Hi Jim,
On 2008-10-13 you wrote:
> Does anyone object to my installing a server-side hook
> that would prevent pushing merge commits on master?
> In our experience here (with gnulib.git), pushing a merge
> commit is always unintentional.
The one that I pushed today was unintentional again. Sorry
Bruno Haible writes:
> @@ -133,10 +133,10 @@
> struct tm t;
> char buf[16];
> int main () {
> - /* On BeOS, locales are not implemented in libc. Rather, libintl
> + /* On BeOS nd Haiku, locales are not implemented in libc. Rather, libintl
s/nd/and/
> imitates locale dependent behav
The doc of many functions that take a wchar_t argument or return wchar_t
mentions the danger of using them on platforms where wchar_t is only a
16-bit type. This patch applies this boilerplate warnings to more functions.
2008-12-22 Bruno Haible
* doc/posix-functions/iswalnum_l.texi: M
Hi Simon,
I propose this change to the getaddrinfo module with two effects:
- Allow package maintainers to decide which programs to link with -lsocket
(on Solaris) or -lws2_32 (on native Windows). Not all programs in a
package need to link with the library that contains getaddrinfo.
Hi Jim,
The Haiku port prompted me to look at this comment. Looking at the sources of
statvfs() for the various glibc platforms, I found that it's not BeOS which
is special, but Linux: It's only the Linux specific implementation which looks
at /proc/mounts. OK to commit this update?
2008-12-22
The discussion on 2008-11-05 led to an agreement how to best fix this, but
no one actually committed the fix. Now done as follows:
2008-12-22 Ingo Weinhold(tiny change)
Eric Blake
Paolo Bonzini
Bruno Haible
Make c-stack work on Haiku.
I had this in my pipe for Haiku since 2008-11-16. Applied now:
2008-12-22 Bruno Haible
* m4/locale-fr.m4 (gt_LOCALE_FR_UTF8): Treat Haiku like BeOS.
--- m4/locale-fr.m4.orig2008-12-23 00:48:01.0 +0100
+++ m4/locale-fr.m4 2008-12-23 00:38:43.0 +0100
@@ -1,4
> New module 'mbrlen'.
mbrlen() has essentially the same set of platform dependent bugs as mbrtowc().
This patch provides a workaround:
2008-12-22 Bruno Haible
Work around mbrlen() bugs on AIX, HP-UX, OSF/1, Solaris.
* m4/mbrlen.m4 (gl_FUNC_MBRLEN): Set REPLACE_MBRLEN i
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Paolo Bonzini on 12/22/2008 2:59 AM:
> I'm committing this:
>
> 2008-12-22 Paolo Bonzini
>
> * modules/regex: Request emulations for the mb*/wc* functions we
> need.
> * m4/regex.m4: Don't look for those functions he
On Mon, 2008-12-22 at 12:45 +0100, Bruno Haible wrote:
> Alan Hourihane wrote:
> > The only commitment I'm asking from gnulib is to apply patches that we
> > submit. They'll be #ifdef __MINT__ anyway, isn't that acceptable ??
>
> A patch with #ifdef __MINT__ certainly helps us understand the issue
On Mon, 2008-12-22 at 12:41 +0100, Bruno Haible wrote:
> Alan Hourihane wrote:
> > > Did you try to run the complete gnulib unit tests?
> > > $ ./gnulib-tool --test --with-tests
> >
> > I have run individual tests, but I can run the whole suite.
>
> In your situation, it is easier (less time co
Alan Hourihane wrote:
> The only commitment I'm asking from gnulib is to apply patches that we
> submit. They'll be #ifdef __MINT__ anyway, isn't that acceptable ??
A patch with #ifdef __MINT__ certainly helps us understand the issue. But
we need a plain-text description of the issue as well. Just
Alan Hourihane wrote:
> > Did you try to run the complete gnulib unit tests?
> > $ ./gnulib-tool --test --with-tests
>
> I have run individual tests, but I can run the whole suite.
In your situation, it is easier (less time consuming) to do
./gnulib-tool --create-testdir --with-tests --dir=/
Hi,
Now, here is a patch to make use of the mbrtowc emulation on platforms that
lack it (essentially HP-UX 11.00, IRIX 6.5, Solaris 2.6).
It causes an increase of code size on these platforms. But this is actually
a bug fix: gnulib code was assuming that is !HAVE_MBRTOWC, there are only
unibyte l
On Mon, 2008-12-22 at 11:48 +0100, Bruno Haible wrote:
> Alan Hourihane wrote:
> > > > I have more patches to gnulib for MINT. Shall I just file them as bugs ?
> > >
> > > It depends how serious MINT as a platform is. What is MINT at all? Why
> > > does
> > > it lack basic functions like mbrtowc,
Alan Hourihane wrote:
> > > I have more patches to gnulib for MINT. Shall I just file them as bugs ?
> >
> > It depends how serious MINT as a platform is. What is MINT at all? Why does
> > it lack basic functions like mbrtowc, standardized in ANSI C Amendment 1?
>
> MINT runs on the Atari ST.
Oh
Bruno Haible ha escrit:
> This is done.
Thanks, Bruno!
Regards,
Sergey
I'm committing this:
2008-12-22 Paolo Bonzini
* modules/regex: Request emulations for the mb*/wc* functions we
need.
* m4/regex.m4: Don't look for those functions here.
diff --git a/m4/regex.m4 b/m4/regex.m4
index 38fb7ba..f3a7b42 100644
--- a/m4/regex.m4
+++ b/m4/rege
On Mon, 2008-12-22 at 10:49 +0100, Bruno Haible wrote:
> Alan Hourihane wrote:
> > I have more patches to gnulib for MINT. Shall I just file them as bugs ?
>
> It depends how serious MINT as a platform is. What is MINT at all? Why does
> it lack basic functions like mbrtowc, standardized in ANSI C
Alan Hourihane wrote:
> I have more patches to gnulib for MINT. Shall I just file them as bugs ?
It depends how serious MINT as a platform is. What is MINT at all? Why does
it lack basic functions like mbrtowc, standardized in ANSI C Amendment 1?
Is this platform in active development? If so, it
On Mon, 2008-12-22 at 03:55 +0100, Bruno Haible wrote:
> Hello Sergey, Alan,
>
> On 2008-10-30 I wrote:
> > I'll therefore extend the modules in a way that they work also on older
> > systems without HAVE_MBRTOWC.
>
> This is done. When you do a "gnulib-tool --update" from texinfo, you should
> g
21 matches
Mail list logo
