kdesupport/kdewin32/include/msvc

[Christian Ehrlicher]

> Von: "Jarosław Staniek"
> Christian Ehrlicher said the following, On 2008-01-30 12:15:
> >> Von: "Jarosław Staniek"
> >> Christian Ehrlicher said the following, On 2008-01-30 11:27:
> >>>> Von: Jaroslaw Staniek
> >>>> SVN commit 768534 by staniek:
> >>>>
> >>>> add a number of HAVE_* flags so we can easier check existence of
> these
> >>>> functions e.g. in strigi
> >>>>
> >>> That's not good. You'll get conflicts now when anywhere else HAVE_FOO
> is
> >> defined to 1 or something else. Please use a kdewin prefix or something
> >> like this. I also don't understand why you need this at all - cmake
> configure
> >> checks are the correct place to check for such things.
> >>
> >> Configure checks no longer work because functions like strcasecmp() are
> >> defined in kdewin32 and the checks does not link against kdewin32. This
> is
> >> the 
> >> case for instance in strigi. Some other checks do not define 
> >> WIN32_LEAN_AND_MEAN and we end up without important HAVE_* macros
> defined
> >> too.
> >>
> >> Perhaps we should add #ifndef HAVE_FOO ?
> >>
> > Your system is screwed up. strigi was never supposed to use headers from
> kdewin32 nor link against kdewin32.
> > It compiled fine for the last months and I never heard complains from
> emerge users that strigi is uncompileable. What changed?
> 
> I have used emerge with strigi from trunk. Ask Partick how many compile 
> problems we've encountered there last night...
> 
Then don't use strigi/trunk for now. 0.5.7 works fine (+ is the default in emerge afaics) and the rest will be fixed somehow - but not by using a new dependency (kdewin32). I can't think of any big change in strigi which would required this step.
Also the HAVE_foo in public headers have to be removed - they're wrong. You'll only get trouble with them upstream.


Christian
-- 
GMX FreeMail: 1 GB Postfach, 5 E-Mail-Adressen, 10 Free SMS.
Alle Infos und kostenlose Anmeldung: http://www.gmx.net/de/go/freemail