2002-06-17 Sven Neumann <sven@gimp.org>
Obviously I disagree with Tim about the symbol_name parameter in
g_scanner_unexp_token(). As it seems it's not supposed to be used
to tell the user about the expected symbol (see bug #81850):
* gimpconfig.[ch] (gimp_config_deserialize_return): removed the
symbol_name parameter.
* app/config/gimpconfig-deserialize.c
* app/core/gimpcontainer.c
* app/core/gimpdocumentlist.c
* app/core/gimpparasitelist.c: changed the callers accordingly.
* app/core/gimpunits.c (gimp_unitrc_load): pass NULL as symbol_name
to g_scanner_unexp_token().
2002-05-31 Michael Natterer <mitch@gimp.org>
* app/config/gimpscanner.[ch]: configure the scanner the same
way GimpConfig did (scanner->user_data is, if set, a GError).
Added gimp_scanner_parse_color(). Added "GError **error" to
gimp_scanner_new().
* app/config/gimpconfig.c: use gimp_scanner_new().
* app/config/gimpconfig-deserialize.c: use gimp_scanner_parse_color().
* app/core/gimpunits.c
* app/plug-in/plug-in-rc.c: changed accordingly.
* app/plug-in/plug-in-rc.c: made the main parse loops consistent
with the other places using GScanner.
2002-05-27 Michael Natterer <mitch@gimp.org>
* app/config/gimpconfig.[ch]: added "gint nest_level" to the
deserialization functions, so nested calls to deserialize()
don't error on the trailing ')'. Made the scanner config parse
one-character identifiers and allow G_CSET_A_2_Z for all
identifiers.
Added gimp_config_deserialize_return() because returning the
correct return value from a deserialize() function is not
trivial any more with nested calls.
* app/config/gimpconfig-deserialize.[ch]
* app/config/gimprc.c
* app/core/gimpdocumentlist.c
* app/core/gimpparasitelist.c: use the new return value utility
function and made the main parsing loops simpler.
* app/core/gimpunits.c: made the main parsing loops consistent
with the stuff that uses GimpConfig.
* app/config/gimpconfig-deserialize.c
* app/config/gimpconfig-serialize.c: call the
(de)serialize_property() functions only if the property's class
implements the method itself (does *not* inherit it from one of
it's parents).
* app/core/gimpcontainer.c: implemented deserialization of
containers. For each child entry encountered in the input, check
if it's already contained in the container and create it if not.
If a "gimp" pointer is passed as user_data pass it as construct
property to g_object_new() when creating the object.
* app/core/gimpcontext.c: changed deserialization of brush,
pattern etc. to honor "no_data".
* app/widgets/gimpdeviceinfo.c: added a "gimp" construct_only
property which overrides GimpContext's "gimp" property. Moved all
initialisation code from gimp_device_info_new() to
gimp_device_info_set_property(PROP_GIMP) so it is properly
performed if the object is created by GimpContainer's
deserialize() implementation.
* app/widgets/gimpdevices.c: made gimp_devices_restore_test() work.
2002-05-15 Sven Neumann <sven@gimp.org>
* app/core/gimpdocumentlist.c
* app/core/gimpparasitelist.c
* app/core/gimpunits.c: improved parser error messages. Relies on a
fix in glib (see bug #81850). However it doesn't hurt if you don't
have the fix.
2002-03-28 Michael Natterer <mitch@gimp.org>
* app/config/Makefile.am
* app/config/gimpscanner.[ch]: new files containing a convenience
constructor for GScanner and some parse utility functions, mostly
cut out of app/plug-in/plug-in-rc.c
* app/plug-in/plug-in-rc.c: removed the stuff here, added scopes
to the scanner symbols.
* app/core/Makefile.am
* app/core/gimpunits.[ch]: moved the user_unit list handling and
unitrc stuff to this file. Parse unitrc here, using the new
utility functions instead of using gimprc.
* app/core/gimpunit.[ch]: removed here.
* app/core/gimpdocuments.c: same here: added a scanner for the
document history instead of using gimprc.
* app/gimprc.c: removed unitrc and document history parsing stuff
along with some old unused cruft.
* app/app_procs.c
* app/core/gimp.c
* app/gui/user-install-dialog.c: #include "core/gimpunits.h".
* app/core/gimpdrawable-bucket-fill.c: don't include "gimprc.h".
Sven Neumann