Alan Knowles [Sun, 18 Apr 2010 07:37:14 +0000 (15:37 +0800)]
Add void* userdata support for gobjects
gtk_menu_popup void* userdata when used with gtk_statusicon_position_menu
expect the userdata to be a pointer to the status icon. This change checks void*
arguments and if they are gobjects, sends them.
Tim Horton [Mon, 29 Mar 2010 04:12:00 +0000 (00:12 -0400)]
Postrelease version bump.
Tim Horton [Mon, 29 Mar 2010 04:04:27 +0000 (00:04 -0400)]
Update NEWS for 2.31.1
Tim Horton [Mon, 29 Mar 2010 03:33:55 +0000 (23:33 -0400)]
Update NEWS for 2.30
Alexandre Mazari [Wed, 17 Mar 2010 04:12:49 +0000 (12:12 +0800)]
Add gtkdocs for Request #612590 - initialization using existing context
gtkdoc comments for 2 new methods to initialize using existing context
Alan Knowles [Fri, 12 Mar 2010 04:31:57 +0000 (12:31 +0800)]
Nicer warning messages on our temporary warning about not_null check
As we do not throw Exceptions currently, it is difficult to work out what caused a warning,
This just adds the Container/Namespace of the method that caused the problem
Alexandre Mazari [Fri, 12 Mar 2010 04:17:24 +0000 (12:17 +0800)]
Feature Request #612590 - permit initialization using an existing JSGlobalContext
Usage example:
SeedEngine *eng;
void window_object_cleared (WebKitWebView *web_view,
WebKitWebFrame *frame,
gpointer context,
gpointer arg3,
gpointer user_data) {
eng = seed_init_with_context(NULL, NULL, context);
}
....
g_signal_connect(G_OBJECT(web_view), "window-object-cleared",
G_CALLBACK(window_object_cleared), NULL);
Alan Knowles [Thu, 11 Mar 2010 15:03:47 +0000 (23:03 +0800)]
change callback release_in_arg to release_arg otherwise segfaults occur
ClipboardTextReceivedFunc - char* text got freed before it was used.
The logic in release_in_arg is inconsitant with what I think is being attempted here.
Alan Knowles [Thu, 11 Mar 2010 07:09:19 +0000 (15:09 +0800)]
Allow access to global object of sandbox
x = new imports.sandbox.Context();
x.add_globals();
x.get_global_object().a = "hello world";
x.eval("Seed.print(a);");
Very usefull to implement a Console...
Alan Knowles [Tue, 9 Mar 2010 09:15:36 +0000 (17:15 +0800)]
Use fully qualified paths when sending to Evaluate script, so error messages are clearer
If you have two files called Abc.js in two different directories, the error message for syntax error (and others)
does not distinguish between them. The fix is to send the fully qualified path to the method.
Alan Knowles [Mon, 1 Mar 2010 09:06:01 +0000 (17:06 +0800)]
Change null into a warning for the time being, due to gir's need to catch up
Introspection tells us which arguments can be null, however the gir's for most packages have not been
updated to show this (Gtk had a big patchfest last week to add this, but it will not be reflected on end users
desktop gir's for a while) - until then we issue g_warnings.
alan [Fri, 26 Feb 2010 09:20:42 +0000 (17:20 +0800)]
More null checks - arguments that are structs/unions need the pointer checking to see if they are null
This may need moving upwards into make_argument, but that would mean the error messages would be less useful.
alan [Wed, 24 Feb 2010 05:42:16 +0000 (13:42 +0800)]
More detail on introspection of functions
Adds name, direction and allow_none
root [Wed, 24 Feb 2010 05:32:54 +0000 (13:32 +0800)]
Change Gtype conversion to use long
After reading the commit history - one of the notes mentioned that Gtypes should be long..
alan [Wed, 24 Feb 2010 05:07:39 +0000 (13:07 +0800)]
Fix Bug #609508 - Support Out args as return object if possible
Fixes issues with in/out arguments
Better debug message (names the argument)
For calls that have 'out' arguments
a) If return is VOID - the return is an object, with properties set to the out values
b) If the first out argument is an object, then all the out values are set as properties on that.
c) still supports out_argument.value setting (previous way of doing this)
alan [Wed, 24 Feb 2010 04:54:45 +0000 (12:54 +0800)]
White space fix
White space fix
alan [Wed, 24 Feb 2010 04:49:58 +0000 (12:49 +0800)]
Fix Bug #610786 - support for GError types in seed_gi_argument_make_js
Used by gst_message_parse_error() - turns gerror into non-thrown exception
alan [Wed, 24 Feb 2010 04:31:30 +0000 (12:31 +0800)]
Fix Bug# 610765 - test g_arg_info_may_be_null before accepting null
Correct behaviour is to read gir hints on null for arguments - otherwise segfaults happen...
This does however break a few function calls, that have got broken gir notes,
eg. gtk_drag_source_set gtk_drag_dest_set and gst_bus_add_watch
alan [Wed, 24 Feb 2010 03:35:00 +0000 (11:35 +0800)]
Fix Bug #610793 - support for array of GTypes
used by g_type_get_interfaces - which returns an array of Gtypes
Tim Horton [Tue, 23 Feb 2010 01:59:47 +0000 (20:59 -0500)]
Post-release version increment.
Tim Horton [Tue, 23 Feb 2010 01:58:52 +0000 (20:58 -0500)]
2.29.91.1 release (oops!)
Alan Knowles [Sun, 21 Feb 2010 03:32:35 +0000 (11:32 +0800)]
Fix Bug #610533 - Support void* pointers
Gio.outputstream.write() uses void* pointers to write generic data.
Tim Horton [Thu, 18 Feb 2010 23:19:19 +0000 (18:19 -0500)]
Ignore m4 dir (which we don't need in Git)
Tim Horton [Thu, 18 Feb 2010 23:19:04 +0000 (18:19 -0500)]
Fix missing signal test
Diego Escalante Urrelo [Wed, 17 Feb 2010 22:04:54 +0000 (17:04 -0500)]
seed-signals: allow notify:: signals
g_signal_query/g_signal_lookup don't recognize notify::x signal names,
they do recognize notify signal of course. Take this into account and
enable connecting to notify::x signals like shown by documentation.
Also add a SeedException when signal name is not valid instead of just
returning NULL.
Bug #610311
Diego Escalante Urrelo [Wed, 17 Feb 2010 21:56:43 +0000 (16:56 -0500)]
m4 dir is not needed at all
Bug #610314
Tim Horton [Mon, 8 Feb 2010 23:46:49 +0000 (18:46 -0500)]
Post-release version increment
Tim Horton [Mon, 8 Feb 2010 23:44:23 +0000 (18:44 -0500)]
NEWS for 2.29.90
Alan Knowles [Sat, 6 Feb 2010 02:10:58 +0000 (21:10 -0500)]
imports['somedir/somefile.js'] works now
Closes BGO#607987
Alan Knowles [Sat, 6 Feb 2010 01:57:24 +0000 (20:57 -0500)]
Fix handling of GValue structs; fixes segfault
Fixes gtk_tree_model_get_value calls eventually segfaulting
Fixes BGO#608015
Rob Taylor [Tue, 2 Feb 2010 23:22:40 +0000 (23:22 +0000)]
Fix subtle crashed when a dbus reply has no message content.
In the DBus spec, its perfectly valid to send empty reply messages, even
if a method call message has been set 'no_reply'. In this case, the dbus
module was invoking a closure with uninitialised data, causing all sorts
of interesting behaviour.
This patch fixes this by not invoking the
closure when a reply is empty.
Tim Horton [Mon, 25 Jan 2010 23:10:35 +0000 (18:10 -0500)]
Post-release version bump.
Tim Horton [Fri, 22 Jan 2010 21:25:38 +0000 (16:25 -0500)]
Update manpage and --help output with -e, etc.
Matt Arsenault [Sun, 17 Jan 2010 06:46:34 +0000 (01:46 -0500)]
Add -e option to repl.
Matt Arsenault [Wed, 6 Jan 2010 19:11:54 +0000 (14:11 -0500)]
Minor mpfr style change
Matt Arsenault [Wed, 6 Jan 2010 19:03:33 +0000 (14:03 -0500)]
Change copyright header of mpfr module
I wrote this useless part, not Robb, and it's 2010.
Matt Arsenault [Wed, 6 Jan 2010 17:34:59 +0000 (12:34 -0500)]
Add printerr
Tim Horton [Sat, 16 Jan 2010 10:03:19 +0000 (05:03 -0500)]
The manpage had completely incorrect command line arguments (triple dashes?!)
Tim Horton [Fri, 15 Jan 2010 04:36:17 +0000 (23:36 -0500)]
libseed: Properly bubble exceptions raised during class init
Before, we were simply catching the exception, printing it, and continuing
on happily. This completely defeats the point of exceptions. Now, we pop
the exception into the class's qdata (a little ugly) and bubble it up and
clear it the first time we encounter it (when the class is created, the
first time it's instantiated).
Tim Horton [Fri, 15 Jan 2010 01:37:02 +0000 (20:37 -0500)]
Revert "tests: make check should test against version in build tree, not installed"
This seems to make some things angry, and it doesn't totally make sense
(if build tree is elsewhere, or one uses 'clean', or whatever)
This reverts commit
018aaa4edbc49f1b1db1ea54417388f857ee3f12.
Tim Horton [Tue, 12 Jan 2010 06:35:35 +0000 (01:35 -0500)]
libseed: Fix passing a union out to a C function
The parent class of seed_union was set to seed_union (undefined at that point);
I believe it should be seed_pointer, just like seed_struct
Tim Horton [Mon, 11 Jan 2010 22:27:44 +0000 (17:27 -0500)]
Postrelease increment
Tim Horton [Mon, 11 Jan 2010 22:24:33 +0000 (17:24 -0500)]
Add 2.29.5.3 NEWS, and fix 2009->2010 problems
Tim Horton [Mon, 11 Jan 2010 22:15:26 +0000 (17:15 -0500)]
tests: add test for null struct return values
Ryan Brown [Mon, 11 Jan 2010 03:46:09 +0000 (21:46 -0600)]
Allow null struct return values
Tim Horton [Sat, 9 Jan 2010 00:12:58 +0000 (19:12 -0500)]
Revert "Fix incorrect type assumption"
We'll have to revisit this; it breaks things on Linux, and is all mired
in my lack of understanding
This reverts commit
07db8b4966ca85a256137d32eeca2a634f473825.
Tim Horton [Fri, 8 Jan 2010 01:40:05 +0000 (20:40 -0500)]
modules: Use NULL instead of 0 to ensure type safety
Tim Horton [Fri, 8 Jan 2010 00:17:09 +0000 (19:17 -0500)]
Add null sentinels to the end of seed_static_function and seed_static_value
arrays in all modules; fixes BGO #592585
Tim Horton [Fri, 8 Jan 2010 00:11:30 +0000 (19:11 -0500)]
Post-release increment.
Tim Horton [Fri, 8 Jan 2010 00:08:51 +0000 (19:08 -0500)]
Release 2.29.5.2
Tim Horton [Wed, 6 Jan 2010 09:06:59 +0000 (04:06 -0500)]
tests: make check should test against version in build tree, not installed
Tim Horton [Wed, 6 Jan 2010 08:53:37 +0000 (03:53 -0500)]
tests: Make C-based test run too
Tim Horton [Wed, 6 Jan 2010 08:47:37 +0000 (03:47 -0500)]
tests: Ditch the python test framework, use standard make check with a small
set of JavaScript functions to deal with assertions and unreachable code
and whatnot. Also, clean up some tests, and remove some that made no sense
whatsoever.
Three or four tests still don't work (and print output to that effect)
because of strange exception-chaining problems which I'm going to look into.
Tim Horton [Wed, 6 Jan 2010 07:05:05 +0000 (02:05 -0500)]
Return non-zero exit code if an exception occurs
Tim Horton [Wed, 6 Jan 2010 07:04:54 +0000 (02:04 -0500)]
Remove extraneous space in output
Tim Horton [Wed, 6 Jan 2010 05:04:07 +0000 (00:04 -0500)]
Note to self: need to properly integrate our exceptions with JS
Tim Horton [Mon, 4 Jan 2010 05:56:29 +0000 (00:56 -0500)]
Fix incorrect type assumption
seed_gi_argument_make_js was assuming that enums and flags are always longs,
which is (apparently) incorrect. Check the type and convert accordingly.
Tim Horton [Mon, 4 Jan 2010 05:56:15 +0000 (00:56 -0500)]
Fix spacing in warning (double space)
Tim Horton [Mon, 4 Jan 2010 03:11:03 +0000 (22:11 -0500)]
GType can be bigger than int32; use seed_value_to_long for now
This is probably not correct, since the size of the type is variable, and all...
Tim Horton [Sun, 3 Jan 2010 03:45:36 +0000 (22:45 -0500)]
Update OS X build instructions (MacPorts!)
Tim Horton [Sun, 3 Jan 2010 03:06:35 +0000 (22:06 -0500)]
libseed: Out arguments should use out_values, not out_args, for finding values!
Fixes bug introduced in commit
249a26bd077
Tim Horton [Sun, 3 Jan 2010 00:56:46 +0000 (19:56 -0500)]
tests: Fix struct-union-enumerate test
Tim Horton [Sun, 3 Jan 2010 00:40:30 +0000 (19:40 -0500)]
Fix readline tests
Tim Horton [Sat, 2 Jan 2010 23:40:24 +0000 (18:40 -0500)]
Fix completely incorrect nested struct/union set test
Tim Horton [Sat, 2 Jan 2010 23:19:27 +0000 (18:19 -0500)]
Ignore missing X extensions when testing
This fixes testing over X forwarding to a machine lacking expected extensions
Tim Horton [Sat, 2 Jan 2010 21:55:07 +0000 (16:55 -0500)]
Fix glong/gint "mismatch" elsewhere
Tim Horton [Sat, 2 Jan 2010 21:51:57 +0000 (16:51 -0500)]
Fix lack of gnome-js-common directory in default search path
I accidentally broke this when fixing the pkglibdir/datadir substitution
Tim Horton [Sat, 2 Jan 2010 21:43:23 +0000 (16:43 -0500)]
Fix enum typing bug from commit
cedbd78451
A glong/gint mismatch was wreaking havoc on various enums.
It's clear why this was broken; g_value_info_get_value says
that it returns a glong.
Tim Horton [Fri, 1 Jan 2010 22:26:51 +0000 (17:26 -0500)]
Post-release version increment
Tim Horton [Fri, 1 Jan 2010 22:25:58 +0000 (17:25 -0500)]
Release 2.29.5.1
Tim Horton [Fri, 1 Jan 2010 22:23:16 +0000 (17:23 -0500)]
Fix some 'duplicate symbol' errors on other platforms
Tim Horton [Fri, 1 Jan 2010 21:47:57 +0000 (16:47 -0500)]
Post-release increment
Tim Horton [Fri, 1 Jan 2010 21:38:56 +0000 (16:38 -0500)]
2.29.5 release notes
Tim Horton [Thu, 31 Dec 2009 10:44:51 +0000 (05:44 -0500)]
os module: oops! other GNU extensions were in use; reenable these
Tim Horton [Thu, 31 Dec 2009 10:40:29 +0000 (05:40 -0500)]
Use realpath() instead of the GNU libc extension canonicalize_file_path (again)
Tim Horton [Thu, 31 Dec 2009 10:36:09 +0000 (05:36 -0500)]
extensions: Fix substitution of paths into Seed.js
This fixes the failure to find native modules when installing to unexpected
prefixes.
Tim Horton [Thu, 31 Dec 2009 10:12:28 +0000 (05:12 -0500)]
extensions: fix some indentation oddities
Tim Horton [Wed, 30 Dec 2009 08:40:25 +0000 (03:40 -0500)]
interpreter: display exceptions (if any) when trying to execute the repl
Ryan Brown [Sat, 19 Dec 2009 21:04:51 +0000 (16:04 -0500)]
Use the right type when creating GValues for GObject subtypes
Tim Horton [Wed, 30 Dec 2009 07:56:42 +0000 (02:56 -0500)]
Distribute the repl snippet along with seed
Tim Horton [Wed, 30 Dec 2009 07:54:29 +0000 (02:54 -0500)]
Use realpath() instead of the GNU libc extension canonicalize_file_path
Tim Horton [Wed, 30 Dec 2009 07:53:33 +0000 (02:53 -0500)]
Remove Mac OS X specific configuration
There's no reason to use the system's WebKit, since it has to be a nightly anyway.
Instead, use webkit-gtk; the version in macports is now recent enough.
Tim Horton [Wed, 30 Dec 2009 07:39:26 +0000 (02:39 -0500)]
Use POSIX thread-local storage API instead of GCC extensions
This is necessary to make Seed build on platforms where the GCC
TLS API is not available, such as Mac OS X.
Robert Carr [Thu, 17 Dec 2009 22:32:36 +0000 (17:32 -0500)]
Improve handling of case where constructing struct with 0 size
Robert Carr [Thu, 17 Dec 2009 22:25:21 +0000 (17:25 -0500)]
[importer] Struct/union constructors need to take a reference to the GIBaseInfo they store in their privates...
Robert Carr [Thu, 17 Dec 2009 22:09:48 +0000 (17:09 -0500)]
Built in REPL was broken because it was installing from examples tree (now seed-examples on git.gnome.org). Move a copy of REPL snippet into seed
Robert Carr [Thu, 17 Dec 2009 20:40:07 +0000 (15:40 -0500)]
Add regression test for BGO 593182
Robert Carr [Thu, 17 Dec 2009 20:23:39 +0000 (15:23 -0500)]
seed_get_class_info_for_type had a logic error, causing GIBaseInfos to be unreffed, in the case of "They didn't exist".
This is obviously not a good idea and has been rectified. Closes BGO #593182
Robert Carr [Thu, 17 Dec 2009 20:23:34 +0000 (15:23 -0500)]
Update some autotools junk? I guess. I could only really say if I understood autotools
Iain Nicol [Mon, 2 Nov 2009 03:59:35 +0000 (03:59 +0000)]
libseed: do not call JSValueProtect on values the GC might have freed
Closes BGO #599666
Tim Horton [Wed, 16 Dec 2009 22:04:06 +0000 (17:04 -0500)]
Post-release increment
Tim Horton [Wed, 16 Dec 2009 22:01:05 +0000 (17:01 -0500)]
Update NEWS and README for 2.29.4
Steve Frécinaux [Sun, 15 Nov 2009 23:35:52 +0000 (00:35 +0100)]
reference: Add a note about search path ownership.
The current reference documentation of seed_engine_get_search_path()
does not specify whether the return value should be freed or not, and
that return value is not const so it's ambiguous.
https://bugzilla.gnome.org/show_bug.cgi?id=602032
Iain Nicol [Thu, 19 Nov 2009 13:03:35 +0000 (13:03 +0000)]
libseed: Improve integral type conversions; mostly portability issues
Partial fix for BGO #602404.
Steve Frécinaux [Tue, 15 Dec 2009 13:33:27 +0000 (14:33 +0100)]
Remove unused "arg_type" argument in make_native_closure.
This argument is kept around in the closure but isn't used anywhere,
so let's just drop it.
https://bugzilla.gnome.org/show_bug.cgi?id=604617
Steve Frécinaux [Tue, 15 Dec 2009 13:39:21 +0000 (14:39 +0100)]
Remove arg_info param from seed_gi_make_argument().
This arg_info value was only used when triggering make_native_closure(),
and was even set to NULL in most calls of this function, so let's just
remove it.
https://bugzilla.gnome.org/show_bug.cgi?id=604617
Tim Horton [Mon, 30 Nov 2009 22:24:18 +0000 (17:24 -0500)]
Postincrement after release (now 2.29.4)
Tim Horton [Tue, 17 Nov 2009 00:03:39 +0000 (19:03 -0500)]
Postincrement version number to 2.29.3
Tim Horton [Tue, 17 Nov 2009 00:01:18 +0000 (19:01 -0500)]
More 2.29.2 changes (update NEWS and version)
Tim Horton [Mon, 16 Nov 2009 23:52:27 +0000 (18:52 -0500)]
Seed 2.29.2 changes