Re: Reevaluating the 2.4 schedule

[Matthias Clasen <maclas gmx de>]

Am Don, 2003-07-17 um 16.52 schrieb Owen Taylor:
> One definitely possibility is that we should just move it into GTK+ 
> as soon as possible, and at least move gtk-demo to it, write docs, etc,
> after all, we aren't going to get any experience with the API
> in practice as long as it languishes in libegg.

This gets my vote. Lets move it in. The libbonobo cognoscendi have had
plenty of time to review the stuff. If they haven't done till now, its
unlikely that we will get feedback before 2.3.0 gets out.

> Pretty clearly. I'm going to suggest that we push back all the
> dates by two weeks (to August 1...). While continually making
> small schedule slippages is less comfortable than whacking back the
> schedule and giving lots of breathing room, I'd rather keep
> up the pressure on myself and others to keep things movings.

I've moved all dates back by two weeks.

> Another thing that needs doing is to go through 2.4 API milestones
> for glib and gtk+ in bugzilla, there are:
> 
>  - Big things that aren't going to make 2.4
>  - Little things that we might as well do for 2.4
>  - Little things that aren't going make 2.4, because we have
>    to many big things to do even if the little things are easy.
>  - Patches sitting around ready for review
> 
> It would be really nice if we could get these milestones down to
> the things that we really think are going to make 2.4, instead
> of the clutter that is there now.

Ok. As a start, here is my breakdown of the glib 2.4 API milestone. It
has 24 bugs, of which 13 have a patch attached (marked with a '*' below)

 - Big things that could / should / must make 2.4

*105894 Support properties on interfaces
*63621 Add GAtomicInt for atomic integer operations

 - Big things that aren't going to make 2.4

80498 Added function g_io_channel_new_pipe()
*110528 collecting value arrays is inconvenient
100948 g_value_set_string_take_ownership should be official
*111848 function to canonicalize file names

 - Little things that we might as well do for 2.4

72545 g_readlink
*78414 g_queue_remove() and g_queue_remove_all()
*80487 A generalized accumulator for signals that return gboolean
*94879 can't remove more than one item at a time from an array
98536 Add g_timer_continue()
99226 Add G_HIDDEN
100763 Add g_unsetenv()?
*88329 Add g_strtokenize()
*97556 Eel functions for adding message comments should be moved to
101053 Should be able to block signal handlers on signal id

 - Little things that aren't going make 2.4, because we have
   to many big things to do even if the little things are easy.

81040 G_TYPE_GTYPE is missing from fundamental types
*83729 Patches to add iterators to GLib  
87761 read-write locks have no upgrade mechanism
*103710 Why is there no G_MININT32/16/8, G_MAXINT32/16/8, G_MAXUINT3
107397 G_TYPE_BOILERPLATE wanted, as replacement of GNOME_CLASS_BOI
113697 g_ptr_array_sort_[with_data] is twisted
113075 support "nonnull" attribute
*51639 Equivalent of __FUNCTION__ for Sun's compiler

Matthias