[CMake] Re: [Chicken-users] Re: Cygwin via CMake: it works!

Brandon J. Van Every bvanevery at gmail.com
Tue Aug 1 18:27:49 EDT 2006


John Cowan wrote:
> Brandon J. Van Every scripsit:
>
>   
>> Why does libchicken-static.a require renaming?  What evidence do you 
>> have for this?
>>     
>
> Because if you do "csc -static" or "csc -static-libs", and you haven't renamed,
> you get an error.  More fundamentally, Cygwin ld expects -lfoo in a static
> build to look in /usr/lib for libfoo.a, not libfoo-static.a.
>   

Ok, I've accepted that MSVC is the odd man out.  I'm using postfixes 
like libchickens.lib and libchickend.lib on MSVC, as that's the way 
people typically do things.  I don't know if my postfixes obey good 
conventions; that information is difficult to Google up.  Anyways, 
Unix-y compilers are now expected to fend for themselves as far as 
keeping their static vs. dynamic nomenclatures straight.  So it seems 
they do the right thing, i.e. libchicken.a.

>   
>> Well, I suppose the argument is whether further testing shows that it 
>> breaks something.  But that would require further testing, and your 
>> initial method does get us farther along.  I will CMakeify it.
>>     
>
> Thanks.  As far as I'm concerned, it gets us over the hump between
> "unusable" and "usable".
>   

I modified your solution.  I install cygchicken-0.dll and 
libchicken-0.dll.a, then I make a symlink for libchicken.dll.a.  I can 
do "csi -R srfi-1" no problem now.  It's in Darcs; please test.


Cheers,
Brandon Van Every




-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://public.kitware.com/pipermail/cmake/attachments/20060801/13d5ba78/attachment.htm


More information about the CMake mailing list