Greetings, Host OS: Solaris 9 I am trying to use: autoconf version 2.59 automake version 1.9.2 bash version 2.05.0(1)-release coreutils version 5.0 libtool version 1.5.10 m4 version 1.4 make version
Just to be clear in advance, the following is mostly not intended to be obstinate argument but more an attempt to understand the GNULIB design goals further... libraries haven't they How does the gen
Thanks. I'm using this patch, which looks a little bit more symmetrical. diff -c -3 -r1.16 localcharset.c ** localcharset.c 24 Dec 2003 13:07:08 -0000 1.16 -- localcharset.c 20 Apr 2004 10:56:26 -000
Bruno, When realloc fails in localcharset's get_charset_aliases, it can clobber the sole pointer to a block of malloc'd memory, thus leaking that block. Here's a proposed patch: 2004-04-20 Jim Meyeri
I installed this: 2004-03-30 Paul Eggert <address@hidden> * modules/getdate: Depend on timespec, stdbool, gettime. * lib/getdate.h: Include stdbool.h, and timespec.h instead of the usual <time.h> dan
I checked it for compatibility with gnulib, and it seems OK. coreutils builds and tests with it. I also took the liberty of merging gnulib's changes with yours. gnulib assumes C89 or better (and has
Hi, There were no comments about 'execute', so I committed it. Now comes the module that allows to create subprocesses connected with pipes (in or out or both). Again, it's tested within gettext for
Hi, Continuing in the series of module for creating subprocesses, here comes a very basic module: Executing a program. It's in use within gettext for ca. 2 years. Any objections or comments? Bruno ==
Hi, After wait-process, here is the proposed module for executing subprocesses in a synchronous way. It's a single function, which - unlike system() - doesn't require an intermediate /bin/sh. A minim
Sorry, I must have missed this in August, but I'll reply now. My kneejerk reaction is that this sort of thing may be heading down the wrong path. The GNU coding standards say: As for systems that are
This patch breaks the linkage between glibc and gnulib for getopt.h and getopt1.c (getopt.c was already not linked). I hope to merge this back into libc. 2003-09-09 Paul Eggert <address@hidden> * src
After installing your getloadavg.c gnulib-Emacs merge <http://mail.gnu.org/archive/html/bug-gnulib/2003-06/msg00023.html> I installed the following minor change, to sync one white space change with c
The latest Bison test version has problems with PARAMS mismatch. This motivated me to propose the following patch. gnulib assumes C89 or better now, so it doesn't need PARAMS any more. Index: lib/Cha
2003-06-05 Dave Love <address@hidden> [From Emacs.] * getloadavg.c: Check `__unix' as well as `unix'. Use #ifdef, not F_SETFD, not FD_SETFD. Use HAVE_STRUCT_NLIST_N_UN_N_NAME, not obsolete NLIST_NAME
Usually it is. If not, there should be a good reason and a comment describing the dependency. IMHO, the way to do it is like this: include <stddef.h> include <stdlib.h> The others may be changed even
There must be a reason why the inclusion of stddef.h isn't done whenever we determine that it's available, but I don't know what it is. This will certainly work; AC_HEADER_STDC causes STDC_HEADERS to
I don't know if the idea is to make each header file provided by gnulib self-contained or not, but if it is then dirname.h needs some #include for size_t. I guess the needed header file is <stddef.h>
That's good with me, but what would you think of going further and defining an xexit function that checks the ferror status of stdout, and translates 0/nonzero into EXIT_SUCCESS/EXIT_FAILURE? I adap