Re: recent argp changes

2007-05-01 Thread Sergey Poznyakoff
Bruno Haible <[EMAIL PROTECTED]> ha escrit: > - Add a field 'array_index' to the hol_entry struct. Fill these fields > right before calling qsort. Actually, that was my first idea, but the trick with pointers seemed easier:) I have rewritten it this way. Thank you. Regards, Sergey

Re: recent argp changes

2007-05-01 Thread Ben Pfaff
Paul Eggert <[EMAIL PROTECTED]> writes: > Bruno Haible <[EMAIL PROTECTED]> writes: > >> you are now assuming an in-place sorting algorithm. If a libc were >> to implement qsort by use of mergesort (and temporary storage), such >> a sorting algorithm would pass you two pointers to hol_entrys in >>

Re: doc: new chapter about functions

2007-05-01 Thread Paul Eggert
Bruno Haible <[EMAIL PROTECTED]> writes: > Jim, Paul, I don't understand whether the 'strftime' module is useful when > someone only wants a POSIX compliant strftime() function. Does it fix known > POSIX incompatibilities on some systems, or does it only provide GNU > extensions? Currently it is

Re: recent argp changes

2007-05-01 Thread Paul Eggert
Bruno Haible <[EMAIL PROTECTED]> writes: > you are now assuming an in-place sorting algorithm. If a libc were > to implement qsort by use of mergesort (and temporary storage), such > a sorting algorithm would pass you two pointers to hol_entrys in > different array. A minor quibble: the C standar

Re: doc: new chapter about functions

2007-05-01 Thread Bruno Haible
Eric Blake wrote: > The creation of both doc/functions/_exit.texi and doc/functions/_Exit.texi > breaks checkout on case insensitive file systems. We need to come up with an > alternate name for one of these files. Ah, right. I rename it to _Exit_C99.texi. Bruno

argp on mingw

2007-05-01 Thread Bruno Haible
The test-argp-2.sh test fails on mingw, because the program name comes out as "test-argp.exe", whereas the test expects "test-argp". This fixes it. OK to commit? 2007-05-01 Bruno Haible <[EMAIL PROTECTED]> * tests/test-argp-2.sh (func_compare): Drop .exe suffix. *** tests/test-argp-2.

argp needs sleep

2007-05-01 Thread Bruno Haible
On mingw 2005, the argp module fails to link: gllib/argp-parse.c:128: undefined reference to 'sleep' This fixes it; I'm committing it. 2007-05-01 Bruno Haible <[EMAIL PROTECTED]> * modules/argp (Depends-on): Add sleep. *** modules/argp9 Sep 2006 05:14:25 - 1.15 ---

new module 'sleep'

2007-05-01 Thread Bruno Haible
The function sleep() is missing on mingw 2005 (although it was present in some older versions of mingw). This provides a substitute. 2007-05-01 Bruno Haible <[EMAIL PROTECTED]> * modules/sleep: New file. * lib/sleep.c: New file. * m4/sleep.m4: New file. * lib/uni

Re: doc: new chapter about functions

2007-05-01 Thread Eric Blake
Bruno Haible clisp.org> writes: > > Like the documentation of the header files, I'm adding a documentation of > each ISO C 99 or POSIX function to the gnulib documentation. > > + * _Exit:: > + * _exit:: The creation of both doc/functions/_exit.texi and doc/functions/_Exit.texi breaks checkou

Re: recent argp changes

2007-05-01 Thread Ben Pfaff
Bruno Haible <[EMAIL PROTECTED]> writes: >> The option sorting in argp-help was based on the stability >> of sort algorithm, therefore the test failed on systems with >> quicksort qsort implementations. It went unnoticed on glibc, >> because its qsort uses adaptive algorithm, defaulting to >> ins

Re: yet another snprintf bug

2007-05-01 Thread Bruno Haible
Eric Blake wrote: > For strings larger than 1024 bytes, there was a > separate bug, where the NUL byte would be appended with a buffer overflow Is it possible to detect this with an autoconf test? Or not, because the buffer overflow will always hit memory that was not provided by the caller? Brun

Re: recent argp changes

2007-05-01 Thread Bruno Haible
Hello Sergey, > The option sorting in argp-help was based on the stability > of sort algorithm, therefore the test failed on systems with > quicksort qsort implementations. It went unnoticed on glibc, > because its qsort uses adaptive algorithm, defaulting to > insertion sort if the number of par

drop sigprocmask.h, use signal.h instead

2007-05-01 Thread Bruno Haible
This replaces the sigprocmask.h (gnulib specific include file) with , using the well-known idioms. 2007-05-01 Bruno Haible <[EMAIL PROTECTED]> * lib/sigprocmask.h: Remove file. * lib/signal_.h: Incorporate the previous contents of sigprocmask.h. * lib/sigprocmask.c: Incl

new module 'signal'

2007-05-01 Thread Bruno Haible
This module provides a replacement header for when needed. 2007-05-01 Bruno Haible <[EMAIL PROTECTED]> * modules/signal: New file. * lib/signal_.h: New file. * m4/signal_h.m4: New file. = modules/signal = Desc

make wctype.h self-contained

2007-05-01 Thread Bruno Haible
wctype.h too can be made more self-contained. 2007-05-01 Bruno Haible <[EMAIL PROTECTED]> * lib/wctype_.h: Test HAVE_ISWCNTRL at configure time. * m4/wctype.m4 (gl_WCTYPE_H): Substitute HAVE_ISWCNTRL. * modules/wctype (Makefile.am): Substitute HAVE_ISWCNTRL instead of

make sys/stat.h self-contained

2007-05-01 Thread Bruno Haible
This makes the generated sys/stat.h more self-contained (no need to include first, except for typedefs). 2007-05-01 Bruno Haible <[EMAIL PROTECTED]> * lib/sys_stat_.h: Test HAVE_LSTAT, HAVE_DECL_MKDIR, HAVE_IO_H at configure time. * m4/sys_stat_h.m4 (gl_HEADER_SYS_STAT_

make glob.h self-contained

2007-05-01 Thread Bruno Haible
The generated glob.h should be as self-contained as possible. 2007-05-01 Bruno Haible <[EMAIL PROTECTED]> * lib/glob_.h: Test HAVE_SYS_CDEFS_H at configure time. * m4/glob.m4 (gl_PREREQ_GLOB): Substitute HAVE_SYS_CDEFS_H via AC_SUBST. * modules/glob (Makefile.am): Put HA

Re: doc: new chapter about header files

2007-05-01 Thread Bruno Haible
Eric Blake wrote: > 2007-04-27 Eric Blake <[EMAIL PROTECTED]> > > * doc/headers/assert.texi (assert.h): Document assert module use. After adding per-function documentation, the mention of the 'assert' module fits better in the per-function documentation rather than in the per-header docum

mkdir: depend on sys_stat

2007-05-01 Thread Bruno Haible
mkdir has two portability problems: one with the trailing slash, one with the declaration on mingw. The former is solved by the 'mkdir' module, the latter by the 'sys_stat' module. This is confusing. How can anyone remember this? If someone wants a working mkdir() function, one module should be eno

new various-symlists release

2007-05-01 Thread Bruno Haible
A new version of the "show-portability" script & data is at http://www.haible.de/bruno/gnu/various-symlists.tar.gz Changelog: * pips: New directory. PIPS is the POSIX layer for smartphones like Symbian v9.3 or Nokia. It has already been used to port Apache 2. Most notab