platform/upstream/glib.git
24 years agoFix
Owen Taylor [Thu, 29 Jun 2000 02:53:12 +0000 (02:53 +0000)]
Fix

Wed Jun 28 22:52:00 2000  Owen Taylor  <otaylor@redhat.com>

* Makefile.am (libglib_la_SOURCES): Fix

24 years agospelling
Tim Janik [Wed, 28 Jun 2000 16:18:39 +0000 (16:18 +0000)]
spelling

24 years agodefine gstring in terms of gchar*. this typedef reflects the type name of
Tim Janik [Sat, 24 Jun 2000 22:30:10 +0000 (22:30 +0000)]
define gstring in terms of gchar*. this typedef reflects the type name of

Fri Jun 23 17:20:26 2000  Tim Janik  <timj@gtk.org>

        * glib.h: define gstring in terms of gchar*. this typedef reflects
        the type name of the primitive G_TYPE_STRING in the gobject module.

Sat Jun 24 23:03:04 2000  Tim Janik  <timj@gtk.org>

        * gtype.[hc]: provide G_TYPE_CHAR, G_TYPE_UCHAR, G_TYPE_BOOLEAN,
        G_TYPE_INT, G_TYPE_UINT, G_TYPE_LONG, G_TYPE_ULONG, G_TYPE_FLOAT,
        G_TYPE_DOUBLE and G_TYPE_STRING fundamental types.
        added a GTypeValueTable* pointer to GTypeInfo structure for types
        to implement GValue handling functions.
        GTypeValueTable contains the following members:
          value_init(): initialize a GValue structure.
          value_free(): free GValue structure contents (optional).
          value_copy(): copy one GValue contents to another GValue structure of
          collect_type: varargs collection type for the first variable argument
          to be collected by collect_value().
          collect_value(): variable arguments collection function (optional).
          lcopy_type: varargs collection type for the first variable argument
          to be location copyied by lcopy_value().
          lcopy_value(): variable arguments location copy function (optional).
        g_type_value_table_peek(): new function to retrive the GTypeValueTable*
        for a type node. ValueTables get inherited from parent types, unless
        overridden through the GTypeInfo structure. internally, GTypeValueTable
        support means an added overhead of one pointer per static or used
        dynamic type node.
        g_type_add_class_cache_func(): provide a cache_func/data pair to be
        called  prior to a type nodes last_unref() function, this can be used
        to prevent premature class destruction. multiple installed cache_func()
        will be chained upon last_unref() untill one of them returns TRUE.
        the cache_func()s have to check the type id passed in to figure whether
        they actually wants to cache the class of this type (since any types are
        routed through the cache_func() chain).
        g_type_remove_class_cache_func(): remove a previously installed
        cache_func/data pair. the cache maintained by this function has to be
        clear when calling g_type_remove_class_cache_func() to avoid leaks.
        g_type_class_unref_uncached(): class unref function for cache_func()
        implementations, unreferences a class omitting the cache chain (and
        therefore unref->cache->unref->... loops).

        * gvaluetypes.[hc]: provide the value setters/getters for the primitive
        fundamental types boolean, char, uchar, int, uint, long, ulong, float,
        double and string.

        * gvalue.[hc]: provide G_TYPE_IS_VALUE() in terms of whether a
        GTypeValueTable is provided for this type.
        removed g_value_init_default(), g_value_validate(), g_value_defaults(),
        g_value_set_default() and g_values_cmp() as these are supplied by the
        GParamSpec API now.
        moved g_values_exchange() into the "implementation details" section,
        since it just provides the underlying functionality for
        g_value_convert().

        * gvaluecollector.h: renamed the varargs value container from
        GParamCValue to GTypeCValue as the value collection methods are
        supplied by the type system now.
        G_PARAM_COLLECT_VALUE() and G_PARAM_LCOPY_VALUE() got renamed to
        G_VALUE_COLLECT() and G_VALUE_LCOPY() and operate without a
        GParamSpec structure now.

        * genums.h: macros cleanups/fixes.

        * genum.c: provide G_TYPE_ENUM and G_TYPE_FLAGS type
        and assorted g_value_{s|g}et_{enum|flags}() implementations.

        * gobject.[hc]:
        provide G_IS_VALUE_OBJECT(), G_TYPE_OBJECT ValueTable methods
        and g_value_{s|g}et_object().

        * gparam.[hc]: reduced class to value_set_default(),
        value_validate() and values_cmp(). also parameters now need to fill
        in a GType value_type; field to indicate the GValue type they
        are handling. provide g_param_value_set_default(),
        g_param_value_defaults(), g_param_value_validate() and
        g_param_values_cmp().

        * gparamspecs.[hc]: got rid of the g_value_* functions and
        the G_IS_VALUE_* macros. adapted param spec implementations
        according to the GParamSpecClass changes.

24 years agoInitial pass at adding unicode support functions. A few things still need
Owen Taylor [Wed, 21 Jun 2000 16:11:21 +0000 (16:11 +0000)]
Initial pass at adding unicode support functions. A few things still need

Wed Jun 21 12:09:03 2000  Owen Taylor  <otaylor@redhat.com>

* gunicode.h gutf8.c guniprop.c gunidecomp.[ch] gunichartables.h
Makefile.am glib.h: Initial pass at adding unicode support
functions. A few things still need to be implemented, a bit
of cleanup needs to be done, tests need to be added, and
the docs need to be finished, but this should allow replacing
most or all use of libunicode.

24 years agoupdated a bit more.
Damon Chaplin [Sun, 11 Jun 2000 18:32:13 +0000 (18:32 +0000)]
updated a bit more.

2000-06-11  Damon Chaplin  <damon@helixcode.com>

* tmpl/messages.sgml: updated a bit more.

24 years agoupdated a few bits about log handlers & flags.
Damon Chaplin [Sun, 11 Jun 2000 17:28:20 +0000 (17:28 +0000)]
updated a few bits about log handlers & flags.

2000-06-11  Damon Chaplin  <damon@helixcode.com>

* tmpl/messages.sgml: updated a few bits about log handlers & flags.

24 years agofetch the nth iface entry of the type node in the nth loop iteration, not
Tim Janik [Sat, 10 Jun 2000 06:41:55 +0000 (06:41 +0000)]
fetch the nth iface entry of the type node in the nth loop iteration, not

Sat Jun 10 08:38:27 2000  Tim Janik  <timj@gtk.org>

        * gtype.c (type_class_init): fetch the nth iface entry of the
        type node in the nth loop iteration, not alwys the first.
        bug discovered by Walt Pohl <cher@suitware.com>.
        (type_data_finalize_class_ifaces): same here, cut and paste
        rulez.

24 years agoAdd top_builddir so glibconfig.h is found properly when build srcdir !=
Owen Taylor [Wed, 7 Jun 2000 13:22:41 +0000 (13:22 +0000)]
Add top_builddir so glibconfig.h is found properly when build srcdir !=

Wed Jun  7 09:21:05 2000  Owen Taylor  <otaylor@redhat.com>

* Makefile.am (INCLUDES): Add top_builddir so glibconfig.h
is found properly when build srcdir != builddir. (Patch from
Eric Lemings)

24 years agoRephrase the ChangeLog comment a bit.
Tor Lillqvist [Tue, 6 Jun 2000 06:57:41 +0000 (06:57 +0000)]
Rephrase the ChangeLog comment a bit.

24 years agoIf we are watching the same fd several times (for different conditions,
Tor Lillqvist [Tue, 6 Jun 2000 06:56:03 +0000 (06:56 +0000)]
If we are watching the same fd several times (for different conditions,

2000-06-06  Tor Lillqvist  <tml@iki.fi>

* giowin32.c (g_io_channel_win32_pipe_readable): If we are
watching the same fd several times (for different conditions, with
different callbacks), check them all. Only call the callback for
G_IO_IN from here. (This bug popped up when a watch for
G_IO_ERR|G_IO_HUP was added to gimplib.)

24 years agoImplement on Win32.
Tor Lillqvist [Tue, 30 May 2000 20:46:03 +0000 (20:46 +0000)]
Implement on Win32.

2000-05-30  Tor Lillqvist  <tml@iki.fi>

* gutils.c (g_locale_get_codeset): Implement on Win32.

* glib.def: Add g_get_codeset.

* tests/Makefile.am (EXTRA_DIST): makefile.cygwin* has been
renamed to makefile.mingw*.

24 years agoMove the g_locale_get_codeset() up in the header file to correspond to to
Owen Taylor [Tue, 30 May 2000 20:05:23 +0000 (20:05 +0000)]
Move the g_locale_get_codeset() up in the header file to correspond to to

Tue May 30 16:01:32 2000  Owen Taylor  <otaylor@redhat.com>

* glib.h gutils.c: Move the g_locale_get_codeset() up in the
header file to correspond to to comments about memory
management. Rename to g_get_codeset() to avoid
polluting the g_locale_* namespace, which probably
would have g_locale_get_codeset (GLocale *locale).
Add a doc comment.

24 years agoAdd function to get the codeset name for the current locale.
Owen Taylor [Mon, 29 May 2000 18:48:29 +0000 (18:48 +0000)]
Add function to get the codeset name for the current locale.

Mon May 29 14:10:35 2000  Owen Taylor  <otaylor@redhat.com>

* gutils.c (g_locale_get_codeset): Add function to get the
codeset name for the current locale.

* configure.in acconfig.h: Add check for nl_langinfo(CODESET);

24 years agoadded argument assertments.
Tim Janik [Fri, 19 May 2000 10:01:08 +0000 (10:01 +0000)]
added argument assertments.

Fri May 19 11:39:29 2000  Tim Janik  <timj@gtk.org>

        * gutils.c (g_snprintf):
        (g_vsnprintf): added argument assertments.

        * gstring.c (g_string_assign): added argument assertments.
        (g_string_truncate): make len a guint.

24 years agofixed SIZEOF_LONG==4 assumption with ENABLE_MEM_CHECK, from Art Haas
Tim Janik [Fri, 19 May 2000 08:18:29 +0000 (08:18 +0000)]
fixed SIZEOF_LONG==4 assumption with ENABLE_MEM_CHECK, from Art Haas

Fri May 19 09:00:44 2000  Tim Janik  <timj@gtk.org>

        * gmem.c (g_free): fixed SIZEOF_LONG==4 assumption with
        ENABLE_MEM_CHECK, from Art Haas <ahaas@neosoft.com>.

        * gslist.c (g_slist_reverse): shut up compiler.

        * gscanner.c (g_scanner_get_token_ll): removed inline assignment.

        * garray.c: remove index>=0 checks for unsigned indices.

        * gmain.c (g_idle_prepare): timeout assignment fix.

        * gtree.c (g_tree_node_rotate_right): shut up compiler.

24 years agoSmall fixes to test progam build.
Tor Lillqvist [Sat, 13 May 2000 19:38:37 +0000 (19:38 +0000)]
Small fixes to test progam build.

24 years agomakefile.mingw.in tests/makefile.mingw.in Rename makefile.cygwin(.in) to
Tor Lillqvist [Sat, 13 May 2000 19:30:58 +0000 (19:30 +0000)]
makefile.mingw.in tests/makefile.mingw.in Rename makefile.cygwin(.in) to

2000-05-13  Tor Lillqvist  <tml@iki.fi>

* makefile.mingw.in
* tests/makefile.mingw.in
* build-dll: Rename makefile.cygwin(.in) to
makefile.mingw(.in), which better describes what it is. Move the
build of gmodule, gthread and gobject DLLs to makefiles in those
directories. Move resource file handling and build number bump to
build-dll, where it sits much cleaner.

* README.win32
* Makefile.am (EXTRA_DIST): Update accordingly.

* glib.h: Add G_PI, G_PI_2, G_PI_4, G_E, G_LN2, G_LN10 and
G_SQRT2. M_PI etc aren't necessarily in <math.h> in strict ISO C
implementations.

* glib.def: Add g_strcanon.

* gtree.c (g_tree_node_rotate_left): Remove unused variables.

* gwin32.c (g_win32_opendir): Remove unneeded statement.

gmodule:

* makefile.mingw.in: New file, with gmodule stuff
moved from ../makefile.mingw.in.

* Makefile.am: Add to EXTRA_DIST, and add rule to make makefile.mingw.

gobject:
* makefile.mingw.in
* gobject.def
* gobject.rc.in: New files, for Win32 (mingw) build.

* Makefile.am: Add to EXTRA_DIST. Add rules to produce the
corresponding non-*.in files.

* gtype.h: (Win32:) Mark _g_type_fundamental_last for
export/import from DLL.

gthread:
* makefile.mingw.in: New file, with gthread stuff moved from
../makefile.mingw.in.

* Makefile.am: Add to EXTRA_DIST, add rule to build makefile.mingw.

24 years agoshutup
Tim Janik [Fri, 12 May 2000 15:29:04 +0000 (15:29 +0000)]
shutup

24 years agoadded gobject
Tim Janik [Fri, 12 May 2000 15:24:42 +0000 (15:24 +0000)]
added gobject

Fri Apr 28 23:54:35 2000  Tim Janik  <timj@gtk.org>

        * setup things for a new sub-library libgobject:

        * Makefile.am (SUBDIRS): added gobject

        * glib-config.in: feature -lgobject.

        * configure.in (AC_OUTPUT): generate gobject/Makefile.

        * glib.m4 (AM_PATH_GLIB): feature gobject module.

        * glib.spec.in: added %{prefix}/lib/libgobject-1.3.so.*

24 years agoadded gobject
Tim Janik [Fri, 12 May 2000 15:23:16 +0000 (15:23 +0000)]
added gobject

Fri Apr 28 23:54:35 2000  Tim Janik  <timj@gtk.org>

        * setup things for a new sub-library libgobject:

        * Makefile.am (SUBDIRS): added gobject

        * glib-config.in: feature -lgobject.

        * configure.in (AC_OUTPUT): generate gobject/Makefile.

        * glib.m4 (AM_PATH_GLIB): feature gobject module.

        * glib.spec.in: added %{prefix}/lib/libgobject-1.3.so.*

Fri Apr 28 21:41:49 2000  Tim Janik  <timj@gtk.org>

        * glib.h: added G_STRLOC macro.
        G_STRUCT_OFFSET(): signedness corrections.
        (G_CSET_DIGITS): list 0-9.
        * gscanner.c (g_scanner_config_template): use G_CSET_DIGITS.

        * glib.h:
        * gstrfuncs.c:
        (g_strdown):
        (g_strup):
        (g_strreverse): return the modified string instead of void, so
        calls to these functions can be nested.
        (g_strcanon): new function, canonicalizes string according to
        a given character set.

Fri Apr 28 19:45:16 2000  Tim Janik  <timj@gtk.org>

        * gasyncqueue.c (g_async_queue_unref): get rid of an unused variable.

24 years agoadd reserved fundamental ids for gtk types (for transition time). added
Tim Janik [Fri, 12 May 2000 15:22:31 +0000 (15:22 +0000)]
add reserved fundamental ids for gtk types (for transition time). added

Fri May  5 01:15:48 2000  Tim Janik  <timj@gtk.org>

* gtype.h: add reserved fundamental ids for gtk types (for transition
time). added G_TYPE_FUNDAMENTAL_MAX for gtk.

Mon Apr 17 20:45:50 2000  Tim Janik  <timj@gtk.org>

* glib-gobject.c (g_object_base_class_finalize): oops, don't unset
n_params prior to destructing them.

Tue Apr 11 04:28:10 2000  Tim Janik  <timj@gtk.org>

* fixed a couple of bugs in the initial parameter/object
implementations, after getting beast running on GObject and GValue.

Fri Apr  7 04:27:49 2000  Tim Janik  <timj@gtk.org>

* glib-gobject.[hc]: completed parameter set/get implementations,
along with asyncronous parameter changed notification queue.

Sun Apr  2 04:54:36 2000  Tim Janik  <timj@gtk.org>

* glib-gobject.[hc]: GObject implementation, that is facilities
for setting/getting quarked data and reference counting.

* glib-gparamspecs.[hc]: first actuall parameter implementations
for GLib, so far we have: char, uchar, bool, int, uint, long,
ulong, enum, flags, float, double, string and object. each of these
GParamSpecs is a new instantiatable type in its own respect,
so the .c file derives 13 new types from G_TYPE_PARAM and
defines over 50 (*2) conversion facilities.

* glib-gvaluecollector.h: generic varargs handling stubs for
GParamSpecs, private header file (does get installed for
inclusion into user code though).

* glib-gvalue.[hc]: GValue functionality implementation.

* glib-gparam.[hc]: basis GParamSpec implementation for
the virtual base type G_TYPE_PARAM.

        * glib-genums.[hc]: enum/flags type implementation, based on
bseenum.[hc].

* glib-extra.[hc]: GLib additions, including 1.3 compatibility
routines and various other functions, from string manipulation
over list manipulation up to a unix signal GSource.

* glib-gtype.[hc]: GLib Type System implementation, heavily
based on BSE's dynamic type system.

24 years ago <entry size="small" status="90%" target="1.4">
Havoc Pennington [Thu, 11 May 2000 04:12:04 +0000 (04:12 +0000)]
   <entry size="small" status="90%" target="1.4">
         <title>Signal source for main loop</title>
       <description>
       <p>
                 Tim has a signal source for the main loop in bse,
           have to remember to get this in glib.
   </p>

         </description>
       <contact>timj@gtk.org</contact>
           </entry>

24 years agoPrint found version when test succeeds.
Owen Taylor [Thu, 11 May 2000 00:23:45 +0000 (00:23 +0000)]
Print found version when test succeeds.

Wed May 10 19:52:44 2000  Owen Taylor  <otaylor@redhat.com>

* glib.m4: Print found version when test succeeds.

24 years agomakefile.cygwin.in Include the common makefile snippet from
Tor Lillqvist [Thu, 4 May 2000 19:51:48 +0000 (19:51 +0000)]
makefile.cygwin.in Include the common makefile snippet from

2000-05-04  Tor Lillqvist  <tml@iki.fi>

* makefile.cygwin.in
* tests/makefile.cygwin.in: Include the common makefile snippet
from ../build/win32.

Maybe CVSROOT/modules should be changed so that the 'build' module
is included within the glib module (and gtk+, and gimp, and maybe
others later), in the same way as the 'macros' module is included
in lots of GNOME CVS modules?

24 years agoAdd new functions. Add new object files. Add threadpool-test. No need to
Tor Lillqvist [Mon, 1 May 2000 21:13:02 +0000 (21:13 +0000)]
Add new functions. Add new object files. Add threadpool-test. No need to

2000-05-02  Tor Lillqvist  <tml@iki.fi>

* glib.def: Add new functions.
* makefile.{cygwin,msc}.in (glib_OBJECTS): Add new object files.
* tests/makefile.{cygwin,msc}.in: Add threadpool-test.
* gmodule-win32.c: No need to include <psapi.h>

24 years agoNew File implementing an asynchronous queue to be used for asynchronous
Sebastian Wilhelmi [Fri, 28 Apr 2000 12:24:53 +0000 (12:24 +0000)]
New File implementing an asynchronous queue to be used for asynchronous

2000-04-28  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* gasyncqueue.c: New File implementing an asynchronous queue to be
used for asynchronous inter-thread communication.

* gthreadpool.c: New File implementing a thread pool to be used
for distributing work among several threads.

* glib.h: Added the type and function declarations for these two
types.

* tests/threadpool-test.c: New File implementing a test for the
thread pool. This also checks the asynchronous queue underlying
the thread pool.

* tests/Makefile.am: Changed accordingly.

24 years agoLook for both pthread_create and pthread_join in the thread library. Some
Sebastian Wilhelmi [Wed, 26 Apr 2000 13:03:49 +0000 (13:03 +0000)]
Look for both pthread_create and pthread_join in the thread library. Some

2000-04-26  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* configure.in: Look for both pthread_create and pthread_join in
the thread library. Some systems define one of them, but not both
in libc. Arghh. Now we really start a thread and join it later and
check, whether the thread to actually ran.

24 years agoChangeLog update: This fixes Bug #8267. Thanks to Juan Toledo
Sebastian Wilhelmi [Wed, 26 Apr 2000 08:48:24 +0000 (08:48 +0000)]
ChangeLog update: This fixes Bug #8267. Thanks to Juan Toledo

ChangeLog update:
This fixes Bug #8267. Thanks to Juan Toledo
        <toledo@users.sourceforge.net> for + pointing that out.

24 years agoChanged the 'value' parameter of g_cache_remove from gpointer to
Sebastian Wilhelmi [Wed, 26 Apr 2000 08:42:19 +0000 (08:42 +0000)]
Changed the 'value' parameter of g_cache_remove from gpointer to

2000-04-26  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* glib.h, gcache.c, gtree.c: Changed the 'value' parameter of
g_cache_remove from gpointer to gconstpointer. Dito for the 'key'
parameter of g_tree_lookup and g_tree_remove and the 'data'
parameter of g_tree_serach. This function now takes a function of
type GCompareFunc instead of GSearchFunc.

* glib.h: Removed declaration of GSearchFunc.

* gmem.c: s/GSearchFunc/GCompareFunc/.

24 years agoChanged the scope of the initialized mutex to USYNC_THREAD. Thanks to
Sebastian Wilhelmi [Tue, 25 Apr 2000 13:44:43 +0000 (13:44 +0000)]
Changed the scope of the initialized mutex to USYNC_THREAD. Thanks to

2000-04-25  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* gthread-solaris.c (g_mutex_new_solaris_impl): Changed the scope
of the initialized mutex to USYNC_THREAD. Thanks to Soeren
Sandmann <sandmann@daimi.au.dk> for pointing that out.

24 years agoUpdate entry point list.
Tor Lillqvist [Wed, 19 Apr 2000 18:40:31 +0000 (18:40 +0000)]
Update entry point list.

2000-04-19  Tor Lillqvist  <tml@iki.fi>

* glib.def: Update entry point list.

24 years agoMade the debugging G_TRYLOCK call also work for compilers with funnt
Sebastian Wilhelmi [Wed, 19 Apr 2000 09:29:19 +0000 (09:29 +0000)]
Made the debugging G_TRYLOCK call also work for compilers with funnt

2000-04-19  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* glib.h (G_TRYLOCK): Made the debugging G_TRYLOCK call also work
for compilers with funnt G_STMT_(START|END) macros.

* tests/thread-test.c: Implemented a check for that.

24 years agoChanged the win32 part of this function to be thread safe and to make the
Sebastian Wilhelmi [Wed, 19 Apr 2000 08:43:52 +0000 (08:43 +0000)]
Changed the win32 part of this function to be thread safe and to make the

2000-04-19  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* gutils.c (g_getenv): Changed the win32 part of this function to
be thread safe and to make the returned environment string
persistent to match the UN*X behavior. This is again a response to
Bug #8983.

* glib.h (G_LOCK_NAME): Removed parentheses around the lock name,
as that seems to cause problems for some compilers and really
isn't necessary.

24 years agomake sure that scanner->config->cset_skip_characters is "" instead of
Tim Janik [Wed, 19 Apr 2000 06:49:39 +0000 (06:49 +0000)]
make sure that scanner->config->cset_skip_characters is "" instead of

Wed Apr 19 08:32:32 2000  Tim Janik  <timj@gtk.org>

        * gscanner.c (g_scanner_new): make sure that
        scanner->config->cset_skip_characters is "" instead of NULL, so we
        don't segfault further on.

24 years agoChanged the 'data' parameters from gpointer to gconstpointer for the
Sebastian Wilhelmi [Tue, 18 Apr 2000 14:01:33 +0000 (14:01 +0000)]
Changed the 'data' parameters from gpointer to gconstpointer for the

2000-04-18  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* glib.h, glist.h, gslist.h: Changed the 'data' parameters from
gpointer to gconstpointer for the functions
g_(list|slist)_(remove|find|find_custom|index), as they do not
change this parameter. This fixes bug #4836.

24 years agoChanged comment for g_getenv to reflect, that the returned memory must not
Sebastian Wilhelmi [Tue, 18 Apr 2000 13:00:35 +0000 (13:00 +0000)]
Changed comment for g_getenv to reflect, that the returned memory must not

2000-04-18  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* glib.h: Changed comment for g_getenv to reflect, that the
returned memory must not be freed. Fixes Bug #8983.

24 years agoAdd configure test for garbage collector friendliness for GLib. If
Sebastian Wilhelmi [Mon, 17 Apr 2000 13:23:27 +0000 (13:23 +0000)]
Add configure test for garbage collector friendliness for GLib. If

2000-04-17  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* configure.in, acconfig.h: Add configure test for garbage
collector friendliness for GLib. If enabled, ENABLE_GC_FRIENDLY
will be defined.

* garray.c, ghash.c, glist.c, gmain.c, gmem.c, gnode.c, gqueue.c,
gslist.c, gtree.c: If ENABLE_GC_FRIENDLY is defined, NULLify all
memory released by the user, but cached by GLib. This lets a
garbage collector have a more correct view of the actually used
memory.

24 years agoAdded g_(array|ptr_array|byte_array)_sized_new functions, that reserve a
Sebastian Wilhelmi [Mon, 17 Apr 2000 10:59:46 +0000 (10:59 +0000)]
Added g_(array|ptr_array|byte_array)_sized_new functions, that reserve a

2000-04-17  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* garray.c, glib.h: Added g_(array|ptr_array|byte_array)_sized_new
functions, that reserve a certain amount of memeory for the array
at creation time to avoid reallocation. Fixes bug #6707 from
Charles Kerr <ckerr@osserver1.nssl.noaa.gov>.

24 years agoRenamed g_queue_create to g_queue_new in conformance to all other GLib
Sebastian Wilhelmi [Mon, 17 Apr 2000 10:31:10 +0000 (10:31 +0000)]
Renamed g_queue_create to g_queue_new in conformance to all other GLib

2000-04-17  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* glib.h, gqueue.c, tests/queue-test.c (main): Renamed
g_queue_create to g_queue_new in conformance to all other GLib
data types.

24 years agoupdated.
Damon Chaplin [Sat, 15 Apr 2000 23:34:34 +0000 (23:34 +0000)]
updated.

2000-04-16  Damon Chaplin  <damon@helixcode.com>

        * tmpl/linked_lists_single.sgml:
        * tmpl/linked_lists_double.sgml:
        * tmpl/trees-nary.sgml: updated.

        * tmpl/modules.sgml: described g_module_build_path().

        * tmpl/date.sgml: made short description lower case and end in a '.'.

        * glib-sections.txt: rearranged GDate section.

        * tmpl/arrays.sgml:
        * tmpl/arrays_byte.sgml:
        * tmpl/arrays_pointer.sgml: updated.

24 years agoFixed bug. Thanks to Marko Kreen <marko@l-t.ee> for reporting that.
Sebastian Wilhelmi [Fri, 7 Apr 2000 11:44:43 +0000 (11:44 +0000)]
Fixed bug. Thanks to Marko Kreen <marko@l-t.ee> for reporting that.

2000-04-07  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* grand.c (g_rand_new): Fixed bug. Thanks to Marko Kreen
<marko@l-t.ee> for reporting that.

24 years agoAdd note about command line arg parsing, and fix percent done on regexp
Havoc Pennington [Mon, 27 Mar 2000 23:46:00 +0000 (23:46 +0000)]
Add note about command line arg parsing, and fix percent done on regexp
support

24 years agoAdd TODO file
Havoc Pennington [Mon, 27 Mar 2000 23:19:33 +0000 (23:19 +0000)]
Add TODO file

24 years agoTell about using the mingw-based gcc, which is much easier than modifying
Tor Lillqvist [Sun, 26 Mar 2000 20:56:07 +0000 (20:56 +0000)]
Tell about using the mingw-based gcc, which is much easier than modifying

2000-03-26  Tor Lillqvist  <tml@iki.fi>

* README.win32: Tell about using the mingw-based gcc, which is
much easier than modifying the cygwin gcc to product mingw code
for the msvcrt runtime.

* makefile.cygwin.in (WIN32APIHEADERS): Kludge to make it work
with a "pure" mingw gcc, too.

24 years agotypo alarm
Sebastian Wilhelmi [Fri, 24 Mar 2000 15:37:24 +0000 (15:37 +0000)]
typo alarm

typo alarm

24 years agoMade GArray behave correct. Now zero_terminated really means, that the
Sebastian Wilhelmi [Fri, 24 Mar 2000 15:36:03 +0000 (15:36 +0000)]
Made GArray behave correct. Now zero_terminated really means, that the

2000-03-24  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* garray.c: Made GArray behave correct. Now zero_terminated really
means, that the element array->data[array->len] exists and is
zeroed, and clear means that any unassigned elements obtained
through g_array_set_size (the only way to get unassigned elements
AFAICT) are zeroed. Added some macors to mak the code more
obvoius. Also made GPtrArray zero elements after
g_ptr_array_set_size. This is done in a portbale way (assignment
of NULL instead of just memsetting it to zero), though that might
be more portability than we actually want.

24 years ago.cvsignore adaptation
Sebastian Wilhelmi [Fri, 24 Mar 2000 15:09:02 +0000 (15:09 +0000)]
.cvsignore adaptation

.cvsignore adaptation

24 years agoAdded various win32 related *.in files to EXTRA_DIST to let 'make
Sebastian Wilhelmi [Fri, 24 Mar 2000 12:56:56 +0000 (12:56 +0000)]
Added various win32 related *.in files to EXTRA_DIST to let 'make

2000-03-24  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* Makefile.am, gthread/Makefile.am, gmodule/Makefile.am,
tests/Makefile.am: Added various win32 related *.in files to
EXTRA_DIST to let 'make distcheck' procude all the corresponding
files, which it silently fails to do currently.

24 years agoAfter finding the right thread library (containing e.g. pthread_create) we
Sebastian Wilhelmi [Thu, 23 Mar 2000 16:23:50 +0000 (16:23 +0000)]
After finding the right thread library (containing e.g. pthread_create) we

2000-03-23  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* configure.in: After finding the right thread library (containing
e.g. pthread_create) we now search for the right realtime library
(containing e.g. sched_get_priority_max). Makes the output of the
thread related libraries correct.

* gtimer.c (g_usleep): The current implementation of g_usleep
(simply calling select) doesn't work reliable for multi-threaded
programs on some platforms (bad omen for the main loop....), so I
changed the implementation for thread-using programs to wait for a
GCond for the specified amount of time (NB: sleep and usleep are
not MT-safe in general, because they often use signals).

24 years agoIn the "null" module case _g_module_symbol should still first search the main
Tor Lillqvist [Wed, 22 Mar 2000 22:51:54 +0000 (22:51 +0000)]
In the "null" module case _g_module_symbol should still first search the main
program before searching all the loaded DLLs.

24 years agoWhen looking for symbols in the "main" module we must search both the main
Tor Lillqvist [Wed, 22 Mar 2000 22:34:48 +0000 (22:34 +0000)]
When looking for symbols in the "main" module we must search both the main

2000-03-23  Tor Lillqvist  <tml@iki.fi>

* gmodule-win32.c (_g_module_symbol): When looking for symbols in
the "main" module we must search both the main program and all
currently loaded DLLs. Not only the main program, or even just the
DLLs loaded as gmodules.  Libglade requires this.

Thus we need to get a list of all modules in the current
process. There are two alternative APIs to do this: PSAPI and
Toolhelp. The former is only available on NT (including Win2k),
the latter on Win9x and Win2k. Check which one works, and use
that.

Code for using PSAPI and Toolhelp was borrowed from the Dr. Mingw
tool written by José Fonseca <em96115@fe.up.pt>. Thanks.

24 years agoFix indentation.
Owen Taylor [Wed, 22 Mar 2000 21:59:16 +0000 (21:59 +0000)]
Fix indentation.

Wed Mar 22 16:49:57 2000  Owen Taylor  <otaylor@redhat.com>

* gmem.c (g_mem_chunk_area_compare): Fix indentation.

24 years agogmem.c (g_mem_chunk_area_compare): Fix 64-bitness bug in comparing two
Elliot Lee [Wed, 22 Mar 2000 21:41:45 +0000 (21:41 +0000)]
gmem.c (g_mem_chunk_area_compare): Fix 64-bitness bug in comparing two

gmem.c (g_mem_chunk_area_compare): Fix 64-bitness bug in comparing two
pointers more than 4G apart.

24 years agoMove Win32-only includes after inclusion of glib.h, so that G_OS_WIN32 is
Tor Lillqvist [Wed, 22 Mar 2000 20:46:19 +0000 (20:46 +0000)]
Move Win32-only includes after inclusion of glib.h, so that G_OS_WIN32 is

2000-03-22  Tor Lillqvist  <tml@iki.fi>

* gutils.c: Move Win32-only includes after inclusion of glib.h, so
that G_OS_WIN32 is defined.

* glibconfig.h.win32.in: Add GSystemThread.

24 years agobetter ChangeLog Entry
Sebastian Wilhelmi [Wed, 22 Mar 2000 15:18:57 +0000 (15:18 +0000)]
better ChangeLog Entry

better ChangeLog Entry

24 years agoOk, Marcus Brinkmann <Marcus.Brinkmann@ruhr-uni-bochum.de> convinced me,
Sebastian Wilhelmi [Wed, 22 Mar 2000 15:18:11 +0000 (15:18 +0000)]
Ok, Marcus Brinkmann <Marcus.Brinkmann@ruhr-uni-bochum.de> convinced me,

2000-03-22  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* gutils.c (g_get_current_dir): Ok, Marcus Brinkmann
<Marcus.Brinkmann@ruhr-uni-bochum.de> convinced me, that 128 KB
path length might not be enough for the HURD. So I changed the
loop to at least avoid an integer overflow, which could happen at
beyond 2GB size ;-)

24 years agoTest for sched_yield as the native yield function first. Corrected typo
Sebastian Wilhelmi [Wed, 22 Mar 2000 10:04:29 +0000 (10:04 +0000)]
Test for sched_yield as the native yield function first. Corrected typo

2000-03-22  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* configure.in: Test for sched_yield as the native yield function
first. Corrected typo g_thread_sleep -> g_usleep. Corrected
message for the pthread_create test. Negative Priorities are
allowed (and used on Solaris), so consider
sched_get_priority_min failed only if it returns -1, not <0. Check
for sched_get_priority_min also in -lrt, if not found in -lpthread
alone and add -lrt to G_THREAD_LIBS then. Remove special case
handling of priorities for older solaris versions and posix
threads. Thanks to Wan-Teh Chang <wtc@netscape.com> for suggesting
some of those changes.

* config.guess, config.sub, ltconfig, ltmain.sh:
Upgrade to libtool 1.3.4.

24 years ago.cvsignore updates
Sebastian Wilhelmi [Wed, 22 Mar 2000 10:02:34 +0000 (10:02 +0000)]
.cvsignore updates

.cvsignore updates

24 years agoalways define G_GNUC_EXTENSION, even when not needed by GLib. That's
Sebastian Wilhelmi [Tue, 21 Mar 2000 15:21:41 +0000 (15:21 +0000)]
always define G_GNUC_EXTENSION, even when not needed by GLib. That's

2000-03-21  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* glib.h, configure.in, gutils.h: always define G_GNUC_EXTENSION,
even when not needed by GLib. That's actually also the way, the
GLib reference manual describes that macro. Therefore I had to
remove the lonesome #include <glibconfig.h> in gutils.c, which
doesn't seem to be needed there however. This change should make
Ben Gertzfield <che@debian.org> happy.

* gutils.c: Furthermore two warnings in gutils.c were voided,
which crept in due to my last change.

24 years agoAllocate only up to 128KB for a pathname. While this is an arbitrary value
Sebastian Wilhelmi [Tue, 21 Mar 2000 13:46:59 +0000 (13:46 +0000)]
Allocate only up to 128KB for a pathname. While this is an arbitrary value

2000-03-21  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* gutils.c (g_get_current_dir): Allocate only up to 128KB for a
pathname. While this is an arbitrary value just like 2048, it
seems to be enough (after all, even 4GB is an arbitrary value).

24 years agoForgot a '}' in a macro for DCE-threads. Thanks to Karl Nelson
Sebastian Wilhelmi [Mon, 20 Mar 2000 16:43:28 +0000 (16:43 +0000)]
Forgot a '}' in a macro for DCE-threads. Thanks to Karl Nelson

2000-03-20  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* gthread-posix.c (posix_check_for_error): Forgot a '}' in a macro
for DCE-threads. Thanks to Karl Nelson <kenelson@ece.ucdavis.edu>
for pointing that out.

24 years agoWarn in case of an error during the call to poll(2). Closes Bug#7564 as
Sebastian Wilhelmi [Mon, 20 Mar 2000 16:19:58 +0000 (16:19 +0000)]
Warn in case of an error during the call to poll(2). Closes Bug#7564 as

2000-03-20  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* gmain.c (g_main_poll): Warn in case of an error during the call
to poll(2). Closes Bug#7564 as reported by David Helder
<dhelder@umich.edu>.

24 years agoMake g_get_current_dir work on systems with unlimited pathname length like
Sebastian Wilhelmi [Mon, 20 Mar 2000 16:01:41 +0000 (16:01 +0000)]
Make g_get_current_dir work on systems with unlimited pathname length like

2000-03-20  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* gutils.c (g_get_current_dir): Make g_get_current_dir work on
systems with unlimited pathname length like the HURD (It worked
there before, but only for pathes shorter than 2048). Closes
Bug#4525 as reported by Marcus Brinkmann
<Marcus.Brinkmann@ruhr-uni-bochum.de>.

24 years agoInterpret EINTR as G_IO_ERROR_AGAIN.
Sebastian Wilhelmi [Mon, 20 Mar 2000 13:19:20 +0000 (13:19 +0000)]
Interpret EINTR as G_IO_ERROR_AGAIN.

2000-03-17  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* giounix.c (g_io_unix_write, g_io_unix_read): Interpret EINTR as
G_IO_ERROR_AGAIN.

24 years agoAdded the missing POSIX_NO_YIELD and POSIX_NO_PRIORITIES warning messages.
Sebastian Wilhelmi [Fri, 17 Mar 2000 14:49:59 +0000 (14:49 +0000)]
Added the missing POSIX_NO_YIELD and POSIX_NO_PRIORITIES warning messages.

2000-03-17  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* configure.in: Added the missing POSIX_NO_YIELD and
POSIX_NO_PRIORITIES warning messages.

* configure.in: Use AC_TRY_RUN instead of AC_TRY_LINK, to test for
real thread support. On solaris pthread_create can be linked to
even in -lc, but it doesn't work then.

* configure.in: Don't use priorities for threads, when the
minimal/maximal priorities couldn't be determined at configure
time.

* configure.in, gthread.c: Always define GSystemThread in
glibconfig.h to represent a system thread.

* configure.in: Do not use native recursive threads, when
possibe. We use some features, that they do not expose (namely the
depth counter).

* glib.h, gthread.c: Redefined GStaticRecMutex. The functions are
now implemented in a different way, which should be way
faster. Alsothere are now functions g_static_rec_mutex_unlock_full
and g_static_rec_mutex_lock_full to leave/enter a recursive mutex
completly.

* gthread.c (g_thread_self): Do not test the system_thread to be
non-zero to speed things up.

* gthread.c (g_mutex_init): Therefore set the system_thread of the
main thread here.

* tests/thread-test.c: Rerun all tests once again, but this time
we fool the system into thinking, that the available thread system
is not native, but userprovided.

* gthread/gthread-posix.c: Don't use priorities for threads,
when the minimal/maximal priorities couldn't be determined at
configure time.

* gthread/gthread-posix.c: Don't check for errors, when
setting the scope of a tread to system, as some posix
implementations can't do that and we don't want the thing to
fail because of that.

24 years agoWe want the next and prev pointer of the inserted link to be NULL.
Sebastian Wilhelmi [Mon, 13 Mar 2000 16:49:40 +0000 (16:49 +0000)]
We want the next and prev pointer of the inserted link to be NULL.

2000-03-13  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* gqueue.c (g_queue_push_tail_link, g_queue_push_head_link): We
want the next and prev pointer of the inserted link to be NULL.

24 years agoAnother small change to the pthread_.. search pattern. Should work *now*
Sebastian Wilhelmi [Mon, 6 Mar 2000 09:09:57 +0000 (09:09 +0000)]
Another small change to the pthread_.. search pattern. Should work *now*

2000-03-06  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* configure.in: Another small change to the pthread_.. search
pattern. Should work *now* for AIX.

24 years agoNew function that returns the message string for a Win32 error code.
Tor Lillqvist [Fri, 3 Mar 2000 09:36:32 +0000 (09:36 +0000)]
New function that returns the message string for a Win32 error code.

2000-03-04  Tor Lillqvist  <tml@iki.fi>

* gwin32.c (g_win32_error_message): New function that returns the
message string for a Win32 error code.

* glib.h: Declare it.

* glib.def: Export it, plus g_node_copy.

* gmodule-win32.c: Call g_win32_error_message() to get the error
message strings.

* libgplugin_a.c
* libgplugin_b.c
* gmodule.c: Remove LibMain functions that were needed by LCC,
which is not a supported compiler.

* testgmodule.c (main): Test for G_MODULE_IMPL ==
G_MODULE_IMPL_WIN32, not G_OS_WIN32.

* gmoduleconf.h.win32: Remove LCC stuff from here, too.

24 years agoMake the search for pthread_attr_... prototypes find names at the start of
Sebastian Wilhelmi [Fri, 3 Mar 2000 09:36:32 +0000 (09:36 +0000)]
Make the search for pthread_attr_... prototypes find names at the start of

2000-03-03  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* configure.in: Make the search for pthread_attr_... prototypes
find names at the start of a line also, like it is on AIX. Thanks
to Valdis Kletnieks <Valdis.Kletnieks@vt.edu> for the info.

24 years agoshut up cvs
Tim Janik [Wed, 1 Mar 2000 09:44:32 +0000 (09:44 +0000)]
shut up cvs

24 years agoNew function that returns the message string for a Win32 error code.
Tor Lillqvist [Wed, 1 Mar 2000 09:44:10 +0000 (09:44 +0000)]
New function that returns the message string for a Win32 error code.

2000-03-04  Tor Lillqvist  <tml@iki.fi>

* gwin32.c (g_win32_error_message): New function that returns the
message string for a Win32 error code.

* glib.h: Declare it.

* glib.def: Export it, plus g_node_copy.

* gmodule-win32.c: Call g_win32_error_message() to get the error
message strings.

* libgplugin_a.c
* libgplugin_b.c
* gmodule.c: Remove LibMain functions that were needed by LCC,
which is not a supported compiler.

* testgmodule.c (main): Test for G_MODULE_IMPL ==
G_MODULE_IMPL_WIN32, not G_OS_WIN32.

* gmoduleconf.h.win32: Remove LCC stuff from here, too.

24 years agominor optimization.
Tim Janik [Wed, 1 Mar 2000 09:44:10 +0000 (09:44 +0000)]
minor optimization.

Wed Mar  1 10:39:39 2000  Tim Janik  <timj@gtk.org>

        * gslist.c (g_slist_reverse): minor optimization.

        * testglib.c (g_node_test): added a couple of tests for
        g_node_copy().

        * glib.h:
        * gnode.c (g_node_copy): new function to copy subtrees,
        supplied by dbsears@ix.netcom.com.
        changed iterator to walk the children list backwards, so
        we get down from O(n^2) to O(n).

        * gnode.c (g_node_first_sibling): applied patch from
        dbsears@ix.netcom.com to optimize access if node->parent
        is present.

        * gutils.c (g_get_any_init): backed out HAVE_PW_GECOS check around
        assignment of g_real_name, sicne HAVE_PW_GECOS is never defined and
        thus breaks the original code.

        * merged changes from 1.2.7.

Sat Feb 19 19:43:29 2000  Tim Janik  <timj@gtk.org>

        * testgmodule.c (main): added test to check that not yet bound symbols
        in shared libraries of the main module are retrievable, from David Gero.

Fri Jan 28 11:37:41 2000  Owen Taylor  <otaylor@redhat.com>

        Bug #4156 - Changes vaguely modelled after Scott Gifford's patch

        * gtimer.c (g_timer_elapsed): Never report negative times -
        clip times to 0.

        * gmain.c (g_timeout_prepare): Guard against unexpected
        clock shifts by never setting a timeout of more than
        data->interval msecs.

24 years agodo not return NULL symbols.
Tim Janik [Wed, 1 Mar 2000 04:57:07 +0000 (04:57 +0000)]
do not return NULL symbols.

Wed Mar  1 05:34:47 2000  Tim Janik  <timj@gtk.org>

        * gmodule-beos.c (_g_module_symbol): do not return NULL symbols.

        * gmodule-os2.c: removed NetBSD specific defines.
        (_g_module_self): set an error message for unsupported behaviour.

        * gmodule-beos.c: many coding style fixups.
        (_g_module_open):
        (_g_module_self):
        (_g_module_close):
        (_g_module_symbol): bunch of memory leaks plugged.

        * gmodule-dl.c: make sure the error message returned from dlerror()
        is always != NULL, by using a wrapper function fetch_dlerror(). based
        on a patch to fix _g_module_symbol() for NetBSD from Scott Presnell
        <srp@zgi.com>.

        * gmodule-dld.c: minor indentation.

        * gmodule-win32.c: minor cleanups.

        * merges from glib-1-2.

24 years agoNew function that returns the message string for a Win32 error code.
Tor Lillqvist [Wed, 1 Mar 2000 04:57:07 +0000 (04:57 +0000)]
New function that returns the message string for a Win32 error code.

2000-03-04  Tor Lillqvist  <tml@iki.fi>

* gwin32.c (g_win32_error_message): New function that returns the
message string for a Win32 error code.

* glib.h: Declare it.

* glib.def: Export it, plus g_node_copy.

* gmodule-win32.c: Call g_win32_error_message() to get the error
message strings.

* libgplugin_a.c
* libgplugin_b.c
* gmodule.c: Remove LibMain functions that were needed by LCC,
which is not a supported compiler.

* testgmodule.c (main): Test for G_MODULE_IMPL ==
G_MODULE_IMPL_WIN32, not G_OS_WIN32.

* gmoduleconf.h.win32: Remove LCC stuff from here, too.

24 years agodo not return NULL symbols.
Tim Janik [Wed, 1 Mar 2000 04:57:07 +0000 (04:57 +0000)]
do not return NULL symbols.

Wed Mar  1 05:34:47 2000  Tim Janik  <timj@gtk.org>

        * gmodule-beos.c (_g_module_symbol): do not return NULL symbols.

        * gmodule-os2.c: removed NetBSD specific defines.
        (_g_module_self): set an error message for unsupported behaviour.

        * gmodule-beos.c: many coding style fixups.
        (_g_module_open):
        (_g_module_self):
        (_g_module_close):
        (_g_module_symbol): bunch of memory leaks plugged.

        * gmodule-dl.c: make sure the error message returned from dlerror()
        is always != NULL, by using a wrapper function fetch_dlerror(). based
        on a patch to fix _g_module_symbol() for NetBSD from Scott Presnell
        <srp@zgi.com>.

        * gmodule-dld.c: minor indentation.

        * gmodule-win32.c: minor cleanups.

        * merges from glib-1-2.

24 years agoNew function that returns the message string for a Win32 error code.
Tor Lillqvist [Sun, 27 Feb 2000 10:21:30 +0000 (10:21 +0000)]
New function that returns the message string for a Win32 error code.

2000-03-04  Tor Lillqvist  <tml@iki.fi>

* gwin32.c (g_win32_error_message): New function that returns the
message string for a Win32 error code.

* glib.h: Declare it.

* glib.def: Export it, plus g_node_copy.

* gmodule-win32.c: Call g_win32_error_message() to get the error
message strings.

* libgplugin_a.c
* libgplugin_b.c
* gmodule.c: Remove LibMain functions that were needed by LCC,
which is not a supported compiler.

* testgmodule.c (main): Test for G_MODULE_IMPL ==
G_MODULE_IMPL_WIN32, not G_OS_WIN32.

* gmoduleconf.h.win32: Remove LCC stuff from here, too.

24 years agoAdd new functions.
Tor Lillqvist [Sun, 27 Feb 2000 10:21:30 +0000 (10:21 +0000)]
Add new functions.

2000-02-27  Tor Lillqvist  <tml@iki.fi>

* glib.def: Add new functions.

24 years agoAdd a missing step to the setup instructions for gcc-2.95.2. Thanks to
Tor Lillqvist [Wed, 23 Feb 2000 20:09:16 +0000 (20:09 +0000)]
Add a missing step to the setup instructions for gcc-2.95.2. Thanks to

2000-02-23  Tor Lillqvist  <tml@iki.fi>

* README.win32: Add a missing step to the setup instructions for
gcc-2.95.2. Thanks to Arnaud Charlet.

* glib.def: Add missing entry point.

24 years agocheck for sysconf (_SC_THREAD_STACK_MIN), which returns the minimal stack
Sebastian Wilhelmi [Tue, 22 Feb 2000 13:11:24 +0000 (13:11 +0000)]
check for sysconf (_SC_THREAD_STACK_MIN), which returns the minimal stack

2000-02-22  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* gthread-posix.c, gthread-solaris.c: check for sysconf
(_SC_THREAD_STACK_MIN), which returns the minimal stack size for
new threads. Patch from Soeren Sandmann <sandmann@daimi.au.dk>.

24 years agoupdated the g_source_remove_by_XXX() descriptions to note that only the
Damon Chaplin [Mon, 21 Feb 2000 21:08:17 +0000 (21:08 +0000)]
updated the g_source_remove_by_XXX() descriptions to note that only the

2000-02-21  Damon Chaplin  <damon@helixcode.com>

        * tmpl/main.sgml: updated the g_source_remove_by_XXX() descriptions
        to note that only the first source found is removed.

24 years agoChanged GCC version test to also accept major versions > 2. Thanks to Ben
Sebastian Wilhelmi [Fri, 18 Feb 2000 09:57:19 +0000 (09:57 +0000)]
Changed GCC version test to also accept major versions > 2. Thanks to Ben

2000-02-18  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* configure.in: Changed GCC version test to also accept major
versions > 2. Thanks to Ben Gertzfield <che@debian.org> for
pointing this out.

24 years agochanged g_str_hash() to a 31 bit version based on a submission by Karl
Havoc Pennington [Thu, 17 Feb 2000 11:57:35 +0000 (11:57 +0000)]
changed g_str_hash() to a 31 bit version based on a submission by Karl

Thu Feb 17 12:53:44 2000  Tim Janik  <timj@gtk.org>

        * gstring.c: changed g_str_hash() to a 31 bit version based on
        a submission by Karl Nelson and hand optimized ad absurdum by
        various people ;)

        * gstring.c: applied patch from havoc for new gstring functions,
        added some more sanity checks, coding style fixups.

2000-02-13  Havoc Pennington  <hp@pobox.com>

        * tests/string-test.c (main): Add tests for the new GString
        features

        * testglib.c (main): Add tests for the new GString features

        * gstring.c (g_string_insert_len): New function; insert
        a given length of string at a given position.
        (g_string_append): reimplement in terms of g_string_insert_len
        (g_string_append_len): new function
        (g_string_insert_c): accept -1 for "pos" arg to mean "append"
        (g_string_append_c): reimplement in terms of g_string_insert_c
        (g_string_prepend): reimplement in terms of g_string_insert_len
        (g_string_prepend_len): new function
        (g_string_prepend_c): reimplement in terms of g_string_insert_c
        (g_string_insert): reimplement in terms of g_string_insert_len

        * glib.h: Declare g_string_insert_len, g_string_append_len,
        g_string_prepend_len

24 years agowtf??? someone destroyed the configure.in, reverting to an older version
Tim Janik [Sun, 13 Feb 2000 07:18:23 +0000 (07:18 +0000)]
wtf??? someone destroyed the configure.in, reverting to an older version

Sun Feb 13 08:16:47 2000  Tim Janik  <timj@gtk.org>

        * configure.in: wtf??? someone destroyed the configure.in, reverting to
        an older version from Feb 4 which apears to still work.

24 years agoMove to corresponding subdirectories.
Tor Lillqvist [Mon, 7 Feb 2000 21:22:30 +0000 (21:22 +0000)]
Move to corresponding subdirectories.

2000-02-07  Tor Lillqvist  <tml@iki.fi>

* gmodule.rc.in gthread.rc.in: Move to corresponding subdirectories.

* Makefile.am
* gmodule/Makefile.am
* gthread/Makefile.am: Change accordingly.

* makefile.cygwin: Corresponding changes, some cleanup.

24 years agoNew files, for putting version info in the DLLs on Win32.
Tor Lillqvist [Fri, 4 Feb 2000 13:57:46 +0000 (13:57 +0000)]
New files, for putting version info in the DLLs on Win32.

2000-02-05  Tor Lillqvist  <tml@iki.fi>

* glib.rc.in gmodule.rc.in gthread.rc.in: New files, for putting
version info in the DLLs on Win32.

* Makefile.am: Generate corresponding *.rc files and distribute
them.

* makefile.cygwin.in: Add rules to automatically bump a "build
number" in the version info in the rc files each time the DLL is
built. But do this only for the person who releases binaries. If
others build the DLLs, the build number is set to zero.

24 years agoreturn stolen data from g_datalist_id_remove_no_notify() and
Tim Janik [Thu, 3 Feb 2000 20:43:00 +0000 (20:43 +0000)]
return stolen data from g_datalist_id_remove_no_notify() and

Fri Feb  4 19:36:05 2000  Tim Janik  <timj@gtk.org>

        * glib.h:
        * gdataset.c: return stolen data from g_datalist_id_remove_no_notify()
        and g_dataset_id_remove_no_notify() to avoid second lookup for common
        use.

24 years agoglib.h New functions for conversion between UTF-8 and the encoding
Tor Lillqvist [Wed, 2 Feb 2000 23:39:32 +0000 (23:39 +0000)]
glib.h New functions for conversion between UTF-8 and the encoding

2000-02-01  Tor Lillqvist  <tml@iki.fi>

* glib.h
* gstrfuncs.c (g_filename_to_utf8, g_filename_from_utf8): New
functions for conversion between UTF-8 and the encoding expected
by C runtime functions like open() and stat(), and returned by
readdir().

Implement them on Win32 where we use the system "ANSI" codepage,
which might be single-byte or double-byte. On Unix, just skip the
issue for now and provide dummy implementations that return a copy
of the argument.

* README.win32
* build-dll
* glib.def: Minor updates.

24 years agog_bit_nth_lsf/msf docs from Nils Rennebarth <nils@ipe.uni-stuttgart.de>,
Damon Chaplin [Thu, 27 Jan 2000 22:14:26 +0000 (22:14 +0000)]
g_bit_nth_lsf/msf docs from Nils Rennebarth <nils@ipe.uni-stuttgart.de>,

2000-01-25  Damon Chaplin  <damon@karuna.freeserve.co.uk>

        * tmpl/misc_utils.sgml: g_bit_nth_lsf/msf docs from
         Nils Rennebarth <nils@ipe.uni-stuttgart.de>, and updates for
        GVoidFunc & GFreeFunc.

24 years agos/current_time/dispatch_time/ for the dispatch() handlers. refetch the
Tim Janik [Wed, 26 Jan 2000 04:29:06 +0000 (04:29 +0000)]
s/current_time/dispatch_time/ for the dispatch() handlers. refetch the

Wed Jan 26 05:24:38 2000  Tim Janik  <timj@gtk.org>

        * glib.h:
        * gmain.c: s/current_time/dispatch_time/ for the dispatch() handlers.
        refetch the current time after invocation of poll() to cover up for
        the time spent in that function call.

24 years agoAdded G_N_ELEMENTS macro to determine the number of elements in an array.
Owen Taylor [Fri, 21 Jan 2000 15:27:39 +0000 (15:27 +0000)]
Added G_N_ELEMENTS macro to determine the number of elements in an array.

Fri Jan 21 10:18:24 2000  Owen Taylor  <otaylor@redhat.com>

* glib.h (G_N_ELEMENTS): Added G_N_ELEMENTS macro to determine
the number of elements in an array.

24 years ago[FIXME: On NetBSD, dlsym() doesn't set errno != 0 when the system cannot be
Martin Baulig [Thu, 13 Jan 2000 22:11:39 +0000 (22:11 +0000)]
[FIXME: On NetBSD, dlsym() doesn't set errno != 0 when the system cannot be
        found; this leads to an incorrect return value of g_module_symbol.]

2000-01-13  Martin Baulig  <martin@home-of-linux.org>

* gmodule.c (g_module_open): Check whether `check_init' is not NULL
before we attempt to call it.

24 years agocorrectly fetch the current locale, fix from owen.
Tim Janik [Sun, 9 Jan 2000 10:58:55 +0000 (10:58 +0000)]
correctly fetch the current locale, fix from owen.

Sun Jan  9 13:28:36 2000  Tim Janik  <timj@gtk.org>

        * gstrfuncs.c (g_strtod): correctly fetch the current locale,
        fix from owen.

25 years agoUse FormatMessage to translate system error codes into textual messages.
Tor Lillqvist [Thu, 16 Dec 1999 20:29:12 +0000 (20:29 +0000)]
Use FormatMessage to translate system error codes into textual messages.

1999-12-16  Tor Lillqvist  <tml@iki.fi>

* gmodule/gmodule-win32.c: Use FormatMessage to translate system
error codes into textual messages.

25 years agoRemoved G_THREADS_IMPL_NSPR. Shouldn't be used anyway.
Sebastian Wilhelmi [Thu, 2 Dec 1999 14:32:41 +0000 (14:32 +0000)]
Removed G_THREADS_IMPL_NSPR. Shouldn't be used anyway.

1999-12-02  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* glib-sections.txt, tmpl/threads.sgml: Removed
G_THREADS_IMPL_NSPR. Shouldn't be used anyway.

* glib-overrides.txt: New file, that makes most of the thread
related macros look like functions.

* Makefile.am: Added glib-overrides.txt to EXTRA_DIST.

* tmpl/threads.sgml: Minor updates.

25 years agoThis of course should return TRUE in a program with a thread-disabled
Sebastian Wilhelmi [Thu, 25 Nov 1999 13:47:55 +0000 (13:47 +0000)]
This of course should return TRUE in a program with a thread-disabled

1999-11-25  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* glib.h (G_TRYLOCK): This of course should return TRUE in a
program with a thread-disabled GLib.

25 years agog_strjoin was missing.
Tor Lillqvist [Thu, 18 Nov 1999 22:15:51 +0000 (22:15 +0000)]
g_strjoin was missing.

1999-11-18  Tor Lillqvist  <tml@iki.fi>

* glib.def: g_strjoin was missing.

25 years agoupdated .cvsignore
Sebastian Wilhelmi [Tue, 16 Nov 1999 12:30:49 +0000 (12:30 +0000)]
updated .cvsignore

25 years agoRenamed GLIB_SIZEOF_PTHREAD_T to GLIB_SIZEOF_SYSTEM_THREAD to reflect
Sebastian Wilhelmi [Tue, 16 Nov 1999 10:30:25 +0000 (10:30 +0000)]
Renamed GLIB_SIZEOF_PTHREAD_T to GLIB_SIZEOF_SYSTEM_THREAD to reflect

1999-11-16  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

        * acconfig.h, config.h.win32.in, configure.in: Renamed
        GLIB_SIZEOF_PTHREAD_T to GLIB_SIZEOF_SYSTEM_THREAD to reflect
        changed meaning.

        * configure.in: Cope with systems, that have a pthread_t type,
        that is not a pointer. Hint from Karl Nelson
        <kenelson@ece.ucdavis.edu>. Define GLIB_SIZEOF_SYSTEM_THREAD to 4
        for Solaris. Cope with systems, that have no default mutex
        initialize, like obviously most DCE systems.

        * glib.h, gthread.c: Changed the prototype of thread_create and
        thread_self to return the system thread into provided memory
        instead of a return value. This is necessary, as HPUX has a
        pthread_t, that is bigger than the biggest integral type there.

        * gthread.c: system_thread is no longer a pointer, but an memory
        area of size GLIB_SIZEOF_SYSTEM_THREAD. Changed the
        zeroinitialization and the tests for zeroness accordingly.

25 years agoChanged the prototype of thread_create and thread_self to return the
Sebastian Wilhelmi [Tue, 16 Nov 1999 10:29:27 +0000 (10:29 +0000)]
Changed the prototype of thread_create and thread_self to return the

1999-11-16  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

        * gthread-posix.c, gthread-solaris.c: Changed the prototype of
        thread_create and thread_self to return the system thread into
        provided memory instead of a return value. This is necessary, as
        HPUX has a pthread_t, that is bigger than the biggest integral
        type there. Made some more functions static.

        * gthread-posix.c: Small fixes for DCE threads: Detaching has to
        be done after thread creation for DCE.

25 years agoupdated .cvsignore
Sebastian Wilhelmi [Tue, 16 Nov 1999 10:11:45 +0000 (10:11 +0000)]
updated .cvsignore