From kde-commits  Sat Nov 19 15:23:50 2005
From: Matt Rogers <mattr () kde ! org>
Date: Sat, 19 Nov 2005 15:23:50 +0000
To: kde-commits
Subject: Re: KDE/kdelibs/bksys
Message-Id: <200511190923.50790.mattr () kde ! org>
X-MARC-Message: https://marc.info/?l=kde-commits&m=113241372804568

On Saturday 19 November 2005 04:25, Ralf Habacker wrote:
> Matt Rogers schrieb:
> > SVN commit 481408 by mattr:
> >
> > Make the zlib check pass on unix
> >
> >
> >
> > M +25 -8 libz.py
> >
> >
> > --- trunk/KDE/kdelibs/bksys/libz.py #481407:481408 @@ -21,22 +21,19
> > @@ env['CACHED_Z'] = 1 ret = 1
> >
> > + context.Result(ret) return ret
> >
> > from SCons.Options import Options import os - + optionFile =
> > env['CACHEDIR'] + 'libz.cache.py' opts = Options(optionFile)
> > opts.AddOptions( ('CACHED_Z', 'Whether libz is available'),
> > ('CXXFLAGS_Z',''), - ('CCFLAGS_Z',''), - ('LINKFLAGS_Z',''), -
> > ('CPPPATH_Z',''), - ('LIB_Z',''), - ('LIBPATH_Z',''),
>
> Are you aware, that removing this options break the windows
> implementation ?
>

No I wasn't, I'll add them back.

> > + ('LINKFLAGS_Z','') ) opts.Update(env) opts.Save(optionFile, env)
> > @@ -44,8 +41,28 @@ if not env['HELP'] and (env['_CONFIGURE_'] or
> > not env.has_key('CACHED_Z')): conf = env.Configure(custom_tests =
> > { 'Check_libz' : Check_libz} )
> >
> > - if not conf.Check_libz(): + if env['WINDOWS'] and not
> > conf.Check_libz(): print 'libz not found (mandatory).' + else: +
> > env['CACHED_Z'] = 0 + if conf.CheckHeader('zlib.h') and
> > conf.CheckLib('z'): + env['CXXFLAGS_Z'] = ['-DHAVE_LIBZ']
>
> Here HAVE_LIBZ is set as define on compiling command lines.

which shouldn't be used IMHO. The config.h files should handle everything on 
both platforms.

> > + env['LINKFLAGS_Z'] = ['-lz'] + env['CACHED_Z'] = 1
> >
> > - env = conf.Finish() + dest=open(env.join(env['_BUILDDIR_'],
> > 'config-z.h'), 'w') + dest.write('/* libz configuration created
> > by bksys */\n') + + if env['CACHED_Z']: + dest.write('#define
> > HAVE_LIBZ 1\n');
>
> Here HAVE_LIBZ is added in a config-z.h. Isn't this definition already
> defined in [1] ?
> I think you should use only one way
> Regards
>  Ralf

I'll go the config-z.h route then.

Thanks for the comments!

-- 
Matt