[gcompris-devel] Re: gcompris not building

[Bruno Coudoin]

I would be happy to fix that but I don't exactly understand what I have
to do. Do you mean commiting a new aclocal.m4 ?

For the 1.0.2, we already have enough changes to justify a 1.0.2 (3 new
boards).


le mar 26-03-2002 à 23:27, Yann Dirson a écrit :
> Hi,
> 
> aclocal.m4 is indeed a generated file.  The snippet causing problem
> comes from /usr/share/aclocal/libxml.m4, which happen to include
> <string.h> already on my box.
> 
> I'm not sure it really makes sense to remove the cast - as the value
> is affected to a char*, it may be necessary on strange/history
> platforms where it may return something else (an int ?).
> 
> My bet is:
> 
> - upstream should regenerate aclocal.m4 and configure using more
>   recent support files
> 
> - I can build a new package for 1.0.1 with those regenerated, but this
>   involves a hack (touching all of them from debian/rules) to avoid
>   build-dep on autotools (which is undesirable because of slow
>   autobuilders like m68k and maybe arm)
> 
> Bruno: if you have quick plans for a 1.0.2 or so with this fixed I'll
> wait for it, otherwise I'll go the hackish way - just tell me.
> 
> 
> On Tue, Mar 26, 2002 at 04:22:02PM +0000, Matthew Wilcox wrote:
> > On Tue, Mar 26, 2002 at 12:06:18PM +0100, Yann Dirson wrote:
> > > The problems appears to be that the libxml2 includes location do not
> > > get correctly autodetected (and hence there is no -I
> > > /usr/include/libxml2 on gcc command-lines).
> > 
> > Yep.  It's a %^#&*%^ missing include.  strdup is being implicitly
> > declared to return `int' which gets truncated, so the program segfaults
> > and we lose.  Here's a patch:
> > 
> > --- aclocal.m4.orig	Tue Mar 26 09:18:14 2002
> > +++ aclocal.m4	Tue Mar 26 09:19:31 2002
> > @@ -4889,6 +4889,7 @@
> >        AC_TRY_RUN([
> >  #include <stdlib.h>
> >  #include <stdio.h>
> > +#include <string.h>
> >  #include <xmlversion.h>
> >  
> >  int 
> > @@ -4902,7 +4903,7 @@
> >  
> >    /* Capture xml2-config output via autoconf/configure variables */
> >    /* HP/UX 9 (%@#!) writes to sscanf strings */
> > -  tmp_version = (char *)strdup("$min_xml_version");
> > +  tmp_version = strdup("$min_xml_version");
> >    if (sscanf(tmp_version, "%d.%d.%d", &major, &minor, &micro) != 3) {
> >       printf("%s, bad version string from xml2-config\n", "$min_xml_version");
> >       exit(1);
> > @@ -4910,7 +4911,7 @@
> >     free(tmp_version);
> >  
> >     /* Capture the version information from the header files */
> > -   tmp_version = (char *)strdup(LIBXML_DOTTED_VERSION);
> > +   tmp_version = strdup(LIBXML_DOTTED_VERSION);
> >     if (sscanf(tmp_version, "%d.%d.%d", &xml_major_version, &xml_minor_version, &xml_micro_version) != 3) {
> >       printf("%s, bad version string from libxml includes\n", "LIBXML_DOTTED_VERSION");
> >       exit(1);
> > 
> > -- 
> > Revolutions do not require corporate support.
> 
> Best regards,
> -- 
> Yann Dirson    <ydirson at altern.org> |    Why make M$-Bill richer & richer ?
> Debian-related: <dirson at debian.org> |   Support Debian GNU/Linux:
> Pro:    <yann.dirson at fr.alcove.com> |  Freedom, Power, Stability, Gratuity
>      http://ydirson.free.fr/        | Check <http://www.debian.org/>