[Lsb-infrastructure] First look at April 18 updates

[Denis Silakov]

Wichmann, Mats D wrote:
> I don't know if the builder machine broke, or what
> is happening.
>   

Not clear for me, too. I guess that autobuilders on all platform use the
same headers and all except the IA32 one like the first line of headers.

> GL/glxext.h - it doesn't look like we can use the
> new headers in isolation. I've added the two new GL
> headers to the mkfilelists exclusion for now. The
> error becomes:
>
> In file included from /usr/include/GL/glx.h:300,
>                  from GL_glx_h.c:13:
> /opt/lsb/include/GL/glxext.h:23: error: expected '=', ',', ';', 'asm' or
> '__attribute__' before 'glXGetProcAddressARB'
> GL_glx_h.c: In function 'GL_glx_h':
> GL_glx_h.c:1429: error: '__GLXextFuncPtr' undeclared (first use in this
> function)
> GL_glx_h.c:1429: error: (Each undeclared identifier is reported only
> once
> GL_glx_h.c:1429: error: for each function it appears in.)
>
> I've disabled the two additional GL headers from going
> into the lsb-build package for now.
>
>   

Indeed, '*ext' headers are of no use without normal ones. I actually
suggest to turn on all GL headers with corrections attached in bug 706
(http://bugs.linuxbase.org/show_bug.cgi?id=706).

Generated GL headers look ok and and devchk can be compiled with them
(some corrections from devchk might be still required - such as ATsize,
TMoffset, etc.). It would be useful to have some feedback from tests
that use GL headers.

> Printing test:
>
> conf.c: In function 'get_address':
> conf.c:2112: error: 'h_errno' undeclared (first use in this function)
>
> The definition of h_errno as vanished from netdb.h, it now looks like
> this:
>
> #if __LSB_VERSION__ < 21
> #define h_errno (*__h_errno_location ())
> #endif
>
> it is just outright included in the 3.2-branch headers.
>   

This concerns 'srconly' constants discussed in bug 2039
(http://bugs.linuxbase.org/show_bug.cgi?id=2039) and should be fixed by
the recent patch from that bug.


-- 
Regards,
Denis.