glib/NEWS

204 lines
8.2 KiB
Plaintext
Raw Normal View History

What's new in GLib 1.2.0 (since GLib 1.0.x):
* GLib is distributed seperatedly from Gtk+
* Win32 Portability
* Threading support
* GModule mechanism (implemented in an extra library) which wraps dynamic
object code loading facilities in a portable manner
* GIOChannel structure to encapsulate the IPC mechanism
* GQuarks and datasets
* GRelations for n-way mapping of certain data
* An n-way tree implementation
* GDate functionality for calendar date manipulations
* GAllocator type and assocated functions
* Added generic callback maintenance functions (ghook)
* Generic functions for TAB completions
* Endian defines (G_*_ENDIAN)
* g_log() mechanism for logging of messages at different log levels
* Generic main-loop mechanism
* New glib-config helper script
* Many more API extensions
Overview of Changes in GLib 1.1.16:
* Allocate smaller pools of memory for glists, gslists, gnodes
* Bug Fixes
Overview of Changes in GLib 1.1.15:
* HPUX 11 thread system detection should now work
* Release the main loop lock around calls to prepare() and
check() so it is not held over user code
* A few Win32 fixups
Overview of Changes in GLib 1.1.14:
* Check for dlsym() in system libraries and -dl
* FreeBSD portability fixes
* Random bug fixes and autoconf/automake changes
Overview of Changes in GLib 1.1.13:
* Removed alloca() based function and macro variants again.
* Improved thread related configure tests.
* GSource destruction fixups.
* Fixed up idle function removal based on user_data pointer.
* Advanced Win32 portability.
* Enforced GSource's check(), prepare() and dispatch() constrains,
loop recursions may only happen from dispatch(), and check() as well
as prepare() are called while the main_loop lock is being held.
* GLib development now requires GNU autoconf 2.13, GNU automake 1.4
and GNU libtool 1.2d.
* Lots of random portability and bug fixes.
Overview of Changes in GLib 1.1.12:
* Added alloca functions/macros: g_strdup_a, g_strconcat3_a, g_alloca,
g_new_a, g_new0_a
* New tests structure. Type 'make check' on your system to run them.
* Avoid unnecessary extra hook referencing in g_hook_list_marshal
Overview of Changes in GLib 1.1.11:
* provide defaults for POLL sysdefs
* g_main_is_running: new function to check whether a main loop has been quitted
* a few other enhancement/fixes
Overview of Changes in GLib 1.1.9:
* Check for pthread_attr_init in all cases, Digital Unix 4 requires this
* For G_LOCK_DECLARE_*, if !G_THREADS_ENABLED, eat trailing semicolon better
* Changed g_main_poll_(add|remove) to g_main_(add|remove)_poll
Overview of Changes in GLib 1.1.8:
* Added threading support
- The ability to specify a set of functions to be used for
locking at runtime.
- Default implementations of locking functions for pthreads,
Solaris threads, and (experimentally) NSPR.
- All static variables should now properly locked.
- Enhancements to the generic main-loop mechanism to be thread-safe.
(It is used for the main-loop in GTK+ as of GTK+-1.1.8)
* Portability fixes.
Overview of Changes in GLib 1.1.7:
* Removed multiple define from glibconfig.h
Overview of Changes in GLib 1.1.6:
* New GDate functionality for calendar date manipulations (g_date_*)
* New GAllocator type and assocated functions
* New functions g_slist_copy and g_list_copy to duplicate a list with all
its data pointers.
* New function g_array_insert_vals and new macro g_array_insert_val to
insert elements at an arbitrary index
* GAllocators used for glist, gslist, gnode node allocations
* Incremental freezing in ghash
* New function g_hook_list_marshal_check to eventually destroy hooks after
they got marshalled
* Revised GIOChannel to provide generic virtual-function based interface
* Added generic main-loop abstraction
* Removed GListAllocator type and its g_*_allocator_*() function variants
* Bug fixes
Overview of Changes in GLib 1.1.5:
* Win32 portability
* GIOChannel structure to encapsulate the IPC mechanism
* Reimplemented endian stuff, using inline asm for x86
* New functions:
- g_strescape: escapes backslashes
- g_path_is_absolute and g_path_skip_root
- g_getenv: expands environment variables that contain references
to other environment variables
- g_scanner_sync_file_offset: rewind the filedescriptor to the current
buffer position and blow the file read ahead buffer
- g_array_remove_index: remove an entry, preserving the order
- g_array_remove_index_fast: remove an entry, order might be distorted
- g_ptr_array_remove: remove an entry, preserving the order
- g_ptr_array_remove_fast: remove an entry, order might be distorted
- g_byte_array_remove_index: wrapper for g_array_remove_index
- g_byte_array_remove_index_fast: wrapper for g_array_remove_index_fast
- g_strncasecmp: modeled closely after g_strcasecmp
- g_list_sort, g_slist_sort: to merge sort GLists and GSLists
* New macros:
- G_DIR_SEPARATOR, G_DIR_SEPARATOR_S: platform-dependant file name
syntax elements
- G_SEARCHPATH_SEPARATOR, G_SEARCHPATH_SEPARATOR_S: platform-dependant
search path syntax conventions
- G_STRUCT_OFFSET, G_STRUCT_MEMBER_P, G_STRUCT_MEMBER: for handling
structure fields through their offsets
* Removed G_ENUM, G_FLAGS, G_NV, and G_SV macros
* Bug fixes
Overview of Changes in GLib 1.1.4:
* Added generic callback maintenance functions (ghook)
* New endian defines (G_*_ENDIAN)
* New string join/split/free routines
* Fixes
version bump to 1.1.3, binary age 0, interface age 0. Sun Aug 16 20:28:27 1998 Tim Janik <timj@gtk.org> * version bump to 1.1.3, binary age 0, interface age 0. * glib.h: be nice to platforms that don't have gint64 and don't issue #warning on every compilation. since glib doesn't require gint64 itself, packages that need gint64 should test for this themselves. * glib.h: * gutils.c: added a new function g_vsnprintf(). Fri Aug 14 16:41:53 1998 Tim Janik <timj@gtk.org> * glib.h: added static inline functions for bit mask tests: g_bit_nth_lsf, g_bit_nth_msf and g_bit_storage. Fri Aug 13 14:23:37 1998 Tim Janik <timj@gtk.org> * glib.h: * gmessages.c: revised the message handling system, which is now based on a new mechanism g_log*. most of the assertment macros got adapted to feature the new g_log() call with an additional specification of the log level in a preprocessor macro G_LOG_DOMAIN. if G_LOG_DOMAIN is undefined upon the includion of glib.h, it'll be defined with a value of (NULL) and thus preserves the original bahaviour for warning and error messages. the message handler setting functions for g_warning, g_error and g_message are only provided for backwards compatibility and might get removed somewhen. * Makefile.am: feature the G_LOG_DOMAIN macro to set the log domain to "GLib" upon compilation. we currently have to add this definition to the DEFS variable. * testglib.c: we need an ugly #undef G_LOG_DOMAIN at the start of this file currently, since automake doesn't support per target _CFLAGS yet. * glib.h: changed some gints to gbooleans, made a few const corrections, removed some superfluous G_STMT_START{}G_STMT_END wrappers, added some in other required places. * gnode.c: (g_node_prepend): (g_node_insert_before): (g_node_insert): (g_node_append_data): (g_node_prepend_data): (g_node_insert_data_before): (g_node_insert_data): (g_node_append): return (node), so these macros/functions can be usefully chained with g_node_new(). [GModule] Fri Aug 14 02:24:39 1998 Tim Janik <timj@gtk.org> * Makefile.am: feature the G_LOG_DOMAIN macro to set the log domain to "GModule" upon compilation. we currently have to add this definition to the DEFS variable. * testgmodule.c: we need an ugly #undef G_LOG_DOMAIN at the start of this file currently, since automake doesn't support per target _CFLAGS yet.
1998-08-16 23:14:11 +02:00
Overview of Changes in GLib 1.1.3:
* New GModule mechanism (implemented in an extra library) which wraps dynamic
object code loading facilities in a portable manner.
* glib-config features extra "glib" (old behaviour remains) and "gmodule"
(add libgmodule.so to the --libs output) arguments now. this can also
be specified as fourth argument to the AM_PATH_GLIB() macro.
* Overhaul of the `inline' autoconfiguration stuff, so inlining should be
sufficiently supported on all systems that allow inlining now.
version bump to 1.1.3, binary age 0, interface age 0. Sun Aug 16 20:28:27 1998 Tim Janik <timj@gtk.org> * version bump to 1.1.3, binary age 0, interface age 0. * glib.h: be nice to platforms that don't have gint64 and don't issue #warning on every compilation. since glib doesn't require gint64 itself, packages that need gint64 should test for this themselves. * glib.h: * gutils.c: added a new function g_vsnprintf(). Fri Aug 14 16:41:53 1998 Tim Janik <timj@gtk.org> * glib.h: added static inline functions for bit mask tests: g_bit_nth_lsf, g_bit_nth_msf and g_bit_storage. Fri Aug 13 14:23:37 1998 Tim Janik <timj@gtk.org> * glib.h: * gmessages.c: revised the message handling system, which is now based on a new mechanism g_log*. most of the assertment macros got adapted to feature the new g_log() call with an additional specification of the log level in a preprocessor macro G_LOG_DOMAIN. if G_LOG_DOMAIN is undefined upon the includion of glib.h, it'll be defined with a value of (NULL) and thus preserves the original bahaviour for warning and error messages. the message handler setting functions for g_warning, g_error and g_message are only provided for backwards compatibility and might get removed somewhen. * Makefile.am: feature the G_LOG_DOMAIN macro to set the log domain to "GLib" upon compilation. we currently have to add this definition to the DEFS variable. * testglib.c: we need an ugly #undef G_LOG_DOMAIN at the start of this file currently, since automake doesn't support per target _CFLAGS yet. * glib.h: changed some gints to gbooleans, made a few const corrections, removed some superfluous G_STMT_START{}G_STMT_END wrappers, added some in other required places. * gnode.c: (g_node_prepend): (g_node_insert_before): (g_node_insert): (g_node_append_data): (g_node_prepend_data): (g_node_insert_data_before): (g_node_insert_data): (g_node_append): return (node), so these macros/functions can be usefully chained with g_node_new(). [GModule] Fri Aug 14 02:24:39 1998 Tim Janik <timj@gtk.org> * Makefile.am: feature the G_LOG_DOMAIN macro to set the log domain to "GModule" upon compilation. we currently have to add this definition to the DEFS variable. * testgmodule.c: we need an ugly #undef G_LOG_DOMAIN at the start of this file currently, since automake doesn't support per target _CFLAGS yet.
1998-08-16 23:14:11 +02:00
* New g_log() mechanism for logging of messages at different log levels,
associated with certain log domains (define -DG_LOG_DOMAIN for your library).
removed this function which was not publically exported in glib.h. to Mon Aug 24 02:08:56 1998 Tim Janik <timj@gtk.org> * glib.h: * gstring.c: * gstrfuncs.c: (g_vsprintf): removed this function which was not publically exported in glib.h. to export it, it should have been named differently in the first place, since its semantics differ from vsprintf(). apart from that, it was a possible cause for problems since it worked on a previously allocated memory area and was used in a lot places of glib. exporting it would have been a guararant for problems with threaded programs. (g_printf_string_upper_bound): exported this function to return a string size, guarranteed to be big enough to hold the fully expanded format+args string. added 'q', 'L' and 'll' flag handling. in fact, the newly allocated area is in most cases much bigger than required. (g_strdup_vprintf()): new function returning a newly allocated string containing the contents of *format and associated args (size is calculated with g_printf_string_upper_bound()). (g_strdup_printf): new function which wraps g_strdup_vprintf(). * configure.in: check for va_copy() or __va_copy() alternatively. check whether va_lists can be copyied by value. * glib.h: provide a definition for G_VA_COPY. * glib.h: * gmessages.c: (g_logv): (g_vsnprintf): pass va_lists by value, not by reference, since this causes problems on platforms that implement va_list as as arrays. internaly, use G_VA_COPY (new_arg, org_arg); va_end (new_arg); to produce a second va_list variable, if multiple passes are required. changed all callers. * glib.h: * gerror.h: renamed g_debug() to g_on_error_query(), cleaned up a bit. renamed g_stack_trace() to g_on_error_stack_trace() since both functions cluttered different namespaces. there is an appropriate comment in glib.h now that explains the unix and gdb specific dependencies of both functions. removed g_attach_process(). g_on_error_stack_trace() should probably be handled with caution, i've seem several different linux versions (2.0.x) become unstable after invokation of this function.
1998-08-24 07:26:53 +02:00
* New inline functions for bit masks tests.
* GNode macros (and functions) now return the newly allocated node.
removed this function which was not publically exported in glib.h. to Mon Aug 24 02:08:56 1998 Tim Janik <timj@gtk.org> * glib.h: * gstring.c: * gstrfuncs.c: (g_vsprintf): removed this function which was not publically exported in glib.h. to export it, it should have been named differently in the first place, since its semantics differ from vsprintf(). apart from that, it was a possible cause for problems since it worked on a previously allocated memory area and was used in a lot places of glib. exporting it would have been a guararant for problems with threaded programs. (g_printf_string_upper_bound): exported this function to return a string size, guarranteed to be big enough to hold the fully expanded format+args string. added 'q', 'L' and 'll' flag handling. in fact, the newly allocated area is in most cases much bigger than required. (g_strdup_vprintf()): new function returning a newly allocated string containing the contents of *format and associated args (size is calculated with g_printf_string_upper_bound()). (g_strdup_printf): new function which wraps g_strdup_vprintf(). * configure.in: check for va_copy() or __va_copy() alternatively. check whether va_lists can be copyied by value. * glib.h: provide a definition for G_VA_COPY. * glib.h: * gmessages.c: (g_logv): (g_vsnprintf): pass va_lists by value, not by reference, since this causes problems on platforms that implement va_list as as arrays. internaly, use G_VA_COPY (new_arg, org_arg); va_end (new_arg); to produce a second va_list variable, if multiple passes are required. changed all callers. * glib.h: * gerror.h: renamed g_debug() to g_on_error_query(), cleaned up a bit. renamed g_stack_trace() to g_on_error_stack_trace() since both functions cluttered different namespaces. there is an appropriate comment in glib.h now that explains the unix and gdb specific dependencies of both functions. removed g_attach_process(). g_on_error_stack_trace() should probably be handled with caution, i've seem several different linux versions (2.0.x) become unstable after invokation of this function.
1998-08-24 07:26:53 +02:00
* New macro G_VA_COPY() to work around va_list copying oddities on some
platforms. the non-static g_vsprintf() function vanished in favour of
a publically exported g_strdup_vprintf().
People that used the former g_vsprintf() would definitely want to read the
associated ChangeLog entries (grep for printf).
* New utility functions:
g_strndup(), g_on_error_query(), g_on_error_stack_trace(), g_strdup_printf(),
g_strdup_vprintf(), g_printf_string_upper_bound(), g_spaced_primes_closest(),
g_strnfill(), g_memdup().
* Overhaul of the array implementations, this contains some source incompatible
changes. Again, the ChangeLog is much more informative (grep for garray.c).
* The internals of the g_dataset mechanism are now exported through the
new g_datalist_* API (this is also the underlying implementation for the
keyed data of GtkObjects).
* New function g_atexit(), use of the ATEXIT() macro is discouraged.
* Better configure checks for ansi compliance.
* Libtool update to version 1.2b.
removed this function which was not publically exported in glib.h. to Mon Aug 24 02:08:56 1998 Tim Janik <timj@gtk.org> * glib.h: * gstring.c: * gstrfuncs.c: (g_vsprintf): removed this function which was not publically exported in glib.h. to export it, it should have been named differently in the first place, since its semantics differ from vsprintf(). apart from that, it was a possible cause for problems since it worked on a previously allocated memory area and was used in a lot places of glib. exporting it would have been a guararant for problems with threaded programs. (g_printf_string_upper_bound): exported this function to return a string size, guarranteed to be big enough to hold the fully expanded format+args string. added 'q', 'L' and 'll' flag handling. in fact, the newly allocated area is in most cases much bigger than required. (g_strdup_vprintf()): new function returning a newly allocated string containing the contents of *format and associated args (size is calculated with g_printf_string_upper_bound()). (g_strdup_printf): new function which wraps g_strdup_vprintf(). * configure.in: check for va_copy() or __va_copy() alternatively. check whether va_lists can be copyied by value. * glib.h: provide a definition for G_VA_COPY. * glib.h: * gmessages.c: (g_logv): (g_vsnprintf): pass va_lists by value, not by reference, since this causes problems on platforms that implement va_list as as arrays. internaly, use G_VA_COPY (new_arg, org_arg); va_end (new_arg); to produce a second va_list variable, if multiple passes are required. changed all callers. * glib.h: * gerror.h: renamed g_debug() to g_on_error_query(), cleaned up a bit. renamed g_stack_trace() to g_on_error_stack_trace() since both functions cluttered different namespaces. there is an appropriate comment in glib.h now that explains the unix and gdb specific dependencies of both functions. removed g_attach_process(). g_on_error_stack_trace() should probably be handled with caution, i've seem several different linux versions (2.0.x) become unstable after invokation of this function.
1998-08-24 07:26:53 +02:00
* Lotsa bug fixes and cleanups as always ;)
version bump to 1.1.3, binary age 0, interface age 0. Sun Aug 16 20:28:27 1998 Tim Janik <timj@gtk.org> * version bump to 1.1.3, binary age 0, interface age 0. * glib.h: be nice to platforms that don't have gint64 and don't issue #warning on every compilation. since glib doesn't require gint64 itself, packages that need gint64 should test for this themselves. * glib.h: * gutils.c: added a new function g_vsnprintf(). Fri Aug 14 16:41:53 1998 Tim Janik <timj@gtk.org> * glib.h: added static inline functions for bit mask tests: g_bit_nth_lsf, g_bit_nth_msf and g_bit_storage. Fri Aug 13 14:23:37 1998 Tim Janik <timj@gtk.org> * glib.h: * gmessages.c: revised the message handling system, which is now based on a new mechanism g_log*. most of the assertment macros got adapted to feature the new g_log() call with an additional specification of the log level in a preprocessor macro G_LOG_DOMAIN. if G_LOG_DOMAIN is undefined upon the includion of glib.h, it'll be defined with a value of (NULL) and thus preserves the original bahaviour for warning and error messages. the message handler setting functions for g_warning, g_error and g_message are only provided for backwards compatibility and might get removed somewhen. * Makefile.am: feature the G_LOG_DOMAIN macro to set the log domain to "GLib" upon compilation. we currently have to add this definition to the DEFS variable. * testglib.c: we need an ugly #undef G_LOG_DOMAIN at the start of this file currently, since automake doesn't support per target _CFLAGS yet. * glib.h: changed some gints to gbooleans, made a few const corrections, removed some superfluous G_STMT_START{}G_STMT_END wrappers, added some in other required places. * gnode.c: (g_node_prepend): (g_node_insert_before): (g_node_insert): (g_node_append_data): (g_node_prepend_data): (g_node_insert_data_before): (g_node_insert_data): (g_node_append): return (node), so these macros/functions can be usefully chained with g_node_new(). [GModule] Fri Aug 14 02:24:39 1998 Tim Janik <timj@gtk.org> * Makefile.am: feature the G_LOG_DOMAIN macro to set the log domain to "GModule" upon compilation. we currently have to add this definition to the DEFS variable. * testgmodule.c: we need an ugly #undef G_LOG_DOMAIN at the start of this file currently, since automake doesn't support per target _CFLAGS yet.
1998-08-16 23:14:11 +02:00
Overview of Changes in GLib 1.1.2:
* Fixed packaging mistake which occured in 1.1.1
* fix 64-bitness in g_prints in glibtest
What is new in GLib 1.1.1:
* An n-way tree implementation is provided now, based on the GNode structure.
* Bugfix for pointer arrays.
1998-07-27 03:56:45 +02:00
What is new in GLib 1.1.0:
* GLib is distributed seperatedly from Gtk+ now and uses a sophisticated
shared library versioning scheme to deal with interface and binary
incompatibilities.
* There is a glib-config helper script installed now.
* Fixups all over the place.
* gboolean is now a gint, not a gchar anymore.
* API extensions for GList and GSList.
* New g_str*() functions for simple string handling.
* GScanner extensions for scope, warning and error handling.
* Minor performance improvements for GMemChunks.
* Implementations of GQuarks and datasets (similar to GtkObjects data
mechansim, but works for generic memory locations).
* More convenience macros for GNU C function arguments.
* Const correction all over the place, including a new pointer type
gconstpointer.
* Generic functions for TAB completions.
* GRelations for n-way mapping of certain data.