HAS_
> AUTOCONF -I/home/dias/install/source/cmake-1.6.7/Bootstrap/Source -c
> /home/dias/install/source/cmake-1.6.7/Source/cmSystemTools.cxx -o
> cmSystemTools.o
> /home/dias/install/source/cmake-1.6.7/Source/cmSystemTools.cxx: In function
> `double cmSystemTools::GetTime ()':
> /home/dias/install/source/cmake-1.6.7/Source/cmSystemTools.cxx:124: implicit
> declaration of function `int time (...)'
> make: The error code from the last command is 1.
>
>
> Stop.
> make: The error code from the last command is 2.
>
>
> Stop.
> Problem bootstrapping CMake
>
> I suppose setting compiler flags might work but I have no idea what flags to
> use. Any suggestions?
From rc3@doc.ic.ac.uk Wed May 28 12:49:28 2003
From: rc3@doc.ic.ac.uk (Raghavendra Chandrashekara)
Date: Wed, 28 May 2003 12:49:28 +0100
Subject: [Cmake] packaging source
Message-ID: <3ED4A248.5000900@doc.ic.ac.uk>
Hi Folks,
Is there a way of packaging all the code in a cmake project. Something
like the "make dist" command which is available with gnu
autoconf/automake tools.
Thanks,
Raghavendra.
From brad.king@kitware.com Thu May 29 15:01:55 2003
From: brad.king@kitware.com (Brad King)
Date: Thu, 29 May 2003 10:01:55 -0400 (EDT)
Subject: [Cmake] packaging source
In-Reply-To: <3ED4A248.5000900@doc.ic.ac.uk>
Message-ID:
> Is there a way of packaging all the code in a cmake project. Something
> like the "make dist" command which is available with gnu
> autoconf/automake tools.
What exactly does "make dist" do? Tar up the source tree? CMake does not
support this directly. You can try something like this:
IF(UNIX)
ADD_CUSTOM_TARGET(dist ${PROJECT_SOURCE_DIR}/package.sh)
ENDIF(UNIX)
This will add "dist" as a makefile target that runs package.sh, which can
do whatever packaging you want.
-Brad
From andy.cedilnik@kitware.com Thu May 29 15:22:24 2003
From: andy.cedilnik@kitware.com (Andy Cedilnik)
Date: 29 May 2003 10:22:24 -0400
Subject: [Cmake] packaging source
In-Reply-To:
References:
Message-ID: <1054218144.15363.27.camel@andoria>
Hi,
How about this:
STRING(REGEX REPLACE ".*\\/" DIR_NAME "${PROJECT_SOURCE_DIR}")
SET(TOPLEVEL "${PROJECT_SOURCE_DIR}/..)
ADD_CUSTOM_TARGET(
dist
${CMAKE_COMMAND} -E chdir ${TOPLEVEL}
${TAR_COMMAND} cvf ${PROJECT_NAME}.tar ${DIR_NAME})
Andy
On Thu, 2003-05-29 at 10:01, Brad King wrote:
> > Is there a way of packaging all the code in a cmake project. Something
> > like the "make dist" command which is available with gnu
> > autoconf/automake tools.
>
> What exactly does "make dist" do? Tar up the source tree? CMake does not
> support this directly. You can try something like this:
>
> IF(UNIX)
> ADD_CUSTOM_TARGET(dist ${PROJECT_SOURCE_DIR}/package.sh)
> ENDIF(UNIX)
>
> This will add "dist" as a makefile target that runs package.sh, which can
> do whatever packaging you want.
From rc3@doc.ic.ac.uk Fri May 30 10:04:24 2003
From: rc3@doc.ic.ac.uk (Raghavendra Chandrashekara)
Date: Fri, 30 May 2003 10:04:24 +0100
Subject: [Cmake] packaging source
References: <1054218144.15363.27.camel@andoria>
Message-ID: <3ED71E98.2040907@doc.ic.ac.uk>
Hi,
Thanks for your help. Are there any plans to add such a feature to CMake
in the future?
Raj
Andy Cedilnik wrote:
> Hi,
>
> How about this:
>
> STRING(REGEX REPLACE ".*\\/" DIR_NAME "${PROJECT_SOURCE_DIR}")
> SET(TOPLEVEL "${PROJECT_SOURCE_DIR}/..)
> ADD_CUSTOM_TARGET(
> dist
> ${CMAKE_COMMAND} -E chdir ${TOPLEVEL}
> ${TAR_COMMAND} cvf ${PROJECT_NAME}.tar ${DIR_NAME})
>
> Andy
>
> On Thu, 2003-05-29 at 10:01, Brad King wrote:
>
>>>Is there a way of packaging all the code in a cmake project. Something
>>>like the "make dist" command which is available with gnu
>>>autoconf/automake tools.
>>
>>What exactly does "make dist" do? Tar up the source tree? CMake does not
>>support this directly. You can try something like this:
>>
>>IF(UNIX)
>> ADD_CUSTOM_TARGET(dist ${PROJECT_SOURCE_DIR}/package.sh)
>>ENDIF(UNIX)
>>
>>This will add "dist" as a makefile target that runs package.sh, which can
>>do whatever packaging you want.
>
>
>
> _______________________________________________
> Cmake mailing list
> Cmake@public.kitware.com
> http://public.kitware.com/mailman/listinfo/cmake
From andy.cedilnik@kitware.com Fri May 30 12:12:33 2003
From: andy.cedilnik@kitware.com (Andy Cedilnik)
Date: 30 May 2003 07:12:33 -0400
Subject: [Cmake] packaging source
In-Reply-To: <3ED71E98.2040907@doc.ic.ac.uk>
References:
<1054218144.15363.27.camel@andoria> <3ED71E98.2040907@doc.ic.ac.uk>
Message-ID: <1054293153.15365.48.camel@andoria>
Hi Raj,
Yes, but there are a lot of features and improvements that are on the
list. If you want to do it and submit the patch, you are more than
welcome.
Andy
On Fri, 2003-05-30 at 05:04, Raghavendra Chandrashekara wrote:
> Hi,
>
> Thanks for your help. Are there any plans to add such a feature to CMake
> in the future?
From vanessa.noguesruiz@epfl.ch Fri May 30 16:13:03 2003
From: vanessa.noguesruiz@epfl.ch (vanessa.noguesruiz@epfl.ch)
Date: Fri, 30 May 2003 17:13:03 +0200 (MEST)
Subject: [Cmake] (no subject)
Message-ID: <1054307583.3ed774ff9e947@imapwww.epfl.ch>
Hi everybody.
Does anybody know how may I include a .dll file in a CMakeLists.txt file? I use
the formula:
LINK_DIRECTORIES (
${VTKDIR}/bin
${MIPDIR}/win32/bin
)
TARGET_LINK_LIBRARIES(
rigid_nonrigid
vtkCommon
vtkImaging
vtkIO
vtkFiltering
vtkGraphics
vtkRendering
mipdll.dll
)
mipdll.dll is located in ${MIPDIR}/win32/bin
Thanks a lot.
From bill.hoffman@kitware.com Fri May 30 18:26:53 2003
From: bill.hoffman@kitware.com (Bill Hoffman)
Date: Fri, 30 May 2003 13:26:53 -0400
Subject: [Cmake] (no subject)
In-Reply-To: <1054307583.3ed774ff9e947@imapwww.epfl.ch>
Message-ID: <5.2.0.9.0.20030530132613.046dcc00@pop.biz.rr.com>
Leave off the .dll, and make sure there is a mipdll.lib in ${MIPDIR}/win32/bin because
you can not link to a .dll, but you need the .lib file for the .dll.
-Bill
At 11:13 AM 5/30/2003, vanessa.noguesruiz@epfl.ch wrote:
>Hi everybody.
>Does anybody know how may I include a .dll file in a CMakeLists.txt file? I use
>the formula:
>
>
>LINK_DIRECTORIES (
> ${VTKDIR}/bin
> ${MIPDIR}/win32/bin
>)
>
>TARGET_LINK_LIBRARIES(
> rigid_nonrigid
> vtkCommon
> vtkImaging
> vtkIO
> vtkFiltering
> vtkGraphics
> vtkRendering
> mipdll.dll
>)
>
>mipdll.dll is located in ${MIPDIR}/win32/bin
>
>Thanks a lot.
>
>_______________________________________________
>Cmake mailing list
>Cmake@public.kitware.com
>http://public.kitware.com/mailman/listinfo/cmake