| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
| |
The CCode attribute 'cname' needs to be the canonical representation as it
is expected in C.
https://bugzilla.gnome.org/show_bug.cgi?id=731547
|
|
|
|
|
| |
g_list_free_full and g_slist_free_full are available since 2.28.
g_queue_free_full is available since 2.32.
|
|
|
|
|
|
|
| |
The next stable release of vala will happen 3.5 years after the
release of GLib 2.40.0.
https://bugzilla.gnome.org/show_bug.cgi?id=782005
|
|
|
|
|
|
| |
Additionally make sure types derived from string are freed, e.g. ObjectPath
https://bugzilla.gnome.org/show_bug.cgi?id=782719
|
|
|
|
|
|
|
| |
The Method.yield_count is not correct because in C the finally blocks may
be emitted twice.
https://bugzilla.gnome.org/show_bug.cgi?id=741929
|
|
|
|
|
|
|
|
| |
initializer"
This reverts commit f5c654f76eeb1ee887f3d89eb701cf82f75a430a.
https://bugzilla.gnome.org/show_bug.cgi?id=782056
|
|
|
|
|
|
|
| |
Ignoring the warning of uncaught error in async methods resulted in
leaking a reference to the scopes GLib.AsyncResult.
https://bugzilla.gnome.org/show_bug.cgi?id=641171
|
| |
|
|
|
|
| |
Reduces the size of generated c-sources while avoiding redefinitions.
|
| |
|
|
|
|
| |
valac is the only consumer and will statically link these libraries.
|
|
|
|
|
|
| |
Regression of 6d07669384cdb70c3c657dba67d5048212f25da9
https://bugzilla.gnome.org/show_bug.cgi?id=780426
|
|
|
|
|
|
| |
Set delegate-target to "self" which is available in *_instance_init().
https://bugzilla.gnome.org/show_bug.cgi?id=683925
|
| |
|
|
|
|
|
|
|
| |
Nullable value-types are actually pointers to heap-allocated structures.
Therefore a pointer-based marshaller is required for those types.
https://bugzilla.gnome.org/show_bug.cgi?id=758816
|
|
|
|
|
|
| |
g_cclosure_marshal_VOID__BOXED
g_cclosure_marshal_VOID__VARIANT
g_cclosure_marshal_BOOLEAN__BOXED_BOXED
|
|
|
|
|
|
| |
This is check was introduced with 64b9bfc1bc0abfed45ad07a8ebaef8a5f167f848
https://bugzilla.gnome.org/show_bug.cgi?id=779955
|
|
|
|
|
| |
Avoid criticals while unconditionaly trying to chain-up to non-existing
property-accessors without checking for null.
|
|
|
|
|
|
| |
Based on patch by gandalfn
https://bugzilla.gnome.org/show_bug.cgi?id=764481
|
| |
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=779609
|
|
|
|
|
|
| |
Fixes client-side dbus generation on methods that don't throw any error.
https://bugzilla.gnome.org/show_bug.cgi?id=779652
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=710103
|
|
|
|
|
|
|
| |
If there are arguments that need freeing, we need to fall through the
_error label, if there is any.
https://bugzilla.gnome.org/show_bug.cgi?id=778540
|
|
|
|
|
|
|
| |
Fixes the _error label from being set at the end of a compound statement
if there are no values to unref.
https://bugzilla.gnome.org/show_bug.cgi?id=778540
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=778632
|
|
|
|
|
|
|
|
|
|
| |
The elements of current_required_app_classes end up in the generated
code in g_type_ensure() statements. The order of these can vary
between builds when using a HashSet, breaking reproducibilty.
See https://wiki.debian.org/ReproducibleBuilds
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=802528
|
|
|
|
|
|
|
| |
This make it match vala's generated code of async-module.
To properly account for non-void *_async methods we require some additional
metadata since only the return-type of *_finish method is known.
|
| |
|
| |
|
| |
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=777958
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=777958
|
|
|
|
|
|
|
|
|
| |
Add test-case and fix code generator for call to the setter to pass
instance as a value rather than as a pointer to value.
Based on patches by Nathan Summers.
https://bugzilla.gnome.org/show_bug.cgi?id=657346
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=673389
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=778993
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Commit 09e6818d01 changed dbus server paths so requests fell back
to freeing allocated memory on errors. However certain kinds of
dbus replies issue a final g_dbus_connection_send_message() there.
In case of errors, _reply_message will be NULL, and the error paths
would have already issued g_dbus_method_invocation_return_gerror(),
so the g_dbus_connection_send_message() call is both unnecessary and
warns on the NULL argument. So move the DBus reply before the _error
label.
https://bugzilla.gnome.org/show_bug.cgi?id=778540
|
|
|
|
|
|
|
|
| |
Treat methods with non-void return-type special while they require a return
statement. This does not account for a trailing superfluous "return" which
will still trigger this code-duplication.
https://bugzilla.gnome.org/show_bug.cgi?id=778302
|
|
|
|
|
|
|
|
|
| |
On the right situations (eg. fd exhaustion), both
g_dbus_message_get_unix_fd_list() and g_unix_fd_list_get() should be
considered failable here. Add proper error handling to avoid triggering
glib warnings.
https://bugzilla.gnome.org/show_bug.cgi?id=778540
|
|
|
|
|
|
|
| |
This will make extracted variables properly freed when extracting those
fails at some point.
https://bugzilla.gnome.org/show_bug.cgi?id=778540
|
|
|
|
|
|
|
| |
Those may be left uninitialized after error handling paths are introduced,
so invalid data might be attempted to be freed later on.
https://bugzilla.gnome.org/show_bug.cgi?id=778540
|
|
|
|
|
|
|
|
| |
Also calculate a valid length for the resulting array.
This enables easier usage of common uint8[]-based buffer API.
https://bugzilla.gnome.org/show_bug.cgi?id=777194
|
|
|
|
|
|
|
| |
Assigning values to properties this way leads to a ref/copy of the source
and therefore requires a unref/destroy afterwards.
https://bugzilla.gnome.org/show_bug.cgi?id=766739
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=778224
|
|
|
|
|
|
| |
This reverts commit 5b024e44077b6738855b6b2e92d913b75fda3444.
https://bugzilla.gnome.org/show_bug.cgi?id=777242
|
|
|
|
|
|
|
| |
This was working for methods and signals, but not for properties.
Also add tests for all cases.
https://bugzilla.gnome.org/show_bug.cgi?id=744595
|
|
|
|
|
| |
Doing so in CCodeDeclaration is error-prone. CCodeVariableDeclarator still
requires special handling which isn't done here.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
If a target glib between 2.36 and 2.44 is used, store a boolean var
in the async task data, which is initialized to FALSE and set to
TRUE in the async task callback, so async-forced-to-sync like the
async generator example can block until the task is complete.
There is one special case, if the async task receives no callback
to execute, the boolean flag is set to TRUE right away, as it
will be "finished" by the first time it's checked.
https://bugzilla.gnome.org/show_bug.cgi?id=763345
|
|
|
|
|
|
|
| |
Although avoid this check for simple-generic accesses where this would be
too strict and not wanted.
https://bugzilla.gnome.org/show_bug.cgi?id=775466
|