Re: glob_.h & glibc

[Derek Price]

Paul Eggert wrote:

>Hmm, actually they provisionally accepted the bug-1060 changes except
>for the part about using prototypes when defining external functions.
>  
>

Oh, that's what that meant.  I was hoping someone else would say
something if it was important.  Thanks.  :)

>where glob-libc.h is an unmodified copy of the glibc glob.h.
>
>Or perhaps the "...define our own stuff..." could all be in config.h,
>which would mean we wouldn't need two files.
>  
>

Well, there were 3 #include statements that needed to be in glob_.h,
before including glob-glibc.h, so I went with the extra include file. 
I've attached a gnulib patch for review.  It compiles here, in Linux CVS.

There was also a small subset of changes that I can't see how to remove
from glob-glibc.h.  I've attached the new glibc->gnulib patch.  It is
much smaller now, so perhaps Roland will find it acceptable.  (Actually,
4 lines, the dirent/direct change, might also be avoidable with
something like a global "#define dirent direct".  It would be a small
GNULIB module, but a module that did this, included a few header files,
and defined a few macros like D_NAMLEN could replace 10-15 lines in each
of five or six modules.)

>Could you please follow up on this in bug-gnulib first?  I'd rather
>not burden Roland with gnulib issues more than I can help it.  Thanks.
>  
>

That's why I started this thread on bug-gnulib in the first place.  :)

Regards,

Derek

-- 
Derek R. Price
CVS Solutions Architect
Ximbiot <http://ximbiot.com>
v: +1 717.579.6168
f: +1 717.234.3125
<mailto:address@hidden>