On Wed, Apr 04, 2007 at 09:14:12PM -0600, Eric Blake wrote:
> -BEGIN PGP SIGNED MESSAGE-
> Hash: SHA1
>
> According to Albert Chin on 4/4/2007 8:59 PM:
> >
> > Unfortunately, "#include_next " doesn't include
> > /usr/include/stdio.h. It includes "./stdio.h", the gnulib version of
> > stdi
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Bruno Haible on 4/4/2007 5:47 PM:
>
> OTOH, if it causes tests/test-stdint.c to not compile, then it is best
> to mark the part of the test that doesn't work with "#ifndef __CYGWIN__".
> So that the other tests can still be executed witho
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Albert Chin on 4/4/2007 8:59 PM:
>
> Unfortunately, "#include_next " doesn't include
> /usr/include/stdio.h. It includes "./stdio.h", the gnulib version of
> stdio.h.
If you were to change the gnulib stdio.h to use #include_next instead
Tru64 UNIX ships with a C compiler. A commercial version of the
compiler is also available, with extra includes in /usr/include.dtk,
some of which add to the includes in /usr/include. The includes in
/usr/include.dtk include the files in /usr/include with #include_next.
However, because gnulib now
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Bruno Haible on 4/4/2007 6:28 PM:
> On FreeBSD 6.1, printf of "%010a" of Infinity and NaN yields
> "000inf" and "000nan", respectively.
>
> While on glibc systems, it yields
> " inf" and " nan", respectively.
>
> I fi
FreeBSD 6.1 appeared as one of the very few platforms that implement
printf "%a". Well, it implements it, but not correctly: The testsuite
revealed that for "%.2a" of 1.51 it returns "0x1.82p+0" instead of
"0x1.83p+0".
2007-04-04 Bruno Haible <[EMAIL PROTECTED]>
* m4/printf.m4 (gl_PRINT
On FreeBSD 6.1, printf of "%010a" of Infinity and NaN yields
"000inf" and "000nan", respectively.
While on glibc systems, it yields
" inf" and " nan", respectively.
I find glibc's output more sensible, since strtod will accept it, while
strtod will not grok "000inf". But F
Eric Blake wrote:
> It appears you forgot to commit this?
Yes. Done now.
> > Depends-on:
> > math
> > + fpucw
>
> And is there any reason you are undoing the alphabetic sort, other than that
> is
> what the file used to have before my patch?
Yes, keeping the natural order makes it easier
A while ago I proposed to create a list of things to add to portable
tools. Then I forgot. Apologies. Thanks to Karl for reminding me!
* Paul Eggert wrote on Thu, Apr 05, 2007 at 12:49:26AM CEST:
> Eric Blake <[EMAIL PROTECTED]> writes:
>
> > Is it time to propose that cut and join be added
>
Eric Blake wrote:
> As of today, it is fixed in CVS (the eventual cygwin 1.7.0), but
> tests/test-stdint will fail on the current cygwin 1.5.24.
You mean, tests/test-stdint.c compiles fine but gives a failure during
"make check"? Then a comment near the top of this file is appropriate,
saying "Not
Bruno Haible <[EMAIL PROTECTED]> writes:
> Any objection to removing the mention of 'tar' and 'gzip' here?
No, that's fine. Thanks.
Eric Blake wrote:
> Lately, many packages ship both .gz and .bz2 tarballs, although automake
> still
> defaults to .gz only. Is it time to more heavily encourage .bz2?
bzip2 is about 6 times slower upon decompression:
$ time gunzip -c < coreutils-6.9.tar.gz > /dev/null
real0m1.982s
user
Eric Blake wrote:
> Also, should we mention git and/or CVS?
I don't think so. Once a user has a gnulib copy, he can run gnulib-tool without
needing git nor cvs.
Bruno
Paul Eggert wrote:
> Add shell, coreutils, diffutils, grep, tar, gzip.
What are 'tar' and 'gzip' used for? For building distribution tarballs.
But that's outside the scope of gnulib. It's done through a Makefile rule
produced by 'automake'. Therefore I consider that 'tar' and 'gzip' are
mere
Eric Blake <[EMAIL PROTECTED]> writes:
> Is it time to propose that cut and join be added
join perhaps; it's quite stable so long as you run it in the C locale
and stick to the old features. cut I'm not so sure about; it's kind
of persnickety.
If we were adding portable tools, awk would be the
Paul Eggert CS.UCLA.EDU> writes:
>
> I tweaked some more wording, as follows:
>
> 2007-04-04 Paul Eggert cs.ucla.edu>
>
> * DEPENDENCIES: Give overall description of version dependency
> desirability. Use more-typical names for apps.
> Add shell, coreutils,
Is it time to
I tweaked some more wording, as follows:
2007-04-04 Paul Eggert <[EMAIL PROTECTED]>
* DEPENDENCIES: Give overall description of version dependency
desirability. Use more-typical names for apps.
Add shell, coreutils, diffutils, grep, tar, gzip.
--- DEPENDENCIES4
"Schwarz, Konrad" <[EMAIL PROTECTED]> writes:
> What is the justification for making the return value of read() have type
> ssize_t?
Again, I wasn't there. But when ssize_t was invented, C89 (and
therefore ptrdiff_t) was not required by POSIX, so I expect the
original motivation was that POSIX w
Bruno Haible clisp.org> writes:
> > * modules/printf-frexpl (Depends-on): Depend on ldexpl.
>
> Thanks for the quick fix. But this is overkill: the module 'ldexpl' looks
> for the ldexpl() function also in libm, and printf-frexpl doesn't this test
> result.
>
> 2007-04-03 Bruno Haible cl
[EMAIL PROTECTED] (Karl Berry) writes:
> Running MODULES.html.sh --cvs-urls reports:
>
> gnulib-tool: module md5 doesn't exist
> gnulib-tool: module sha1 doesn't exist
> gnulib-tool: module iconvme doesn't exist
>
> [plus four more instances of each msg]
>
> Help?
Fixed.
/Simon
? MODULES.html
I
Bruno Haible <[EMAIL PROTECTED]> writes:
> Simon Josefsson wrote:
>> > Also, AC_PROG_YACC doesn't do this. It simply prints one line of output,
>> > like most other autoconf tests.
>>
>> Right, but AC_PROG_YACC doesn't test whether there is a yacc tool
>> installed.
>
> Huh? Autoconf's doc says:
Running MODULES.html.sh --cvs-urls reports:
gnulib-tool: module md5 doesn't exist
gnulib-tool: module sha1 doesn't exist
gnulib-tool: module iconvme doesn't exist
[plus four more instances of each msg]
Help?
Thanks,
k
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Paul Eggert on 4/3/2007 5:24 PM:
> Eric Blake <[EMAIL PROTECTED]> writes:
>
>> cygwin currently defines wint_t as unsigned int (legal per POSIX), but then
>> tries to use INT_MIN and INT_MAX for WINT_MIN and WINT_MAX. I've reported
>>
Simon Josefsson wrote:
> > Also, AC_PROG_YACC doesn't do this. It simply prints one line of output,
> > like most other autoconf tests.
>
> Right, but AC_PROG_YACC doesn't test whether there is a yacc tool
> installed.
Huh? Autoconf's doc says:
-- Macro: AC_PROG_YACC
If `bison' is found, s
Ralf Wildenhues wrote:
> > * GNU autoconf 2.59 or newer.
> [...]
> > * GNU automake 1.10 or newer.
>
> This does not make sense. Automake-1.10 requires Autoconf-2.60 or newer.
Yup. Thanks, only automake-1.9.6 should be the minimum requirement. Also
some "good English" tweaks:
--- DEPENDENCIES
Bruno Haible <[EMAIL PROTECTED]> writes:
> Simon Josefsson wrote:
>> For example, in GnuTLS, I recently received a patch that added a
>> configure-time warning about missing non-required tools:
>>
>> AC_PATH_PROG([GAA], [gaa])
>> if test "x$GAA" = "x"; then
>>AC_MSG_WARN([[***
>> *** GAA was
26 matches
Mail list logo
