[Paraview] paraview 3.0.1 build on MacOS X is brittle

[Marc Baaden]

Hi,
I'd like to second Jack's appeal for a fink-based (qt4-x11 dependent) 
paraview build and hopefully provide a valid reason for doing so.

It is great that Kitware now provides an OSX native version, but as 
soon as you need/want to recompile the software you'll have a very 
steep installation curve and some sleepless nights.

In particular I wanted to rebuild a parallel (mpi) version of 
paraview, and could so far not manage to get a proper build with the 
v3.x versions.
Building with fink - given the automatic dependency handling - is a 
pleasure though, and I now use paraview 2.6 as it is in fink and I 
could recompile a functional mpi version (thanks to Martin Costabel).

This is all probably due to my incompetence in Mac software development/
installation issues (although I am not really a newby), but the point 
is if spending a couple of days going through mailing lists and trying 
to build such a big package as paraview don't succeed, many users will 
probably be put off to try and build paraview from source to tune it to 
their needs and eventually also contribute back some improvements.

Fink provides all that.. if paraview 3 is in fink, every fink user will 
be able to recompile it, no issue here. I think this could be seen as an
important benefit.

(of course in an ideal world, the fink-paraview 3 version would than 
eventually have an option to swap-out qt4-x11 against osx-native qt4 
like the choice between Apple's X11 or Fink X11.. but that's an 
entirely different issue, I think)

Just my 2 cents,
Cheers,
Marc Baaden

>>> "Mike Jackson" said:
 >> I still need a _good_ reason for building against X11. Just because it
 >> _can_ be done does not mean it _should_ be done. OS X users _expect_ a
 >> proper OS X application bundle to use. We (OS X Users) have been
 >> screaming at the ParaView folks for years to do this so we did not
 >> have to rely on X11 to run ParaView. Now that we have it you want to
 >> take us back to X11? I don't think so. I would rather see Kitware's
 >> resources put to other tasks than to back port to X11.
 >> 
 >> Now.. having ranted about all that I am going to jump to the other
 >> side. Last week Apple announced that the only 64Bit GUIs going forward
 >> will be Cocoa and X11. This leaves those of us wanting a 64 bit
 >> version of ParaVIew on OS X in a bind. Qt bases there implementation
 >> on Carbon, therefor no 64Bit ParaView for OS X. So the only way I am
 >> seeing to get 64 Bit paraview on OS X would be if we continue to build
 >> against X11, which just completely SUCKS. And I am going to jump out
 >> on a limb and hope that the Qt4-X11 package is NOT based on Carbon.
 >> Further more for me to even consider this is if the Qt4X11 package
 >> becomes STABLE, not UNSTABLE. I have clients to serve and putting
 >> buggy software on their machines and saying "deal with it" is not a
 >> good reflection on me or Kitware.
 >> 
 >> Cheers
 >> Mike Jackson
 >> 
 >> On 6/19/07, Jack Howarth <howarth at bromo.msbb.uc.edu> wrote:
 >> > Berk,
 >> >   I was able to build paraview 3.0.1 against the qt4-x11
 >> > package in fink 10.4 unstable. However the installation
 >> > portion of the build goes astray because it expects to
 >> > create an application. What would be helpful is if there
 >> > existed an option for MacOS X builds to build against
 >> > non-carbon qt4 (that uses X11 instead) and to not
 >> > force the user to always create an application
 >> > (but install the programs ala linux).
 >> >                 Jack
 >> >
 >> > On Tue, Jun 19, 2007 at 10:52:44AM -0400, Berk Geveci wrote:
 >> > >
 >> > > It is actually possible to only build the servers and the python
 >> > > client. However, these components also use Carbon instead of X11 to do
 >> > > their rendering. There is actually no X11 dependency left in paraview
 >> > > whatsoever. Although, it is possible to configure the server build to
 >> > > use X11 instead of Carbon, I don't see any practical reason of doing
 >> > > it. Am I missing something?
 >> > >
 >> > > -berk
 >> _______________________________________________
 >> ParaView mailing list
 >> ParaView at paraview.org
 >> http://www.paraview.org/mailman/listinfo/paraview
 >> 

Marc Baaden

-- 
 Dr. Marc Baaden  - Institut de Biologie Physico-Chimique, Paris
 mailto:baaden at smplinux.de      -      http://www.baaden.ibpc.fr
 FAX: +33 15841 5026  -  Tel: +33 15841 5176  ou  +33 609 843217