Hi Jim,
* Jim Meyering wrote on Mon, Jan 07, 2008 at 11:34:19PM CET:
> Ralf Wildenhues <[EMAIL PROTECTED]> wrote:
> >
> > * DEPENDENCIES: Require Texinfo version 4.6 or newer.
>
> Considering some people (like me :) are using texinfo-4.11,
> and 4.6 is over 4 years old, that is a shoe-in.
Ap
On Monday 07 January 2008, Jim Meyering wrote:
> Mike Frysinger <[EMAIL PROTECTED]> wrote:
> > On Saturday 29 December 2007, Mike Frysinger wrote:
> >> On Wednesday 25 July 2007, Mike Frysinger wrote:
> >> > posted this to the coreutils a list sometime ago ... idea is to be
> >> > able to disable a
Bruno Haible <[EMAIL PROTECTED]> writes:
> It's a moving target, but it can be defined roughly like this:
> - All platforms with a Unix like API, that are not older than ca. 6 years.
> This leads approximately to this list:
> glibc >= 2.1, MacOS X >= 10.2, FreeBSD >= 4.8, NetBSD >= 3.0
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Bruno Haible on 1/7/2008 5:58 PM:
| I'm therefore in favour of putting this description under "Particular
Modules".
|
| If gnulib implements a significant number of functions specified by glibc
| - strsignal was just mentioned - then I'm
I seem to have lost Jim's reply, but here is a patch. Works on
aix-4.3.3 and aix-5.3.
The SBITS thing is the way the libperstat.h header recommends
calculating the loadavg:
/* To calculate the load average, divide the numbers by (1<. */
Of course, it is not defined on aix4.3.3 ...
Peter
diff --
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Peter Fales on 1/7/2008 11:53 AM:
|> | This change fixes the problem in in test-wcwidth. There is one other
|> | failure in test-fseeko.sh when building on RedHat 7.3.
|>
|> More details, please?
|
| I'm guessing this due to an incompati
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Bruno Haible on 1/7/2008 5:54 PM:
|
| Thanks for the comments; they are essential for understanding the code.
| I also find your comments clearer than the text in .../~lecroq/...
Thanks for your review. Also, is there any reason to have
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
[adding bug-gnulib]
According to Peter Miller on 1/7/2008 5:15 PM:
| On Sat, 2008-01-05 at 21:51 -0700, Eric Blake wrote:
|> glibc 2.6.1 is quadratic, gnulib is linear. For worst-case scenarios,
|> gnulib's implementation is hands-down better, altho
Micah Cowan wrote:
> Is there a general guideline for what constitutes "Gnulib's supported
> platforms"?
It's a moving target, but it can be defined roughly like this:
- All platforms with a Unix like API, that are not older than ca. 6 years.
This leads approximately to this list:
glib
Ralf Wildenhues wrote:
> Bruno, I suppose config.rpath and config.libpath would need adjustment
> as well. Should I send a patch?
Normally, I update these files after an official libtool release. Once
a libtool-1.5.26 is released, I'll update these. But in this case, the
diffs are probably small
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
Bruno Haible wrote:
> Colin Watson wrote:
>> Do you happen to know whether I was overdoing it by adding in #ifdefs
>> for SIGHUP, SIGINT, etc.? I decided better safe than sorry, but didn't
>> know whether there was some basic set of signals that Gnulib
Simon Josefsson wrote:
> +2008-01-06 Simon Josefsson <[EMAIL PROTECTED]>
> +
> + * tests/test-localename.c: Don't include obsolete "setenv.h".
> +
> + * modules/localename-tests (Depends-on): Need unsetenv.
> +
It's good. Thank you; it was my mistake. Please apply.
Bruno
Colin Watson wrote:
> --- a/users.txt
> +++ b/users.txt
> @@ -34,6 +34,7 @@ The following packages appear to be using gnulib and
> gnulib-tool:
>libvirt http://libvirt.org/
>http://git.et.redhat.com/?p=libvirt.git;a=summary
>m4 http://git.sv.gnu.org
Colin Watson wrote:
> Do you happen to know whether I was overdoing it by adding in #ifdefs
> for SIGHUP, SIGINT, etc.? I decided better safe than sorry, but didn't
> know whether there was some basic set of signals that Gnulib's supported
> platforms were guaranteed to have.
You were right. The s
Eric Blake wrote:
> According to Bruno Haible on 12/21/2007 6:42 AM:
> | does not define NULL. Better write (void*)0 or "" instead of
> NULL.
>
> Which platforms have a broken that fails to define NULL?
Actually all platforms I have access to today define NULL in .
Probably the latest platform
Karl Berry wrote:
> 2007-12-30 Karl Berry <[EMAIL PROTECTED]>
> * doc/gnulib.texi (Library vs. Reusable Code): remove period, to
> work around defect in Texinfo and/or the standalone Info browser.
>
> Rationale: this is simply the most expedient way to work around the
> present subop
Ralf Wildenhues wrote:
> I think that an lt-program* outside [._]libs/ would be a bug.
Thanks for this statement. On platforms where gnulib is used, the
directory name is '.libs', not '_libs'. I'm doing as Paul suggested,
but with a simpler patch (less micro-optimizations -> easier to understand).
Eric Blake wrote:
> I'm installing these two commits:
>
> * modules/memmem-tests (configure.ac): Check for alarm.
> * tests/test-memmem.c (main): Avoid alarm on platforms that lack it.
Thanks for these.
> * doc/functions/memmem.texi: New file.
> * doc/gnulib.texi (Function Substitutes): Add memm
Colin Watson wrote:
> I use strsignal in man-db, and would like a Gnulib module to cope with
> its portability problems. Here's one which seems to be doing the right
> thing for me so far. This is my first attempt at writing a Gnulib module
> from scratch, so I'd appreciate any comments, style or o
Eric Blake wrote:
> I've finished my implementation of a Two-Way plus Boyer-Moore hybrid
> string search algorithm for the memmmem module. This patch has passed
> everything I've thrown at it so far, and it has the nice properties of
> avoiding alloca/malloc (hence it is async-safe), as well as al
Paul Eggert wrote:
> For example, one could build a program that uses gcc's libiberty along
> with some gnulib code. The libiberty code won't use Gnulib's
> substitute, but that's OK; it will still work just as well
> as it does now.
>
> So I still don't see the downside of making that change.
Hi Sylvain,
Sylvain Beucler wrote:
> > > Attached is a small patch to fix a couple non-fred memory blocks. They
> > > are not a big deal, but they do produce noise when analyzing programs
> > > with memory checkers such as Valgrind.
> I do attach the patch now :)
Thanks for the patch. I can even
Ralf Wildenhues <[EMAIL PROTECTED]> wrote:
> The standards.texi document uses @/ for splitting of long lines
> (in URLs), this syntax requires a newer Texinfo than 4.2.
> OK to apply?
>
> * DEPENDENCIES: Require Texinfo version 4.6 or newer.
Considering some people (like me :) are using texi
Peter Fales wrote:
> This change fixes the problem in in test-wcwidth.
Thanks for confirming this. I'm applying this change to the test. (One
can hardly say "fix", since the behaviour of various Unicode characters in
wcwidth is not clearly standardized.)
2008-01-01 Bruno Haible <[EMAIL PROTECTE
Hello,
The standards.texi document uses @/ for splitting of long lines
(in URLs), this syntax requires a newer Texinfo than 4.2.
OK to apply?
Cheers,
Ralf
* DEPENDENCIES: Require Texinfo version 4.6 or newer.
diff --git a/DEPENDENCIES b/DEPENDENCIES
index ed051be..a8f9a9d 100644
--- a/D
Hello,
FYI, I updated gnulib's copies of the config files, to have AIX 6.1
support.
Bruno, I suppose config.rpath and config.libpath would need adjustment
as well. Should I send a patch?
Cheers,
Ralf
On Mon, Jan 07, 2008 at 01:06:38PM -0800, Paul Eggert wrote:
> Thanks for tackling this. My kneejerk reaction: as a general rule, we
> prefer the differences between gnulib and the GNU C library to be
> localized to small sections of the code, and protected by #if !_LIBC,
> the idea being that ide
Thanks for tackling this. My kneejerk reaction: as a general rule, we
prefer the differences between gnulib and the GNU C library to be
localized to small sections of the code, and protected by #if !_LIBC,
the idea being that ideally the gnulib and GNU C Library versions can
be identical. Please
Hi Eric,
* Eric Blake wrote on Sun, Jan 06, 2008 at 02:23:33AM CET:
>
> I'd appreciate any reviews before checking it in.
Here's a rough glance at it. FWIW, the diff is not very readable
(there was a patch to diffutils out there for --more-readable).
> +/* We use the Two-Way string matching al
On Sun, Jan 06, 2008 at 06:26:16AM -0700, Eric Blake wrote:
> -BEGIN PGP SIGNED MESSAGE-
> Hash: SHA1
>
> According to Peter Fales on 12/30/2007 11:46 PM:
> | This change fixes the problem in in test-wcwidth. There is one other
> | failure in test-fseeko.sh when building on RedHat 7.3.
>
"Peter O'Gorman" <[EMAIL PROTECTED]> wrote:
> While getloadavg works ok on aix4, it returns incorrect results on
> aix5.
>
> Why do we not use libperfstat on aix?
> http://publib.boulder.ibm.com/infocenter/pseries/v5r3/topic/com.ibm.aix.basetechref/doc/basetrf1/perfstat_cputot.htm
> It has the adva
Hi,
While getloadavg works ok on aix4, it returns incorrect results on
aix5.
Why do we not use libperfstat on aix?
http://publib.boulder.ibm.com/infocenter/pseries/v5r3/topic/com.ibm.aix.basetechref/doc/basetrf1/perfstat_cputot.htm
It has the advantage of not requiring root privs to get the load
Eric Blake said:
> Should POSIX require linear (rather than quadratic) worst-case performance
> in strstr? Or maybe we just leave this improvement as a
> quality-of-implementation issue, but perhaps document in the application
> notes document that many implementations use less-than-stellar algori
Mike Frysinger <[EMAIL PROTECTED]> wrote:
> On Saturday 29 December 2007, Mike Frysinger wrote:
>> On Wednesday 25 July 2007, Mike Frysinger wrote:
>> > posted this to the coreutils a list sometime ago ... idea is to be able
>> > to disable automatic ACL detection via --disable-acl since this can o
34 matches
Mail list logo
