* gmessages.c (g_log_default_handler): (Win32:) Don't call
ensure_stdout_valid (which would open an unneeded console window)
if we're calling a logging function. Define ensure_stdout_valid
as an empty macro on Unix.
* gutils.c (g_get_any_init): (Win32:) Trust HOME first if defined,
then try HOMEDRIVE and HOMEPATH.
Mon Feb 1 19:04:28 1999 Owen Taylor <otaylor@redhat.com>
* gmain.c (g_main_iterate): Release the main_loop
lock around calls to prepare() and check() so
that we are not holding the main loop lock
over user code.
1999-01-30 Sebastian Wilhelmi <wilhelmi@ira.uka.de>
* configure.in: Moved the check for MT save function variants
after the determination of the MT cflags and use them there.
1999-01-28 Sebastian Wilhelmi <wilhelmi@ira.uka.de>
* configure.in: Simplified configure.in test for posix threads,
systems, that provide /usr/include/pthread.h should also have the
posix thread library. This is to avoid writing down all the tests
twice. New test for thread system on HPUX 11. Info from Matt
Nottingham <matt@sar.dera.gov.uk>.
* acinclude.m4: Xsed isn't valid here, don't use it
* ltmain.sh
* ltconfig: better file magic regexp for Linux libs
* gmodule/Makefile.am: arg, noinst_LTLIBRARIES doesn't make shared
libs, revert my previous change to this file
* docs/texinfo.tex: add it so automake doesn't whine
-Yosh
Wed Jan 27 01:57:19 1999 Tim Janik <timj@gtk.org>
* configure.in (G_MODULE_HAVE_DLERROR): applied patch from Andrej
Borsenkow <borsenkow.msk@sni.de> to also check for dlsym() in system
libraries and -ldl, to catch systems that provide dlopen() in libc
and dlsym() in libdl. this is at least the case for Reliant UNIX
5.44 (labeling itself as SINIX).
1999-01-26 Sebastian Wilhelmi <wilhelmi@ira.uka.de>
* gdate.c (g_date_set_time): Removed the #warning about MT
unsafety without localtime_r.
* configure.in: Moved it here.
1999-01-25 Sebastian Wilhelmi <wilhelmi@ira.uka.de>
* configure.in: Do not use the thread libs, when searching for
some functions, as this might require glib to always be linked
with the thread libs on some platforms.
* gutils.c (g_get_any_init): Don't set errno to zero and use it
only as the error code, if the function returned a value less
0. It might happen, that the call succeeds, even though the errno
is set during the call (i.e. it first looks for a passwd file,
which is not found). Submitted by Michael Natterer
<mitschel@cs.tu-berlin.de>. BTW: Sorry for all the mess with that
`getpwuid_r' change, but it had to be done once.
Sun Jan 24 10:33:30 1999 Tim Janik <timj@gtk.org>
* gutils.c (g_get_any_init): reverted raja's changes, since they leaked
a struct passwd contents buffer and for the rest mostly substituted a
while loop with a bunch of gotos.
restored the getpwuid() code to what we had after my recent clean ups
("Sat Jan 23 02:14:28 1999 Tim Janik"), module yosh's recent changes.
to feature solaris behaviour of directly returning errno, we don't
modify error anymore if it's > 0 and simply reset errno.
don't reset g_home_dir for !NATIVE_WIN32 && !NATIVE_WIN32.
* gutils.c
(g_get_any_init) [HAVE_GETPWUID_R && HAVE_GETPWUID_R_POSIX]:
Hopefully got the error handling for `getpwuid_r' right.
(g_get_any_init) [HAVE_GETPWUID_R && !HAVE_GETPWUID_R_POSIX]:
Fix typo (change `pw == NULL' to `pw != NULL').
Please let this be right :-/.
Sun Jan 24 00:36:22 EST 1999 Jeff Garzik <jgarzik@pobox.com>
* ghash.c:
- Revert previous "fix" (which really just did things a
different way).
- (g_hash_table_remove): Don't need to support multiple values
for a single key.
* tests/hash-test.c:
Add test where hash function always returns a single value.
Add beginnings of tests for g_hash_table_foreach[_remove] and
g_hash_table_remove.
Sat Jan 23 22:45:59 1999 Jeff Garzik <jgarzik@pobox.com>
* ghash.c (g_hash_table_lookup_node, g_hash_table_lookup,
g_hash_table_insert, g_hash_table_remove,
g_hash_table_lookup_extended):
- Fixed bug that overwrote nodes in hash buckets instead of
adding them to the hash bucket node list.
Hash tables now work as advertised.
(g_hash_table_resize):
- Use g_new0 instead of manual init.
- Space out code a bit for readability.
(g_hash_nodes_destroy):
- Replaced "if (!hash_node) return;" with
"if (hash_node) {do stuff}".
Testing takes up less code space than explicit call to
'return' before end of function. (look at gcc -S)
Updated module header copyright to 1999.
New module macro G_HASH_BUCKET for (table,key)->bucket lookups.
* tests/hash-test.c:
- Add two new tests, one with strings as the keys and values, and
one with ints as the keys and values. Tests indirect (strings)
and direct (ints) hashing.
- Cleanup unused junk left over from testglib.c.
- Converted a g_print call to g_assert_not_reached.
- Updated copyright to 1999.
* testglib.c, tests/string-test.c:
- Init 'tmp_string' var to NULL, silencing uninit-var warning.
Sat Jan 23 02:14:28 1999 Tim Janik <timj@gtk.org>
* gutils.c (g_get_any_init): cleaned up the errno mess for
GETPWUID. we especially don't want to g_error() out here!
the warning for G_THREADS_ENABLED and !HAVE_GETPWUID_R isn't
gcc related.
if !HAVE_PWD_H and !NATIVE_WIN32, g_free the home dir before
resetting it to NULL, why are we doing this anyways?
reordered code a bit so we always provide defaults (except
for g_home_dir).
Thu Jan 21 12:40:11 EST 1999 Jeff Garzik <jgarzik@pobox.com>
* gmodule/gmodule-dl.c (_g_module_build_path):
Add braces to eliminate an ambiguous else warning.
* tests/{Makefile.am, string-test.c, strfunc-test.c}:
Separate string and strfunc tests, working towards goal of
having separate test for each of the GLib modules.
Add a couple GString length tests.
1999-01-21 Sebastian Wilhelmi <wilhelmi@ira.uka.de>
* gutils.c (g_get_any_init): Changed error handling again, as the
error is not always set in errno, it is the return value on some
systems. What an evil world.
Wed Jan 20 22:09:59 EST 1999 Mandrake <mandrake@mandrake.net>
* configure.in / Makefile.am : made some adjustments for automake 1.4
and autoconf 2.13 (per suggestions made by Raja R Harinath
<harinath@cs.umn.edu>)
Wed Jan 20 20:33:14 EST 1999 Mandrake <mandrake@mandrake.net>
* autogen.sh: automake 1.4 and libtool 1.2d notices.
(maybe I'll write an actual test for automake 1.4 later)
* HACKING: new file
* Makefile.am: since we require automake 1.4 now, ditch the build
top-level first and just use SUBDIRS
* gmodule/Makefile.am: use noinst_LTLIBRARIES instead of overriding
the install rule
-Yosh
Wed Jan 20 20:48:58 GMT 1999 Adam D. Moss <adam@gimp.org>
* gutils.c (g_get_any_init): All gtk apps were broken
here on my setup. I changed the error to a warning to at
least get things limping.
1999-01-20 Sebastian Wilhelmi <wilhelmi@ira.uka.de>
* gutils.c (g_get_any_init): Use getpwuid_r with the right
signature, if available.
* configure.in, acconfig.h: Test for existance of getpwuid_r and
its signature.
1999-01-19 Sebastian Wilhelmi <wilhelmi@ira.uka.de>
* configure.in: Moved test for localtime_r and rand_r to a place
after the determination of the G_THREAD_LIBS and use them for
finding those functions, necessary on systems with a different C
library libc_r for threaded progs, like FreeBSD 2.2.x. Info from
Timur I. Bakeyev <timur@comtat.ru>.
Tue Jan 19 00:44:24 1999 Josh MacDonald <jmacd@axis.hip.berkeley.edu>
* gstrfuncs.c (g_strsignal): The "extern char* sys_siglist"
declaration breaks systems with different declarations, like
mine (FreeBSD). So, I added a configuration variable to decide
whether the declaration is neccesary. Change also appears on
line 275 of configure.in.
Sun Jan 17 17:15:59 1999 Tim Janik <timj@gtk.org>
* gtree.c (g_tree_traverse): removed assertment about rtree->root being
NULL, so this function can be applied to unpopulated trees as well, fix
provided by Simon Kagedal <sika8225@csd.uu.se>.
Sun Jan 17 16:14:03 1999 Tim Janik <timj@gtk.org>
* glib.h (g_error): if !defined (G_LOG_DOMAIN) define G_LOG_DOMAIN as
((gchar*) 0) instead of ((void*) 0) to silence egcs C++ mode, reported
by Seth Alves <alves@hungry.com>.
Sun Jan 17 14:13:52 1999 Tim Janik <timj@gtk.org>
* glib.h: added a define G_HOOK_DEFERRED_DESTROY, to substitute a noop
GHookList.hook_destroy function.
* ghook.c (g_hook_destroy_link): don't really call hook_destroy if it
is G_HOOK_DEFERRED_DESTROY. for the case where we invoke hook->destroy()
we now clean up the hook completely afterwards, i.e. data, func and
destroy are immediately set to NULL and hook_free can't play with that
values anymore.
* gmain.c (g_source_add): set hook_destroy to G_HOOK_DEFERRED_DESTROY,
instead of using an ugly _noop() hack, this is to avoid an uneccessary
function invokation. set hook_free to g_source_destroy_func, this way
we always invoke the destroy notifiers for user_data and source_data
after execution of dispatch(). thus, g_source_destroy_func() will always
be called within the main_loop lock (this wasn't really assured
before), and can release and reaquire the look around destroy notifier
invokation.
