Re: gnulib woes & run-octave fails, suggestions?

[Jarno Rajahalme]

On Apr 7, 2010, at 9:53 AM, ext John W. Eaton wrote:

> I guess this is a bug in gnulib then, because it does provide a
> replacement for fchwonat in the openat module.  And it appears to me
> is correctly included in the list of modules, according to the output
> of the bootstrap script that autogen.sh runs.  So I don't understand
> why this does not result in actually compiling a replacement.
> 
> Does your build tree have a file libgnu/.libs/fchownat.o?  What
> symbols are defined there?  I'm guessing that there will be a
> rpl_fchownat function, but it is not being used correctly by whatever
> other function needs it.
> 


I have it, and the function is named rpl_fchownat. I think there is no problem 
USING it, as the compilation succeeds, if I comment out the CXXALIASWARN line. 
I think the problem is that the alias warning is "attached to" an non-existing 
function (fchownat).

> This problem should be reported to the gnulib maintainers, as it needs
> to be fixed in gnulib.
> 
> | Also, it seems that the __warning__ in GCC 4.4.3 (or 4.4.2, tried
> | both) is not working properly, as I get over a hundred of these
> | warnigns, even when the source has the gnulib:: before the
> | respective function name.
> 
> It works correctly for me with GCC 4.4.3 on my Debian system, and I
> really have no clue why it is failing for you.
> 
> jwe