Connect to the GtkComboBox::changed signal after all candidates are
added, to avoid calling e_source_config_check_complete() before the
candidate has been told to insert widgets. This can cause run-time
warnings such as:
(evolution:7106): evolution-cal-config-webcal-CRITICAL **:
cal_config_webcal_check_complete: assertion `context != NULL' failed
I've got UI freeze in a call of e_book_client_view_stop() on contact
store dispose, caused by synchronous D-Bus call. Doing the call
in a dedicated thread makes no UI freeze here.
nautilus-sendto has been stripped of its usefulness in GNOME 3.8.
It no longer has a UI of its own, it just immediately spawns a mail
client with a set of files as attachment arguments for a new message.
There's no reason for Evolution to be invoking it anymore.
Evolution is still occasionally getting stuck on shutdown, and although
the evolution-shell log domain shows debug messages for activities that
are preventing shutdown, they frequently look like this:
(evolution:13534): evolution-shell-DEBUG: 5 active 'mail' activities:
(evolution:13534): evolution-shell-DEBUG: * (no description)
(evolution:13534): evolution-shell-DEBUG: * (no description)
(evolution:13534): evolution-shell-DEBUG: * (no description)
(evolution:13534): evolution-shell-DEBUG: * (no description)
(evolution:13534): evolution-shell-DEBUG: * (no description)
I think the lack of descriptions is from CamelOperations popping all
their pushed messages, which is correct behavior but doesn't help us
debug the problem.
e_activity_get_last_known_text() returns the most recent _non-empty_
text value set on the EActivity. So our debug message can fall back
to that if the EActivity has no description at shutdown:
(evolution:13534): evolution-shell-DEBUG: * (was "blah, blah, blah")
gal_view_collection_new() now takes system and user directory arguments
and loads GalViews during instance initialization.
Removed functions:
gal_view_collection_set_storage_directories()
gal_view_collection_load()
gal_view_collection_loaded()
No longer needed. Instead, use g_type_ensure() to ensure the necessary
GalView subclasses are registered in the GType system before loading a
GalViewCollection. Best place to ensure types is from GClassInitFunc.
Given a type code string from an XML file, find the appropriate GType by
traversing the GType hierarchy from GAL_TYPE_VIEW and checking the class
structures for a matching type code string.
This completely eliminates the need for what's left of GalViewFactory.
Now it's just a matter of cleaning up the remains.
GalViewInstance and EMailPanedView need access. Currently they're just
using the public GalViewCollection members directly, but those are about
to be sealed up.
Just readable for now, but I plan to make it also construct-only.
(Despite the precendence, I prefer "user" over "local" for the name.)
New functions:
gal_view_collection_get_user_directory()
This is the GType of the GalView instances the factory creates.
Having this information will allow GalViewFactoryClass to do more itself
and be less reliant on subclasses.
All GalView subclasses return a static string in their get_type_code()
methods, so replace the method with a static string pointer directly in
the class structure, and remove the "type-code" GalView property.
Instead, lazily load the state file once an ETable or ETree is attached,
since we need its ETableSpecification to create an ETableState instance.
This means GalViewFactoryEtable can lose its ETableSpecification too.
e_table_specification_new() now takes a table specification filename and
a GError and parses the file as part of instance creation. If a file or
parse error occurs, e_table_specification_new() returns NULL.
This replaces e_table_specification_load_from_file().
New functions:
e_table_specification_get_filename()
Removed functions:
e_table_specification_load_from_file()
