Bruno Haible <[EMAIL PROTECTED]> writes:
> Paul Eggert wrote:
>> Clearly you don't write financial applications dealing with 30-year
>> bonds. :-)
>
> So you want that 'time64_t' type to be useful specifically for financial
> applications?
I was kidding! (But I'm afraid others are not: I have h
Albert Chin wrote:
> 1 2 3 4 5 6 7
> OpenBSD 4.0: . # . . . . .
> HP-UX 10.20: # # . . . . #
Thanks, I added this to printf.m4.
> HP-UX 10.20's libc has snprintf() but there is _no_ function pro
FYI, this change removed 1500 lines of ChangeLog content:
Author: Bruno Haible <[EMAIL PROTECTED]>
Date: Sun Feb 18 15:10:28 2007 +
New module 'math'. replaces mathl.h.
I presume that was accidental and have restored those lines.
Using git, you can see the diff like this:
git-diff
Paul Eggert wrote:
Perhaps we can nudge this process forward a bit by designing an
Autoconf macro that causes configure/make to default to 64-bit builds
on platforms that can support either 32 or 64.
...just provide a way to turn it off. :-)
I build on one box for an entire platform (e.g. "x86
Jim Meyering wrote:
> FYI, this change removed 1500 lines of ChangeLog content:
> I presume that was accidental and have restored those lines.
Thanks and sorry about it. This apparently happens when I don't wait long
enough after saving the ChangeLog and before doing the "cvs commit".
Bruno
Hi,
For an fprintf emulation, one needs the ability to set the "error indicator"
bit in a FILE stream. Surprisingly, this is not possible in a portable and
efficient way. Like __fpending, this is one of those functionalities that
the ANSI/ISO C people forgot to standardize.
2007-03-08 Bruno Haib
Albert Chin asked:
> Does gl_cv_func_vsnprintf_posix=yes on AIX 5.1, and Solaris 6-9/SPARC?
No. According to the table at the end of m4/printf.m4, the first and second test
will fail on these platforms.
Bruno
Paul Eggert wrote:
> >- or the bash maintainers to make 'expr' and 'sed' shell built-ins,
> > like 'test' and 'printf' are?
>
> I think it unlikely that either would happen; neither sounds all that
> practical.
Why not? 'expr' and 'sed' could ship its code in the form of shared libraries
Paul Eggert wrote:
> The main reason I want 64-bit time_t is so that my apps can stat any
> file without getting EOVERFLOW. ... on 32-bit platforms
If the time_t type is smaller than the type used for timestamps in the
particular filesystem, you can get EOVERFLOW during stat().
If, on the other h
This is needed at least on glibc-2.3.6 systems, to avoid weird compilation
errors.
2007-03-08 Bruno Haible <[EMAIL PROTECTED]>
* lib/stdio_.h: Treat __need___FILE like __need_FILE.
*** lib/stdio_.h7 Mar 2007 03:47:50 - 1.6
--- lib/stdio_.h9 Mar 2007 02:25:51 -
Here comes the full POSIX variant of vfprintf.
2007-03-08 Bruno Haible <[EMAIL PROTECTED]>
* modules/vfprintf-posix: New file.
* lib/vfprintf.c: New file.
* m4/vfprintf-posix.m4: New file.
* m4/stdio_h.m4 (gl_STDIO_H_DEFAULTS): Set also GNULIB_VFPRINTF_POSIX,
And here's the full POSIX compliant fprintf() module.
2007-03-08 Bruno Haible <[EMAIL PROTECTED]>
* modules/fprintf-posix: New file.
* lib/fprintf.c: New file.
* m4/fprintf-posix.m4: New file.
* m4/stdio_h.m4 (gl_STDIO_H_DEFAULTS): Set also GNULIB_FPRINTF_POSIX,
Here comes the POSIX compatible vprintf() substitute.
2007-03-08 Bruno Haible <[EMAIL PROTECTED]>
* modules/vprintf-posix: New file.
* lib/vprintf.c: New file.
* m4/vprintf-posix.m4: New file.
* m4/stdio_h.m4 (gl_STDIO_H_DEFAULTS): Set also GNULIB_VPRINTF_POSIX,
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Matthew Woehlke on 3/8/2007 10:59 AM:
> Matthew Woehlke wrote:
>> Well, it seems to have gotten past lib/ anyway. Will report back with
>> how the test suite goes...
>
> Ok, 'make check results'... I got this:
>
> Checking ./012.macro_ar
Bruno Haible <[EMAIL PROTECTED]> writes:
> if (dup2 (fd2, fd) < 0)
> /* Whee... we botched the stream and now cannot restore it! */
> abort ();
Hi! Consider making it possible to not use this workaround, but
rather fail with a compilation error. I wouldn't want code that
invo
15 matches
Mail list logo
