1 2008-10-08 Christian Persch <chpe@gnome.org>
3 Bug 555313 – GFileAttribute boxed type get_type function should
4 use usual get_type pattern
6 * gio/gfileattribute.c (g_file_attribute_info_list_get_type): Use
7 g_once_init_enter/leave.
9 2008-10-08 Christian Persch <chpe@gnome.org>
11 Bug 555314 – mem leak in gmarkup
13 * glib/gmarkup.c: (g_markup_parse_context_parse): Plug a mem leak.
15 2008-10-08 Tor Lillqvist <tml@novell.com>
17 Bug 554790 - g_convert() misbehaves with winiconv versions
19 * glib/win_iconv.c (kernel_mbtowc): If converting from ASCII,
20 explicitly check for and reject 8bit chars. MultiByteToWideChar()
21 doesn't, at least not on XP.
23 2008-10-06 Matthias Clasen <mclasen@redhat.com>
25 * glib/gtypes.h: Properly include gmacros.h
27 2008-10-06 Behdad Esfahbod <behdad@gnome.org>
29 Bug 555309 – giochannel breaks on error
30 Patch from Christian Persch
32 * glib/giounix.c (g_io_unix_read), (g_io_unix_write),
33 (g_io_unix_seek), (g_io_unix_close), (g_io_unix_set_flags):
34 Don't shadow err. Oops!
36 2008-10-06 Christophe Fergeau <teuf@gnome.org>
38 Bug 555224 – Improve g_format_size_for_display doc
40 * glib/gfileutils.c: change g_format_size_for_display API doc to
41 explicitly say that the returned string has to be freed. Change
42 spelling of "newly allocated" to "newly-allocated" in g_file_read_link
43 API doc to be more consistent with what is done in that file.
45 2008-10-01 David Zeuthen <davidz@redhat.com>
47 * README.in: Add "Notes about glib 2.20" section detailing the
48 ramifications of the patch from bug #528670.
50 2008-09-30 Behdad Esfahbod <behdad@gnome.org>
52 Bug 554092 – glib doesn't return G_FILE_ERROR_NOENT et al on OS X
54 * glib/giounix.c (g_io_unix_read), (g_io_unix_write),
55 (g_io_unix_seek), (g_io_unix_close), (g_io_unix_set_flags),
56 (g_io_unix_get_flags), (g_io_channel_new_file):
57 Like mclasen says: "well, thats the way errno works...,
58 save it or loose it". Save errno.
60 2008-09-30 Tor Lillqvist <tml@novell.com>
63 * glib/tests/Makefile.am: Bypass gtester related stuff on Windows.
65 2008-09-30 Tor Lillqvist <tml@novell.com>
68 * glib/gnulib/vasnprintf.c: Don't define _GNU_SOURCE on Windows,
69 as _GNU_SOURCE has unintended side effects when compiling against
72 2008-09-26 Dan Winship <danw@gnome.org>
74 Bug 553447 – g_assert_no_error()
76 * glib/gtestutils.h (g_assert_no_error, g_assert_error): Macros to
77 assert that a GError is not set, or else is set to a particular
80 * glib/gtestutils.c (g_assertion_message_error): utility for
83 * glib/tests/keyfile.c:
84 * tests/asyncqueue-test.c:
85 * tests/bookmarkfile-test.c:
86 * tests/convert-test.c:
87 * tests/file-test.c: Use g_assert_error/g_assert_no_error
89 2008-09-26 Dan Winship <danw@gnome.org>
91 * glib/gthreadpool.c (wakeup_thread_marker): make this a "const
92 gpointer" rather than a gconstpointer to avoid warnings later
94 * glib/pcre/pcre_ucp_searchfuncs.c:
95 * glib/pcre/pcre_valid_utf8.c: #include "config.h"
97 * glib/tests/printf.c (test_d): fool gcc into not warning about
98 some printf format strings that we know are dubious
100 2008-09-26 Matthias Clasen <mclasen@redhat.com>
102 Bug 553857 – gbacktrace.h requires signal.h
104 * glib/gbacktrace.h: Include signal.h for raise().
105 Pointed out by Sebastien Bacher
107 2008-09-26 Matthias Clasen <mclasen@redhat.com>
109 Bug 553724 – python interpretter path not patched in correctly
111 * glib/Makefile.am: Fix the sed magic to replace python.
113 2008-09-26 Matthias Clasen <mclasen@redhat.com>
115 * glib/gmain.c: Add some more docs.
117 * glib/giochannel.c: Move more docs inline, and improve them
120 2008-09-25 Tor Lillqvist <tml@novell.com>
122 Bug 553820 - gpoll.c: undeclared identifier
125 * glib/gpoll.c: Make the g_poll() function non-static also on
126 Windows. Prefix an underscore to the g_main_poll_debug variable
127 and make it non-static in gmain.c so that it can be used in
128 gpoll.c. Add back missing variable declaration.
130 2008-09-25 Tor Lillqvist <tml@novell.com>
132 * glib/gspawn-win32.c (do_spawn_with_pipes) (do_spawn_directly):
133 Just ignore the child_setup function, never call it. The is no
134 situation in which it could be useful on Windows. Do print a
135 warning, like before.
137 * glib/gspawn.c (g_spawn_async_with_pipes): Corresponding change
140 2008-09-24 Sven Herzberg <sven@imendio.com>
142 Be a little more explcit in the docs. Includes Owen's requested
145 * glib/gmain.c: improved documentation for g_source_attach() and
148 2008-09-23 Michael Natterer <mitch@imendio.com>
150 * glib/glib.h: #include <glib/gpoll.h>
152 * glib/gpoll.h: #error out if gpoll.h is included directly.
154 * glib/gpoll.c: remove trailing whitespace.
156 2008-09-23 Dan Winship <danw@gnome.org>
158 * glib/gpoll.[ch] (g_poll): Move this out of gmain.c and make it part
159 of the public API. (Part of Bug 505361 - gunixinputstream.c assumes
162 2008-09-23 Tor Lillqvist <tml@novell.com>
164 * glib/gmain.c (poll_rest) [Win32]: Fix embarrassing bug: I was
165 passing an incorrect third parameter to memmove(), had forgotten
166 to multiply by the size of the table entry. Just use a for loop
167 instead, clearer. Odd I didn't notice when testing this code.
169 2008-09-19 Hans Petter Jansson <hpj@novell.com>
171 Rewrite most of GHashTable to use open addressing with quadratic
172 probing instead of chaining. This has the potential to reduce memory
173 fragmentation significantly, while being slightly faster due to
174 better locality and no need to call alloc/free functions for nodes.
175 Benchmarks suggest it also uses less memory overall.
177 * glib/ghash.c (prime_mod): Table of suitable primes for
178 initial-probe distribution.
179 (g_hash_table_set_shift): New function.
180 (g_hash_table_find_closest_shift): New function.
181 (g_hash_table_set_shift_from_size): New function.
182 (g_hash_table_lookup_node_for_insertion): New function.
183 (g_hash_table_lookup_node): Rewritten to return node index instead of
184 pointer, use quadratic probe on flat table, and not return insertion
185 data. The latter saves some computation for read-only lookups.
186 (g_hash_table_remove_node): Rewrite to take a pointer directly to the
187 node structure to remove, and clear that. Remove unlinking code.
188 (g_hash_table_remove_all_nodes): Rewrite to not clear nodes
189 individually, but en masse using memset () after potentially calling
191 (iter_remove_or_steal): Use new data structure and algorithm. Vastly
192 simplified - now just a call to g_hash_table_remove_node ().
193 (g_hash_table_resize): New resize code, re-indexing with new prime
194 and cleaning up tombstones.
195 (g_hash_table_maybe_resize): Table may hold 8 buckets minimum, no less
196 than 1/4 load excluding tombstones, and no more than 15/16 load
197 including tombstones. These numbers are the results of a lot of
198 benchmarking with multiple complex applications, and should not be
200 (g_hash_table_iter_next)
201 (g_hash_table_lookup)
202 (g_hash_table_lookup_extended)
203 (g_hash_table_insert_internal)
204 (g_hash_table_remove_internal)
205 (g_hash_table_foreach_remove_or_steal)
206 (g_hash_table_foreach)
208 (g_hash_table_get_keys)
209 (g_hash_table_get_values): Use new data structure and algorithm,
210 fairly trivial changes.
212 2008-09-19 Tor Lillqvist <tml@novell.com>
214 * glib-zip.in: Look for man pages in share/man.
216 * glib/gutils.c (_glib_get_dll_directory)
217 * glib/gspawn-win32.c (do_spawn_with_pipes): Be a bit less
218 restrictive, look for the helper programs in the same folder where
219 the GLib DLL is, not necessarily in a "bin" subfolder of the top
220 GLib installation folder.
222 2008-09-18 Matthias Clasen <mclasen@redhat.com>
224 * configure.in: Bump version to 2.19.0
226 * ChangeLog.pre-2-18: rotate ChangeLog
228 * === branch for 2.18 ===