platform/upstream/glib.git
23 years ago.cvsignore updates
Sebastian Wilhelmi [Fri, 16 Feb 2001 13:41:27 +0000 (13:41 +0000)]
.cvsignore updates

.cvsignore updates

23 years agoincorporated huge docu patch from Eric Lemings <eric.b.lemings@lmco.com>
Tim Janik [Fri, 16 Feb 2001 07:22:59 +0000 (07:22 +0000)]
incorporated huge docu patch from Eric Lemings <eric.b.lemings@lmco.com>

Fri Feb 16 06:52:20 2001  Tim Janik  <timj@gtk.org>

        * gobject/tmpl/types.sgml: incorporated huge docu patch from Eric
        Lemings <eric.b.lemings@lmco.com> with a bunch of editing on my part.

Fri Feb 16 07:10:44 2001  Tim Janik  <timj@gtk.org>

        * gclosure.c:
        (g_closure_ref):
        (g_closure_sink): make closure sinking explicit.

        * gsignal.c:
        (g_signal_connect_data):
        (g_signal_connect_closure):
        (g_signal_connect_closure_by_id):
        (g_signal_newv): perform explicit closure sinking.

Thu Feb  8 00:31:45 2001  Tim Janik  <timj@gtk.org>

        * gtype.h: added G_TYPE_DEBUG_NONE for/from Eric Lemings ;)

23 years agoReverted the changes necessary to enlarge the system thread for
Sebastian Wilhelmi [Thu, 15 Feb 2001 16:43:02 +0000 (16:43 +0000)]
Reverted the changes necessary to enlarge the system thread for

2001-02-15  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* acconfig.h, configure.in: Reverted the changes necessary to
enlarge the system thread for G_THREAD_USE_PID_SURROGATE.

* gthread.c: Now implement G_THREAD_USE_PID_SURROGATE in gthread.c
instead of gthread/gthread-posix.c. While the latter has the
advantage, that it is conceptually cleaner, it makes
g_thread_self_posix_impl _very_ slow and that hurts
GStaticRecMutex and other things. So the new version is less
clean, but faster.

* gthread-posix.c: Removed the G_THREAD_USE_PID_SURROGATE
implementation, which is now in gthread.c.

23 years agodon't overwrite errors
Havoc Pennington [Tue, 13 Feb 2001 18:20:13 +0000 (18:20 +0000)]
don't overwrite errors

2001-02-09  Havoc Pennington  <hp@redhat.com>

* gconvert.c (g_convert): don't overwrite errors

* gerror.c (g_set_error): improve warning message if an error is
overwritten
(g_propagate_error): ditto

23 years agoAdded functions g_static_rec_mutex_init, g_static_rec_mutex_free,
Sebastian Wilhelmi [Tue, 13 Feb 2001 15:57:44 +0000 (15:57 +0000)]
Added functions g_static_rec_mutex_init, g_static_rec_mutex_free,

2001-02-13  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* gthread.c, gthread.h: Added functions g_static_rec_mutex_init,
g_static_rec_mutex_free, g_static_private_init,
g_static_private_free, g_static_rw_lock_init to allow the usage of
all those types dynamically. (g_static_rw_lock_free already
existed). Aditionally freed static_private indeces are reused
now. Untill now the array would just grow if you would use more
and more static_private. That required adding a slist of all
running threads, which could potentially be of good use later. It
is not exported however. Renamed a LOCK and small indentation
fixes.

* tests/thread-test.c: Test the new static_private index freing
and reusing feature.

23 years agoAdd check for gtk-doc version. GLIB_1_2_9PRE1
Owen Taylor [Mon, 12 Feb 2001 20:25:58 +0000 (20:25 +0000)]
Add check for gtk-doc version.

Mon Feb 12 15:01:09 2001  Owen Taylor  <otaylor@redhat.com>

* configure.in (gtk_doc_min_version): Add check for gtk-doc version.

23 years agoThe ultimate gtk-doc makefile.
Owen Taylor [Mon, 12 Feb 2001 17:43:44 +0000 (17:43 +0000)]
The ultimate gtk-doc makefile.

Mon Feb 12 12:42:45 2001  Owen Taylor  <otaylor@redhat.com>

* {glib,gobject}/Makefile.am: The ultimate gtk-doc makefile.

23 years agogtypes.h Move GLIB_VAR definition from gutils.h to gtypes.h
Tor Lillqvist [Sat, 10 Feb 2001 01:43:25 +0000 (01:43 +0000)]
gtypes.h Move GLIB_VAR definition from gutils.h to gtypes.h

2001-02-10  Tor Lillqvist  <tml@iki.fi>

* gtypes.h
* gutils.h: Move GLIB_VAR definition from gutils.h to gtypes.h

* glib.def
* gmem.h: Mark glib_mem_profiler_table for export.

23 years agoDon't store address of local variable in hashtable.
Tor Lillqvist [Sat, 10 Feb 2001 00:17:06 +0000 (00:17 +0000)]
Don't store address of local variable in hashtable.

2001-02-10  Tor Lillqvist  <tml@iki.fi>

* gwin32.c (get_package_directory_from_module): Don't store
address of local variable in hashtable.

23 years agoAdded tr to ALL_LINGUAS.
Fatih Demir [Thu, 8 Feb 2001 09:43:21 +0000 (09:43 +0000)]
Added tr to ALL_LINGUAS.

23 years agoADded the Turkish translation fiels for glib HEAD
Fatih Demir [Thu, 8 Feb 2001 09:40:44 +0000 (09:40 +0000)]
ADded the Turkish translation fiels for glib HEAD

23 years ago.cvsignore updates
Sebastian Wilhelmi [Tue, 6 Feb 2001 09:24:20 +0000 (09:24 +0000)]
.cvsignore updates

.cvsignore updates

23 years agoRemove glib_debug_objects.
Tor Lillqvist [Sun, 4 Feb 2001 19:39:17 +0000 (19:39 +0000)]
Remove glib_debug_objects.

2001-02-04  Tor Lillqvist  <tml@iki.fi>

* gobject.def: Remove glib_debug_objects.

23 years agoUse "nn" for Nynorsk, as nn.po files are appearing now. If the sublanguage
Tor Lillqvist [Sun, 4 Feb 2001 19:35:04 +0000 (19:35 +0000)]
Use "nn" for Nynorsk, as nn.po files are appearing now. If the sublanguage

2001-02-04  Tor Lillqvist  <tml@iki.fi>

* gwin32.c (g_win32_getlocale): Use "nn" for Nynorsk, as nn.po
files are appearing now. If the sublanguage starts with '@', don't
use a '_' separator. South Africa is 'ZA', not 'SA'.

23 years agos/glib_trap_/g_trap_/. add comments on g_trap_instance_signals,
Tim Janik [Sun, 4 Feb 2001 07:03:52 +0000 (07:03 +0000)]
s/glib_trap_/g_trap_/. add comments on g_trap_instance_signals,

Sun Feb  4 07:38:32 2001  Tim Janik  <timj@gtk.org>

        * docs/debugging.txt: s/glib_trap_/g_trap_/. add comments on
        g_trap_instance_signals, g_trace_instance_signals and
        GRUNTIME_DEBUG.

        * gmem.c: s/glib_trap_/g_trap_/.

Sun Feb  4 07:30:53 2001  Tim Janik  <timj@gtk.org>

        * gtype.[hc]: changed g_type_init() to take debugging flags
        initially, a combination of G_TYPE_DEBUG_OBJECTS and
        G_TYPE_DEBUG_SIGNALS. using the G_TYPE_ prefix is a bit odd
        here, but basically g_type_int() serves as initialization
        fucntion for all of GType, GObject, GSignal, so what the heck.

        * gobject.c: special case debugging code properly.
        changed glib_trap_object_ref to g_trap_object_ref.

        * gsignal.c: add signal emission debugging abilities, along with
        a new trap object g_trap_instance_signals.

23 years agoChange order of libs, put progs_LDADD last. Needed for cygwin, says
Tor Lillqvist [Sat, 3 Feb 2001 22:32:10 +0000 (22:32 +0000)]
Change order of libs, put progs_LDADD last. Needed for cygwin, says

2001-02-04  Tor Lillqvist  <tml@iki.fi>

* tests/Makefile.am (thread_LDADD): Change order of libs,
put progs_LDADD last. Needed for cygwin, says jbdoll@kepri.re.kr.

2001-02-04  Tor Lillqvist  <tml@iki.fi>

* Makefile.am (progs_LDADD): Change order of libs to libgobject
first, then libglib. Needed for cygwin, says jbdoll@kepri.re.kr.

23 years agogiochannel.h giowin32.c Stylistic cleanups. Use G_STRLOC in g_warning()
Tor Lillqvist [Thu, 1 Feb 2001 22:28:48 +0000 (22:28 +0000)]
giochannel.h giowin32.c Stylistic cleanups. Use G_STRLOC in g_warning()

2001-02-02  Tor Lillqvist  <tml@iki.fi>

* giochannel.h
* giowin32.c
* gmain.c: Stylistic cleanups. Use G_STRLOC in g_warning() calls.

* glib.def: Add missing functions.

2001-01-31  Tor Lillqvist  <tml@iki.fi>

* glibconfig.h.win32.in: Corresponding change as below to
GStaticMutex.

23 years agoAdded g_static_mutex_init to allow initialization of a GStaticMutex, that
Sebastian Wilhelmi [Thu, 1 Feb 2001 09:37:28 +0000 (09:37 +0000)]
Added g_static_mutex_init to allow initialization of a GStaticMutex, that

2001-02-01  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* gthread.c, gthread.h: Added g_static_mutex_init to allow
initialization of a GStaticMutex, that can not be initialized with
G_STATIC_MUTEX_INIT, for example in allocated structures.

* glib/tmpl/threads.sgml, glib/glib-sections.txt: Added
documentation for g_static_mutex_init().

WCVS: ----------------------------------------------------------------------

23 years agoRemove useless check. (Mike Chen)
Owen Taylor [Wed, 31 Jan 2001 21:06:27 +0000 (21:06 +0000)]
Remove useless check. (Mike Chen)

Wed Jan 31 16:05:44 2001  Owen Taylor  <otaylor@redhat.com>

* gtree.c (g_tree_node_search): Remove useless check.
(Mike Chen)

23 years agoFix problem with --disable-nls.
Owen Taylor [Wed, 31 Jan 2001 19:18:48 +0000 (19:18 +0000)]
Fix problem with --disable-nls.

Wed Jan 31 13:46:58 2001  Owen Taylor  <otaylor@redhat.com>

* acinclude.m4 glib-gettext.m4: Fix problem with --disable-nls.

23 years agoadapt to work with new CVS gtk-doc, leaving the old rules in place caused
Tim Janik [Wed, 31 Jan 2001 06:27:41 +0000 (06:27 +0000)]
adapt to work with new CVS gtk-doc, leaving the old rules in place caused

Wed Jan 31 07:14:22 2001  Tim Janik  <timj@gtk.org>

        * gobject/Makefile.am: adapt to work with new CVS gtk-doc, leaving the old
        rules in place caused bogus recursions. main changes:
        - add to conditionalized section:
          all-local:
                $(MAKE) scan
                $(MAKE) templates
                $(MAKE) sgml
                $(MAKE) html.stamp
          html.stamp: sgml.stamp $(EXTRA_SGML_FILES)
                $(MAKE) html
          DOC_STAMPS= html.stamp sgml.stamp
        - change:
          maintainer-clean-local: clean
          -     cd $(srcdir) && rm -rf sgml html $(DOC_MODULE)-decl-list.txt $(DOC_MODULE)-decl.txt
          +     cd $(srcdir) && rm -rf sgml html $(DOC_MODULE)-decl-list.txt $(DOC_MODULE)-decl.txt $(DOC_STAMPS)

        * glib/Makefile.am (maintainer-clean-local): dito.

Wed Jan 31 06:21:32 2001  Tim Janik  <timj@gtk.org>

        * gobject/tmpl/types.sgml: applied docu patch from Eric Lemings
        <eric.b.lemings@lmco.com>, did some more editing.

Wed Jan 31 06:19:49 2001  Tim Janik  <timj@gtk.org>

        * gparam.h: gtk-doc sucks for not dealing with #define inside enums.

        * gtype.[hc]: added G_TYPE_FLAG_RESERVED_ID_BIT, a bit in the type
        number that's supposed to be left untouched (preserved mainly
        for the signal code).

        * *.c: added thread safety code, based on an old patch from sebastian.
        the remaining thread safety issues are now datalists on pspecs (to be
        solved im gdataset.c) and gvalue.c where locking concerns value exchange
        functionality only, and that's soon to be revised.

23 years agoCall g_thread_impl_init(), as g_thread_init won't call it.
Sebastian Wilhelmi [Tue, 30 Jan 2001 14:16:56 +0000 (14:16 +0000)]
Call g_thread_impl_init(), as g_thread_init won't call it.

2001-01-30  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* gthread/gthread-impl.c (g_thread_init_with_errorcheck_mutexes): Call
g_thread_impl_init(), as g_thread_init won't call it.

* glib/glib-sections.txt, glib/tmpl/threads.sgml: Added
g_static_mutex_free

* gthread.c, gthread.h: Added g_static_mutex_free to allow using
GStaticMutexes with limited lifetime without leaking.

* configure.in: GStaticMutex doesn't have to provide extra space
for debugging information for G_ERRORCHECK_MUTEXES, as then the
non-default implementation (runtime_mutex) is used anyway.

23 years agoAdded debug information for g_mutex_free calls, if G_ERRORCHECK_MUTEXES is
Sebastian Wilhelmi [Tue, 30 Jan 2001 09:03:22 +0000 (09:03 +0000)]
Added debug information for g_mutex_free calls, if G_ERRORCHECK_MUTEXES is

2001-01-30  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* gthread.h (g_cond_wait): Added debug information for
g_mutex_free calls, if G_ERRORCHECK_MUTEXES is defined.

* gthread/gthread-impl.c (g_mutex_free_errorcheck_impl): Fixed it for
real. Sorry for this mess. It looked like a real obvious fix, so I
didn't check. Bad boy. Added some casts to quiet the compiler.

23 years agoUse the same GMutex structure as the configure-generated glibconfig.h
Tor Lillqvist [Mon, 29 Jan 2001 21:33:37 +0000 (21:33 +0000)]
Use the same GMutex structure as the configure-generated glibconfig.h

2001-01-29  Tor Lillqvist  <tml@iki.fi>

* glibconfig.h.win32.in: Use the same GMutex structure as the
configure-generated glibconfig.h does.

* gstrfuncs.c (g_strsignal): Declare strsignal() on Cygwin, too,
says jbdoll@kepri.re.kr.

23 years agohack this so it compiles, needs fixing for real.
Havoc Pennington [Mon, 29 Jan 2001 17:31:40 +0000 (17:31 +0000)]
hack this so it compiles, needs fixing for real.

2001-01-29  Havoc Pennington  <hp@redhat.com>

* gthread-impl.c (g_mutex_free_errorcheck_impl): hack this so it
compiles, needs fixing for real.

23 years agoBroadcast the condition, if there are waiting readers, as all might read
Sebastian Wilhelmi [Mon, 29 Jan 2001 16:32:26 +0000 (16:32 +0000)]
Broadcast the condition, if there are waiting readers, as all might read

2001-01-29  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* gthread.c: Broadcast the condition, if there are waiting
readers, as all might read at the same time. Only signal the
writer thread, if there are no more readers.

* gthread/gthread-impl.c (g_mutex_free_errorcheck_impl): Add
        new check to errorcheck mutexes to abort, if a locked mutex is freed.

23 years agoNeed <io.h> for open() and close().
Tor Lillqvist [Sat, 27 Jan 2001 15:07:03 +0000 (15:07 +0000)]
Need <io.h> for open() and close().

2001-01-27  Tor Lillqvist  <tml@iki.fi>

* gmodule.c: (Win32) Need <io.h> for open() and close().

2001-01-27  Tor Lillqvist  <tml@iki.fi>

* makefile.msc.in: Don't try to compile gmarshal.c on its own.

23 years agomakefile.mingw.in Protect the rule to rebuild makefile.mingw if
Tor Lillqvist [Thu, 25 Jan 2001 22:08:11 +0000 (22:08 +0000)]
makefile.mingw.in Protect the rule to rebuild makefile.mingw if

2001-01-25  Tor Lillqvist  <tml@iki.fi>

* makefile.mingw.in
* */makefile.mingw.in: Protect the rule to rebuild makefile.mingw
if makefile.mingw.in has changed with a check if said .in file
exists. (This rule is mainly a convenience for yours truly.)

23 years agoTypo.
Tor Lillqvist [Thu, 25 Jan 2001 21:19:45 +0000 (21:19 +0000)]
Typo.

23 years agoSocket support rewritten. It was utterly broken, and untested in fact. We
Tor Lillqvist [Thu, 25 Jan 2001 21:16:46 +0000 (21:16 +0000)]
Socket support rewritten. It was utterly broken, and untested in fact. We

2001-01-25  Tor Lillqvist  <tml@iki.fi>

* giowin32.c: Socket support rewritten. It was utterly broken, and
untested in fact. We still do use a thread for each socket being
watched, but instead of blocking in recv() (which of course was
plain stupid for sockets being liste()ed on), we block in
select(). The read method for sockets calls recv(). It is now
possible for the application to call accept(), recv() or send() in
the callback, just like on Unix. Tested with code kindly provided
by Andrew Lanoix.

Rename g_io_channel_win32_new_stream_socket() to
g_io_channel_win32_new_socket() as it isn't restricted to stream
sockets.

* gmain.c (g_poll): Related changes in the Win32 version of
g_poll(). When polling for messages, always do a PeekMessage()
first. We used to miss messages if several were posted between
calls to g_poll().

* giochannel.h: Improve Win32-related comments.

* gutf8.c: (Win32) Include <stdio.h> for sprintf.

* tests/gio-test.c: (Win32) Add tests for polling for Windows
messages.

* tests/makefile.mingw.in: Remove superfluous compilation command
line.

23 years agoDon't leak closure->notifiers.
4 [Wed, 24 Jan 2001 07:57:39 +0000 (07:57 +0000)]
Don't leak closure->notifiers.

2001-01-24    <alexl@redhat.com>

* gclosure.c (g_closure_unref):
Don't leak closure->notifiers.

23 years agoFree the callback (g_source_set_callback): Initialize the callback
3 [Tue, 23 Jan 2001 07:35:52 +0000 (07:35 +0000)]
Free the callback (g_source_set_callback): Initialize the callback

2001-01-23    <alexl@redhat.com>

* gmain.c (g_source_callback_unref): Free the callback
(g_source_set_callback): Initialize the callback refcount

23 years agoAdd Ukrainian translation
Yuri Syrota [Mon, 22 Jan 2001 07:39:20 +0000 (07:39 +0000)]
Add Ukrainian translation

23 years agoEven if g_get_codeset() is currently commented out from gutils.h, fix it
Tor Lillqvist [Sat, 20 Jan 2001 22:26:32 +0000 (22:26 +0000)]
Even if g_get_codeset() is currently commented out from gutils.h, fix it

2001-01-20  Tor Lillqvist  <tml@iki.fi>

* gutils.c (g_get_codeset): (Win32) Even if g_get_codeset() is
currently commented out from gutils.h, fix it to return the same
CP%d value as g_get_charset().

23 years agoFix typo. Same here.
Kjartan Maraas [Fri, 19 Jan 2001 08:32:46 +0000 (08:32 +0000)]
Fix typo. Same here.

2001-01-19  Kjartan Maraas  <kmaraas@gnome.org>

* gconvert.c: Fix typo.
* gfileutils.c: Same here.

23 years agoAdded Norwegian translation.
Kjartan Maraas [Fri, 19 Jan 2001 08:31:18 +0000 (08:31 +0000)]
Added Norwegian translation.

2001-01-18  Kjartan Maraas  <kmaraas@gnome.org>

* no.po: Added Norwegian translation.

23 years agoAdd doc comments. (g_win32_get_package_installation_directory): Add one
Tor Lillqvist [Wed, 17 Jan 2001 21:37:32 +0000 (21:37 +0000)]
Add doc comments. (g_win32_get_package_installation_directory): Add one

2001-01-17  Tor Lillqvist  <tml@iki.fi>

* gwin32.c (g_win32_getlocale, g_win32_error_message): Add doc
comments.
(g_win32_get_package_installation_directory): Add one parameter,
the name of a DLL in the package. Add possibility to use that to
deduce the installation directory if not entered into the
Registry. Make the return value dynamically allocated.
(g_win32_get_package_installation_subdirectory): New convenience
function.

* gutils.c (GLIB_LOCALE_DIR)
* gwin32.h: Adapt accordingly.

* testglib.c (main): Adapt tests accordingly.

23 years agoOptimize for size rather than speed by making _() always call
Owen Taylor [Wed, 17 Jan 2001 04:31:20 +0000 (04:31 +0000)]
Optimize for size rather than speed by making _() always call

Tue Jan 16 23:20:38 2001  Owen Taylor  <otaylor@redhat.com>

* gutils.c glibintl.h: Optimize for size rather than speed by
making _() always call _glib_gettext() instead of conditionally
calling gettext() or _glib_gettext_init. glib only uses translated
strings in slow error handling code anyways.

* glibintl.h: g'ify types.

* Makefile.am: include glibintl.h in _SOURCES

23 years agoDefine ENABLE_NLS and GETTEXT_PACKAGE.
Tor Lillqvist [Tue, 16 Jan 2001 22:06:23 +0000 (22:06 +0000)]
Define ENABLE_NLS and GETTEXT_PACKAGE.

2001-01-17  Tor Lillqvist  <tml@iki.fi>

* config.h.win32.in: Define ENABLE_NLS and GETTEXT_PACKAGE.

* makefile.mingw.in
* makefile.msc.in: Use the GNU intl library.

* gwin32.c (g_win32_get_package_installation_directory): New
function. To be used by various GLib-using packages to get their
installation directory, which should be stored in the Registry by
some installer.

* gwin32.h: Declare it.

* testglib.c (main): Test it.

* gutils.c: On Win32, define GLIB_LOCALE_DIR using
g_win32_get_package_installation_directory().

* glib.def: Update.

23 years agoAdd @INTLIBS@.
Owen Taylor [Tue, 16 Jan 2001 02:52:13 +0000 (02:52 +0000)]
Add @INTLIBS@.

Mon Jan 15 21:39:06 2001  Owen Taylor  <otaylor@redhat.com>

* glib-2.0.pc.in (Libs) glib-config-2.0.in: Add @INTLIBS@.

23 years agoAdd gettext support.
Owen Taylor [Tue, 16 Jan 2001 02:24:24 +0000 (02:24 +0000)]
Add gettext support.

Mon Jan 15 21:12:49 2001  Owen Taylor  <otaylor@redhat.com>

        * configure.in acconfig.h glibintl.h gutils.c
po/{Makefile.in.in,POTFILES.in,po2tbl.in}: Add gettext
support.

* glib-gettext.m4 acinclude.m4: Clean up the GTK+ gettext macros
some more and put them in this file, though they also need
to be included in acinclude.m4 due to the brokeness of
aclocal.

* gspawn.c gspawn-win32.c gutf8.c gconvert.c gfileutils.c
gshell.c: Remove dummy _() #defines, include glibintl.m4.

23 years agoFixed my debbugs-test-submit@bugzilla.gnome.org mess.
Martin Baulig [Thu, 11 Jan 2001 21:24:29 +0000 (21:24 +0000)]
Fixed my debbugs-test-submit@bugzilla.gnome.org mess.

23 years agoAdded debbugs-test-submit@bugzilla.gnome.org everywhere.
Martin Baulig [Thu, 11 Jan 2001 10:49:19 +0000 (10:49 +0000)]
Added debbugs-test-submit@bugzilla.gnome.org everywhere.

23 years agoMinor cosmetic change to previous ChangeLog entry.
Tor Lillqvist [Tue, 9 Jan 2001 20:48:20 +0000 (20:48 +0000)]
Minor cosmetic change to previous ChangeLog entry.

23 years agoRework the changes to needed to pass mainloop-test. Now we don't need to
Tor Lillqvist [Tue, 9 Jan 2001 20:46:49 +0000 (20:46 +0000)]
Rework the changes to needed to pass mainloop-test. Now we don't need to

2001-01-09  Tor Lillqvist  <tml@iki.fi>

* giowin32.c (create_reader_thread): Rework the changes to
needed to pass mainloop-test. Now we don't need to call
TerminateThread() after all, which is a relief, as the docs
have a BIG RED WARNING SIGN about using that API. Instead,
when closing a fd channel that has a reader thread running,
just mark it as non-running and additionally mark the fd as
ripe for closing. When the reader thread hopefully eventually
gets something (and EOF or some actual data), it will note
that it shouldn't be running, break out of the loop, and close
the fd.

The socket channel closing code should probably be changed
similarily, but that will have to wait until I have a test case.

(g_pipe_readable_msg, g_io_channel_win32_new_pipe,
g_io_channel_win32_new_pipe_with_wakeups,
g_io_channel_win32_pipe_request_wakeups,
g_io_channel_win32_pipe_readable): Remove these, have been
obsolete for some time.

* glib.def: Update.

23 years agoDon't warn about deprecation on Win32. Code written for GLib 1.2 doesn't
Tor Lillqvist [Tue, 9 Jan 2001 16:33:21 +0000 (16:33 +0000)]
Don't warn about deprecation on Win32. Code written for GLib 1.2 doesn't

2001-01-09  Tor Lillqvist  <tml@iki.fi>

* gutils.c (g_basename, g_dirname): Don't warn about deprecation
on Win32. Code written for GLib 1.2 doesn't have much choice but
to use GLib >= 1.3 on Win32.

* glib.def: Update.

23 years agominor markup cleanups
Tim Janik [Tue, 9 Jan 2001 15:29:15 +0000 (15:29 +0000)]
minor markup cleanups

23 years agoMade g_profile_mutex a GMutex* instead of G_LOCK_DEFINE_STATIC to avoid
Sebastian Wilhelmi [Tue, 9 Jan 2001 13:47:12 +0000 (13:47 +0000)]
Made g_profile_mutex a GMutex* instead of G_LOCK_DEFINE_STATIC to avoid

2001-01-09  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* gmem.c: Made g_profile_mutex a GMutex* instead of
G_LOCK_DEFINE_STATIC to avoid deadlock for thread implementations
without native static mutexes. Contruct g_profile_mutex in
g_mem_init().

23 years agoadded some function documentations.
Tim Janik [Tue, 9 Jan 2001 02:18:57 +0000 (02:18 +0000)]
added some function documentations.

Tue Jan  9 03:10:38 2001  Tim Janik  <timj@gtk.org>

        * gobject/tmpl/types.sgml: added some function documentations.

        * gobject/gobject-sections.txt: buncha fixups.

23 years agoGet len using strlen() if arg is negative in the Win32 code, too.
Tor Lillqvist [Sat, 6 Jan 2001 03:09:46 +0000 (03:09 +0000)]
Get len using strlen() if arg is negative in the Win32 code, too.

2001-01-06  Tor Lillqvist  <tml@iki.fi>

* gconvert.c (g_locale_to_utf8, g_locale_from_utf8): Get len using
strlen() if arg is negative in the Win32 code, too.

* giowin32.c: Changes necessary to be able to run
mainloop-test. We can't close the fd that our (internal) reader
thread is sitting doing a blocking read() from. We must terminate
the thread first. Keep track of thread handle, and close it when
thread is dying. Start reader thread with the lower-level
CreateThread() instead of _beginthreadex() from the C runtime, in
order to be able to use TerminateThread(). Hopefuly this isn't
harmful.

* glib.def: Update.

* tests/makefile.{mingw,msc}.in (TESTS): Add mainloop-test and
unicode-encoding.

* tests/mainloop-test.c: Portability: <unistd.h>, need <fcntl.h>
on Win32.

* tests/unicode-encoding.c (process): Add missing "line" argument
to fail(). On Win32, convert UTF-16LE, as libiconv'c ivonf always
converts to UTF-16BE if we ask for unspecific UTF-16.
(main) Handle also '\r'.

23 years agomove $enable_debug down below checks for GCC to avoid setting CFLAGS
Owen Taylor [Fri, 5 Jan 2001 21:22:47 +0000 (21:22 +0000)]
move $enable_debug down below checks for GCC to avoid setting CFLAGS

Fri Jan  5 11:25:42 2001  Owen Taylor  <otaylor@redhat.com>

* configure.in (PACKAGE): move $enable_debug down below
checks for GCC to avoid setting CFLAGS prematurely,
change checks to avoid adding -g twice.

* gutf8.c (g_ucs4_to_utf8): Support len < 0 to mean
0 termination.

* gutf8.c (g_utf8_to_ucs4): Terminate result with 0.

* tests/mainloop-test.c (main): Fix uses of
g_main_loop_destroy().

* tests/unicode-encoding.c tests/Makefile.am tests/utf8.txt:
Tests for unicode-conversion code.

* gconvert.c (g_convert, g_convert_with_fallback): work around
a couple of GNU libc bugs.

* gconvert.[ch] (g_{locale,filename}_{to,from}_utf8): Standardize
arguments to match g_convert(). Document.

* gunicode.[ch]:
  - Implement conversion functions to and from UTF-16
  - Standardize unicode conversion functions on prototype like
    g_convert.
  - Add a lot of error checking to unicode conversion functions.

* gunicode.[ch] (g_utf8_to_ucs4_fast): Add fast, non-checking
variant of g_utf8_to_ucs4.

* gutf8.c (g_utf8_validate):
 - add g_return_if_fail (str != NULL).
 - add checks for overlong strings, non-valid Unicode characters (>= 110000)
   and single surrogates.

23 years agoUpdated.
Owen Taylor [Fri, 5 Jan 2001 20:48:43 +0000 (20:48 +0000)]
Updated.

Fri Jan  5 15:47:10 2001  Owen Taylor  <otaylor@redhat.com>

* glib/glib-sections.txt glib/tmpl/* glib/Makefile.am: Updated.

23 years agoset the value_type in the param spec; closes bug 40210
Havoc Pennington [Fri, 5 Jan 2001 20:26:29 +0000 (20:26 +0000)]
set the value_type in the param spec; closes bug 40210

2001-01-05  Havoc Pennington  <hp@redhat.com>

* gparamspecs.c (g_param_spec_enum): set the value_type in the
param spec; closes bug 40210
(g_param_spec_flags): ditto for flags

23 years agoAdd test for g_path_skip_root().
Tor Lillqvist [Fri, 5 Jan 2001 20:19:37 +0000 (20:19 +0000)]
Add test for g_path_skip_root().

2001-01-05  Tor Lillqvist  <tml@iki.fi>

* testglib.c (main): Add test for g_path_skip_root().

* gutils.c (g_path_skip_root): Fix bugs in code I just added ;-)

23 years agoLook also for (illegal) forward slashes in the template.
Tor Lillqvist [Fri, 5 Jan 2001 19:30:13 +0000 (19:30 +0000)]
Look also for (illegal) forward slashes in the template.

2001-01-05  Tor Lillqvist  <tml@iki.fi>

* gfileutils.c (g_file_open_tmp): (Win32:) Look also for (illegal)
forward slashes in the template.

* gutils.c (g_path_skip_root): On Win32, skip the \\server\share
part of UNC paths. On all platforms, skip several initial
slashes. Add a few comments.
(g_get_any_init): On Win32, in case HOME is Unix-style with
(forward) slashes (some other applications apparently set it up
this way, convert to backslashed form.

* configure.in (glib_os): Remove stray 'v'. Add case for mingw,
although using configure for mingw surely doesn't work yet.

* glib.def: Update.

23 years agoremove glib-config-2.0
Havoc Pennington [Thu, 4 Jan 2001 17:47:00 +0000 (17:47 +0000)]
remove glib-config-2.0

2001-01-02  Havoc Pennington  <hp@redhat.com>

* configure.in: remove glib-config-2.0

* Makefile.am: remove glib-config-2.0

23 years ago-I$(top_srcdir)/gmodule, fix from Michael Meeks
Havoc Pennington [Thu, 4 Jan 2001 03:49:45 +0000 (03:49 +0000)]
-I$(top_srcdir)/gmodule, fix from Michael Meeks

2001-01-03  Havoc Pennington  <hp@pobox.com>

* tests/Makefile.am (INCLUDES): -I$(top_srcdir)/gmodule,
fix from Michael Meeks

* Makefile.am (INCLUDES): DISABLE was spelled wrong

23 years agoSwitch GMainLoop to be ref/unref, use to make dropping reference to
Owen Taylor [Wed, 3 Jan 2001 20:18:40 +0000 (20:18 +0000)]
Switch GMainLoop to be ref/unref, use to make dropping reference to

Wed Jan  3 14:10:49 2001  Owen Taylor  <otaylor@redhat.com>

* gmain.[ch]: Switch GMainLoop to be ref/unref, use to
make dropping reference to running loop safe.

23 years agoMade g_thread_min_stack_size static.
Sebastian Wilhelmi [Wed, 3 Jan 2001 17:00:43 +0000 (17:00 +0000)]
Made g_thread_min_stack_size static.

2001-01-03  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

        * gthread-solaris.c, gthread-posix.c: Made g_thread_min_stack_size
        static.

2CVS: ----------------------------------------------------------------------

23 years agoUnref callback->cb_data if it was still set when the source is freed.
Owen Taylor [Wed, 3 Jan 2001 16:05:39 +0000 (16:05 +0000)]
Unref callback->cb_data if it was still set when the source is freed.

Wed Dec 13 20:41:49 2000  Owen Taylor  <otaylor@redhat.com>

* gmain.c (g_source_unref_internal): Unref callback->cb_data
if it was still set when the source is freed. (Usually, this
will be done by g_source_destroy.)

23 years agoMove glib_debug_objects out of the G_ENABLE_DEBUG #ifdef.
Alexander Larsson [Wed, 3 Jan 2001 10:03:49 +0000 (10:03 +0000)]
Move glib_debug_objects out of the G_ENABLE_DEBUG #ifdef.

2001-01-03  Alexander Larsson  <alexl@redhat.com>

* gobject.c:
Move glib_debug_objects out of the G_ENABLE_DEBUG #ifdef.

23 years agoUse parentheses around an argument to make these cause an error if you
Dan Winship [Tue, 2 Jan 2001 20:52:05 +0000 (20:52 +0000)]
Use parentheses around an argument to make these cause an error if you

* garray.h (g_array_append_val, g_array_prepend_val,
g_array_insert_val): Use parentheses around an argument to make
these cause an error if you pass a non-lvalue for the value,
rather than silently doing the wrong thing.

23 years agoAdd GLIB_SIZEOF_VOID_P and GLIB_SIZEOF_LONG.
Tor Lillqvist [Fri, 29 Dec 2000 19:25:40 +0000 (19:25 +0000)]
Add GLIB_SIZEOF_VOID_P and GLIB_SIZEOF_LONG.

2000-12-29  Tor Lillqvist  <tml@iki.fi>

* glibconfig.h.win32.in: Add GLIB_SIZEOF_VOID_P and GLIB_SIZEOF_LONG.

* glib.def: Update.

* {.,*}/makefile.{mingw,msc}.in: Add -DG_ENABLE_DEBUG.

23 years agowe can't grow _cv_ variables by using a backticked expr that refers back
Tim Janik [Fri, 29 Dec 2000 15:11:03 +0000 (15:11 +0000)]
we can't grow _cv_ variables by using a backticked expr that refers back

Fri Dec 29 14:53:18 2000  Tim Janik  <timj@gtk.org>

        * configure.in: we can't grow _cv_ variables by using a backticked
        expr that refers back to the variable (glib_cv_sizeof_system_thread for
        G_THREAD_USE_PID_SURROGATE), that'd keep the variable growing every time
        it's evaluated. quantum states, anyone?)

23 years agogot rid of outdated dmalloc support. provide g_try_malloc() and
Tim Janik [Fri, 29 Dec 2000 02:16:51 +0000 (02:16 +0000)]
got rid of outdated dmalloc support. provide g_try_malloc() and

Thu Dec 28 10:21:46 2000  Tim Janik  <timj@gtk.org>

        * gmem.[hc]: got rid of outdated dmalloc support. provide g_try_malloc()
        and g_try_realloc() which _may_ fail and return NULL.
        nuked g_mem_check(), provided GMemVTable for memory function
        virtualization, alterable at program startup with g_mem_set_vtable().
        provided glib_mem_profiler_table and g_mem_profile() to support limited
        profiling information out of the box (uses mprotect() for free()ed areas
        on linux).
        provide globally visible G_MEM_ALIGN.
        buncha cleanups.

        * docs/macros.txt: file to get a clue about the various configuration
        macros.

        * docs/debugging.txt: explain debugging traps.

        * configure.in: got rid of --enable-mem-check and --enable-mem-profile,
        define GLIB_SIZEOF_VOID_P and GLIB_SIZEOF_LONG. check malloc prototypes
        and define SANE_MALLOC_PROTOS is we can use them.
        <boy, is this file a mess>

        * gutils.c, gscanner.c: fix up compatibility warnings, use g_message().

Thu Dec 28 11:36:44 2000  Tim Janik  <timj@gtk.org>

        * gbsearcharray.c (upper_power2): disable G_BSEARCH_ALIGN_POWER2
        fucntionality if DISABLE_MEM_POOLS is defined.

        * gtype.c: honour DISABLE_MEM_POOLS.

        * gsignal.c (g_signal_init): flag signal key bsearch array with
        G_BSEARCH_ALIGN_POWER2 to avoid excessive growth time. honour
        DISABLE_MEM_POOLS.

        * gparam.h: added G_PARAM_READWRITE alias for (G_PARAM_READABLE |
                                                       G_PARAM_WRITABLE).

23 years agoUpdate.
Tor Lillqvist [Wed, 27 Dec 2000 20:56:15 +0000 (20:56 +0000)]
Update.

2000-12-27  Tor Lillqvist  <tml@iki.fi>

* README.win32: Update.

23 years agoUse a MessageBox for fatal messages. Collect eror message into a buffer,
Tor Lillqvist [Mon, 25 Dec 2000 22:07:01 +0000 (22:07 +0000)]
Use a MessageBox for fatal messages. Collect eror message into a buffer,

2000-12-25  Tor Lillqvist  <tml@iki.fi>

* gmessages.c: (Win32) Use a MessageBox for fatal
messages. Collect eror message into a buffer, and display that.

* glib.def: Update.

* glibconfig.h.win32.in: Update. Remove unused wchar and wctype
macros, add G_MODULE_SUFFIX.

23 years agoNew gcompat.h header file as recommended by Havoc.
Ali Abdin [Sun, 24 Dec 2000 12:11:06 +0000 (12:11 +0000)]
New gcompat.h header file as recommended by Havoc.

2000-12-24  Ali Abdin  <aliabdin@aucegypt.edu>

        * Makefile.am, gcompat.h, glib.h: New gcompat.h header file
          as recommended by Havoc.

        * gdate.c, gdate.h, testgdate.c,
          docs/reference/glib/glib-sections.txt,
          docs/reference/glib/tmpl/date.sgml, tests/date-test.c: Rename some
          of the gdate functions to use the '_get' in their name. Patch
          reviewed by Havoc.

23 years agoDetermine the suffix of the shared librarries for this system. This is
Sebastian Wilhelmi [Fri, 22 Dec 2000 13:44:25 +0000 (13:44 +0000)]
Determine the suffix of the shared librarries for this system. This is

2000-12-22  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* configure.in: Determine the suffix of the shared librarries for
this system. This is done analogous to
ltconfig.sh. G_MODULE_SUFFIX in glibconfig.h is set to either
"sl", "dll", or (most often) "so".

* tests/Makefile.am, tests/module-test.c,
tests/libmoduletestplugin_a.c, tests/libmoduletestplugin_b.c:
Added new testcase for gmodule. This is mostly copied from
gmodule/testgmodule.c, but unlike that is is quiet. (Why BTW are
some tests that verbose, not to say loquacious...)

* gmodule.c: Make g_module_open more tolerant wrt to the module
name. First it tries to open the module as named, if that fails,
it checks, whether it is a libtool archive and parses it, if that
fails it appends the systems shared library suffix
(i.e. ".so") (if not already found) and tries again and if that
fails it tries to append the ".la" libtool suffix (if not already
found) and parses it.

* gmodule.c: Lock recursive mutex during most module functions for
safety.

* gmodule-dl.c: Return an error from _g_module_symbol only, if
dlerror says so. All other functions return an error as well, if
dlerror returns NULL.

* testgmodule.c: Thanks to the above change the #ifdefs have
vanished.

* glib/glib-sections.txt: Added G_MODULE_SUFFIX.

* glib/tmpl/modules.sgml: Updated.

23 years agoUpdated G_RAND_DOUBLE_TRANSFORM to be more accurate. Redid g_rand_double()
Sebastian Wilhelmi [Tue, 19 Dec 2000 15:57:53 +0000 (15:57 +0000)]
Updated G_RAND_DOUBLE_TRANSFORM to be more accurate. Redid g_rand_double()

2000-12-19  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* grand.c: Updated G_RAND_DOUBLE_TRANSFORM to be more
accurate. Redid g_rand_double() such that it returns 52 bits after
the point instead of 32 as before. That OTOH requires calling
g_rand_int() twice. Overhauled g_rand_int_range(), which is easier
now thanks to the new precision of g_rand_double(). Thanks to
Sverre Johansen <sj@ifi.uio.no> for the hint.

* grand.h: Added g_rand_boolean() and g_random_boolean()
macros. While they could be omitted due to extreme simplicity,
they make intention clearer in code and are therefore good to have.

* grand.c, grand.h: Renamed all 'min' and 'max' parameters to'
begin' and 'end' resp. to avoid making people think, that 'max' is
included in the interval. 'end' now isn't, whereas 'begin'
is. That's similar to the use in the STL.

* glib/glib-sections.txt: Added g_rand_boolean and
g_random_boolean macros.

* glib/tmpl/random_numbers.sgml: Updated.

23 years agoOk, I'm a moron. When I originally implemented ENABLE_GC_FRIENDLY, I
Sebastian Wilhelmi [Tue, 19 Dec 2000 15:40:30 +0000 (15:40 +0000)]
Ok, I'm a moron. When I originally implemented ENABLE_GC_FRIENDLY, I

2000-12-19  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* gslist.c, glist.c: Ok, I'm a moron. When I originally
implemented ENABLE_GC_FRIENDLY, I forgot to include config.h into
the affected files. Now that Alex did that for those two,
inevitable typos surfaced, which are now fixed.

* garray.c, ghash.c, gqueue.c, gtree.c: Include config.h as well,
as ENABLE_GC_FRIENDLY should be known.

23 years agoAdded --disable-mem-pools option.
Alexander Larsson [Tue, 19 Dec 2000 09:35:44 +0000 (09:35 +0000)]
Added --disable-mem-pools option.

2000-12-19  Alexander Larsson  <alexl@redhat.com>

* configure.in:
Added --disable-mem-pools option.

* glist.c:
* gslist.c:
* gnode.c:
* gmem.c:
Disable free list and memory chunks if DISABLE_MEM_POOLS is defined.

23 years agoUse GetACP to get the current ANSI codepage.
Tor Lillqvist [Sun, 17 Dec 2000 18:43:57 +0000 (18:43 +0000)]
Use GetACP to get the current ANSI codepage.

2000-12-17  Tor Lillqvist  <tml@iki.fi>

* gutf8.c (g_utf8_get_charset_internal): (Win32) Use GetACP to get
the current ANSI codepage.

* gunicode.h: Add comment that the static string g_get_charset
sets the parameter to point to should be copied in case the
charset might be changed later in the program.

23 years agoUpdate.
Tor Lillqvist [Sat, 16 Dec 2000 20:06:59 +0000 (20:06 +0000)]
Update.

2000-12-16  Tor Lillqvist  <tml@iki.fi>

* gobject.def: Update.

23 years agoUpdate.
Tor Lillqvist [Fri, 15 Dec 2000 19:49:35 +0000 (19:49 +0000)]
Update.

2000-12-15  Tor Lillqvist  <tml@iki.fi>

* gobject.def: Update.

23 years agouse g_signal_newc
Havoc Pennington [Fri, 15 Dec 2000 07:05:10 +0000 (07:05 +0000)]
use g_signal_newc

2000-12-15  Havoc Pennington  <hp@pobox.com>

* gobject.c (g_object_do_class_init): use g_signal_newc

* gsignal.c (g_signal_newc): convenience function for signals
created from C
(g_signal_new_valist): added
(g_signal_new): removed

23 years agoadd an instance member value_type so the default value of the pspec class
Tim Janik [Fri, 15 Dec 2000 05:51:08 +0000 (05:51 +0000)]
add an instance member value_type so the default value of the pspec class

Fri Dec 15 04:40:23 2000  Tim Janik  <timj@gtk.org>

        * gparam.[hc]: add an instance member value_type so the default
        value of the pspec class can be overridden.

23 years agoUpdate once again.
Tor Lillqvist [Thu, 14 Dec 2000 22:28:15 +0000 (22:28 +0000)]
Update once again.

2000-12-14  Tor Lillqvist  <tml@iki.fi>

* gobject.def: Update once again.

23 years agoUpdate more.
Tor Lillqvist [Thu, 14 Dec 2000 21:34:07 +0000 (21:34 +0000)]
Update more.

2000-12-14  Tor Lillqvist  <tml@iki.fi>

* gobject.def: Update more.

23 years agoNo need to -DGSPAWN_HELPER when compiling gspawn-win32-helper any longer.
Tor Lillqvist [Thu, 14 Dec 2000 21:08:00 +0000 (21:08 +0000)]
No need to -DGSPAWN_HELPER when compiling gspawn-win32-helper any longer.

2000-12-14  Tor Lillqvist  <tml@iki.fi>

* makefile.{mingw,msc}.in: No need to -DGSPAWN_HELPER when
compiling gspawn-win32-helper any longer.

23 years agoWarn if no callback. Call callback correctly. (g_io_win32_create_watch):
Tor Lillqvist [Thu, 14 Dec 2000 21:02:20 +0000 (21:02 +0000)]
Warn if no callback. Call callback correctly. (g_io_win32_create_watch):

2000-12-14  Tor Lillqvist  <tml@iki.fi>

* giowin32.c (g_io_win32_dispatch): Warn if no callback. Call
callback correctly.
(g_io_win32_create_watch): Fix typo.
(g_io_win32_fd_create_watch): Ditto.
(g_io_channel_unix_new): If it is a file descriptor (i.e., a Unix
fd lookalike provided by the C library), call
g_io_channel_win32_new_fd(). If it is a socket (from WinSock),
call g_io_cahnnel_win32_new_stream_socket(). Hopefully sockets and
fds don't overlap. TODO: Implement also datagram sockets.
(g_io_channel_win32_poll): Call g_main_context_get_poll_func().

* gcompletion.h: Include <unistd.h> only on Unix. Is this
inclusion really needed here? OTOH, do include <stddef.h>, for
size_t.

* gmessages.c: (Win32) Don't define a function called "write" that
might clash with the prototype from <io.h>, use a #define.

* glib.def: Update.

* gmain.c (g_source_add_poll): Don't return a value from void
function.
(g_main_context_get_poll_func): Compile also for non-Win32, as
presumably was intended. The result var is a GPollFunc, not a
GPollFunc*. Return the result!

gobject:

 2000-12-14  Tor Lillqvist  <tml@iki.fi>

* makefile.mingw.in: Update, include parts from Makefile.am to
build gmarshal.[ch]. Some day, we won't need these separate
makefiles for Win32 compilation. I hope.

* makefile.msc.in: Update. No use trying to build gmarshal.[ch]
here, it would require Unixish tools. MSVC users building from CVS
sources are out of luck.

* gobject.def: Update.

23 years agomake static
Havoc Pennington [Wed, 13 Dec 2000 23:54:51 +0000 (23:54 +0000)]
make static

2000-12-13  Havoc Pennington  <hp@redhat.com>

* gconvert.c (open_converter): make static

* gutf8.c (g_utf8_validate): Simplify logic a bit, maybe
speeding it up - now we just return FALSE if we had to bail out
for any reason before getting to the end of the string, as defined
by a nul byte if len was -1, defined by the len otherwise. This
also fixes a bug where nul bytes were not treated as invalid
when the length was specified.

23 years agourg managed to forget this last time ;(
Tim Janik [Wed, 13 Dec 2000 08:36:17 +0000 (08:36 +0000)]
urg managed to forget this last time ;(

Wed Dec 13 09:31:26 2000  Tim Janik  <timj@gtk.org>

        * gparamspecs.[hc]: add G_TYPE_PARAM_BOXED implementation.

        * gobject.[hc]: minor fixes.

23 years agodon't try to use thread stuff unless G_THREADS_ENABLED
Havoc Pennington [Wed, 13 Dec 2000 04:23:45 +0000 (04:23 +0000)]
don't try to use thread stuff unless G_THREADS_ENABLED

2000-12-12  Havoc Pennington  <hp@pobox.com>

* gmain.c (g_main_context_destroy): don't try to use thread stuff
unless G_THREADS_ENABLED
(g_main_context_query): ditto
(g_main_context_check): ditto
(g_main_loop_quit): ditto

23 years agoreturn whether a value got removed.
Tim Janik [Wed, 13 Dec 2000 00:44:18 +0000 (00:44 +0000)]
return whether a value got removed.

Tue Dec 12 18:58:22 2000  Tim Janik  <timj@gtk.org>

        * ghash.c (g_hash_table_remove): return whether a value
        got removed.

Tue Dec 12 23:38:02 2000  Tim Janik  <timj@gtk.org>

        * Makefile.am: _never_ touch oldest-source-stamp.

        * gobject.[hc]: construct property handling fixes/improvements.
        fixed trailer handling in get/set property.

        * gparam.[hc]: implement param spec pool, got rid of param spec
        hashtable. the most prominent change is that e deal with type
        prefixes here.

23 years agoclean up after elliot, wonder what his mother has to do on a daily basis.
Tim Janik [Tue, 12 Dec 2000 22:40:00 +0000 (22:40 +0000)]
clean up after elliot, wonder what his mother has to do on a daily basis.

23 years agoRevert unauthorized changes.
Owen Taylor [Tue, 12 Dec 2000 20:23:37 +0000 (20:23 +0000)]
Revert unauthorized changes.

Tue Dec 12 15:18:10 2000  Owen Taylor  <otaylor@redhat.com>

* gmain.[ch]: Revert unauthorized changes.

23 years agoGMainContext useful in implementing some additional styles of main loop
Elliot Lee [Tue, 12 Dec 2000 19:34:41 +0000 (19:34 +0000)]
GMainContext useful in implementing some additional styles of main loop

* gmain.c, gmain.h (g_main_context_new, g_main_context_destroy): GMainContext useful
in implementing some additional styles of main loop usage.  To do this, however, Joe
Hacker needs to be able to create/destroy GMainContext's at will. This is just an
export of existing functionality, rather than new functionality.

They are listed in the "Low level functions for implementing custom main loops"
section of the header file, to avoid confusing people.

* gobject/Makefile.am:
  . You have to 'touch oldest-source-stamp' if you want to avoid having
  the Makefile constantly rebuild itself.
  . Fix marshaller generation rules to work with srcdir != builddir
    (there were issues with trying to run "./glib-genmarshal", etc.)

23 years agoshut up on new marshaller generated stuff
Tim Janik [Tue, 12 Dec 2000 07:53:04 +0000 (07:53 +0000)]
shut up on new marshaller generated stuff

23 years agoi guess some people may need this ;)
Tim Janik [Tue, 12 Dec 2000 07:51:44 +0000 (07:51 +0000)]
i guess some people may need this ;)

23 years agofixed dealing with collection/lcopy of NULL values.
Tim Janik [Tue, 12 Dec 2000 07:32:00 +0000 (07:32 +0000)]
fixed dealing with collection/lcopy of NULL values.

Mon Dec 11 04:44:11 2000  Tim Janik  <timj@gtk.org>

* gboxed.c: fixed dealing with collection/lcopy of NULL values.

* gclosure.h: removed insane ramblings, added G_CALLBACK() a casting
convenience macro.

* Makefile.am: cleanups, marshaller generation rules.

* gmarshal.[hc]: new files with GRuntime standard marshallers.

* glib-genmarshal.c: fix log domain, support gruntime standard
marshallers, suport G_TYPE_PARAM, come with extern "C" and
#include gmarshal.h.

* glib-genmarshal.1: reflect glib-genmarshal.c updates.

* gobject.[hc]: implement object constructor. rework parameter
changed notification queueing, we support queue freezes now and
don't dispatch from an idle handler anymore.
parameter->property rename hassle.
implemented ::properties_changed and ::notify::* signals for
property change notification (the later supports property names
as details). added signal connection and named data properties.
(g_signal_connect_object): new function to setup while_alive
connections.
(g_object_class_install_property): sink properties now, since they
are initially floating.
(g_object_steal_data):
(g_object_set_data_full):
(g_object_set_data):
(g_object_get_data): set/get data by using g_datalist_*() functions
directly.
(g_object_queue_param_changed): nuked.
(g_object_freeze_notify): start queueing of property changes (freeze/
thaw calls stack).
(g_object_notify): announce changes of a certain property directly.
(g_object_thaw_notify): process queue of property changes, therefore
emitting GObject::notify::detail with detail being the changed
properties names.
(G_OBJECT_WARN_INVALID_PROPERTY_ID): saner macro variant of former
G_WARN_INVALID_PARAM_ID().

* gparam.[hc]: param specs are now initially floating and need to be
sunken with g_param_spec_sink(), support G_TYPE_PARAM values.
added G_PARAM_CONSTRUCT and G_PARAM_CONSTRUCT_ONLY parameter flags,
required by GObjectClass.constructor().

* gparamspecs.[hc]: added GParamSpecParam, GParamSpecPointer and
GParamSpecCCallback, param specs for G_TYPE_PARAM, G_TYPE_POINTER
and G_TYPE_CCALLBACK respectively.

* gsignal.[hc]: cleanups.
(signal_id_lookup): after walking the anchestry, try interfaces as well.
(g_signal_new): new function to create signals from varargs type list.
(g_signal_connect_closure): closure connection variant that works from
signal name+detail.
(g_signal_connect_data): c handler connection variant that works from
signal name+detail.
(g_signal_emit_valist): emit signal for an instance with paraneters
collected from a va_list.
(g_signal_emit): emit signal, taking parameters from varargs list.
(g_signal_emit_by_name): same as g_signal_emit, working from
signal name+detail.
(signal_emit_R): return whether return_value needs to be altered.

* gtype.[hc]: set log-domain to GRuntime, i'm slowly getting to all
the points that need to reflect the upcoming rename.
melt g_type_conforms_to() functionality into g_type_is_a(), as that
is what we really want (liskov substitution principle).
assorted changes to other files due to conforms_to->is_a.

* gvalue.[hc]: implemented g_value_set_instance() that sets a value
from an instantiatable type via the value_table's collect_value()
function (based on an idea from James Henstridge <james@daa.com.au>).
cleanups/fixes.

* gvaluetypes.[hc]: implement G_TYPE_CCALLBACK and G_TYPE_PARAM.

23 years agoRemove pollfds from the context here, not when actually freeing the
Owen Taylor [Sun, 10 Dec 2000 16:02:48 +0000 (16:02 +0000)]
Remove pollfds from the context here, not when actually freeing the

Sun Dec 10 10:47:11 2000  Owen Taylor  <otaylor@redhat.com>

* gmain.c (g_source_destroy_internal): Remove pollfds
from the context here, not when actually freeing the
source.

* gmain.c (g_source_unref_internal): Free source list
and source, call source->source_funcs->destroy().

* giochannel.c: Unreference io_channel properly.

23 years agoRename from GLIB_DIVERT_BEFORE_HELP. Update to track autoconf 2.49b.
Raja R Harinath [Fri, 8 Dec 2000 22:48:45 +0000 (22:48 +0000)]
Rename from GLIB_DIVERT_BEFORE_HELP. Update to track autoconf 2.49b.

* acglib.m4 (GLIB_AC_DIVERT_BEFORE_HELP):
Rename from GLIB_DIVERT_BEFORE_HELP.
Update to track autoconf 2.49b.
* configure.in: Reflect above change.
(AC_EGREP_HEADER): Rename from really obselete AC_HEADER_EGREP.
(debug_default): Replace "if test `expr ...`" with "case".

23 years agoFix typo, and fix some formatting.
Raja R Harinath [Fri, 8 Dec 2000 21:23:30 +0000 (21:23 +0000)]
Fix typo, and fix some formatting.

23 years agoRearrange into other veriables, and include run-markup-tests.sh.
Raja R Harinath [Fri, 8 Dec 2000 21:04:52 +0000 (21:04 +0000)]
Rearrange into other veriables, and include run-markup-tests.sh.

* tests/Makefile.am (TESTS): Rearrange into other veriables, and
include run-markup-tests.sh.
(TESTS_ENVIRONMENT): New.  Pass $srcdir to tests.
(noinst_PROGRAMS): Rename to ...
(check_PROGRAMS): ... this.  'automake' ensures that these are
built before running the tests.
* tests/run-markup-tests.sh: Support $srcdir != $builddir.

23 years agomove markup-test to noinst_PROGRAMS; it isn't a proper test, the proper
Havoc Pennington [Fri, 8 Dec 2000 15:03:12 +0000 (15:03 +0000)]
move markup-test to noinst_PROGRAMS; it isn't a proper test, the proper

2000-12-08  Havoc Pennington  <hp@pobox.com>

* tests/Makefile.am (TESTS): move markup-test to noinst_PROGRAMS;
it isn't a proper test, the proper test would be
run-markup-tests.sh, but that can't go in tests, so we need
a manual make check rule. Didn't do that yet.

23 years agoDon't put anything after an #endif. Likewise.
Raja R Harinath [Fri, 8 Dec 2000 03:39:47 +0000 (03:39 +0000)]
Don't put anything after an #endif. Likewise.

* gmain.h: Don't put anything after an #endif.
* gmain.c: Likewise.

23 years agoAdd a bunch of unrefs.
Owen Taylor [Thu, 7 Dec 2000 20:29:58 +0000 (20:29 +0000)]
Add a bunch of unrefs.

Thu Dec  7 15:22:30 2000  Owen Taylor  <otaylor@redhat.com>

* tests/mainloop-test.c (recurser_start): Add a bunch
of unrefs.

* gmain.c (g_source_attach): Reference the source
when adding (pointed out by Elliot)

23 years agoadd g_strdupv test
Havoc Pennington [Thu, 7 Dec 2000 04:48:37 +0000 (04:48 +0000)]
add g_strdupv test

2000-12-06  Havoc Pennington  <hp@pobox.com>

* tests/strfunc-test.c (main): add g_strdupv test

* gstrfuncs.c (g_strdupv): Add a function to copy
an array of strings

23 years agoMajor change in API for creating sources to handle multiple main loops
Owen Taylor [Tue, 5 Dec 2000 20:45:33 +0000 (20:45 +0000)]
Major change in API for creating sources to handle multiple main loops

Tue Dec  5 12:23:04 2000  Owen Taylor  <otaylor@redhat.com>

        * gmain.[hc]: Major change in API for creating sources
to handle multiple main loops (GMainContext *).

GSources are now exposed as GSource * and implemented
with structure derivation.

* giochannel.[ch]: Changed vtable for GIOChannel to correspond
to the new mainloop API, add g_io_channel_create_watch().

* gtypes.h: Move GTimeVal here.

* gthread.h: Remove gmain.h include to avoid circularity.

        * giounix.c: Update for new GMain API.

* giowin32.c: Update for new GMain API. (No check for
proper compilation or working.)

* timeloop.c timeloop-basic.c: A benchmarking program for
the main loop comparing the main loop against a
hand-written (timeloop-basic.c) variant.

* tests/mainloop-test.c: New torture test of mainloop.

* docs/Changes-2.0.txt: Started. Added text about
changes to GMain.

* gmain.c (g_main_add_poll_unlocked): Initial fd->revents
to zero. (#8482, Benjamin Kahn)