Outstanding GTK+ patches
- From: Owen Taylor <otaylor redhat com>
- To: gtk-devel-list redhat com
- Subject: Outstanding GTK+ patches
- Date: 18 Feb 2000 19:18:45 -0500
Here is the (embarrassingly long) list of outstanding patches
on gtk.org. I've roughly classed them by effort they take
to be applied and whether I think they should be applied.
The notes below were written rather quickly, so I apologize
if they are cryptic in places. If you want more details on
any of the notes below, let me know.
I'm posting this here partly as a reference for myself and
the other core GTK+ developers, and partly so that those who
have submitted patches don't feel that there patches have
been completely forgotten. Hopefully, we can go through all
these patches in the next few weeks and either apply them
or get a response out to the author letting them know the
status of their patch.
Patches that should be applied
Adds gdk_draw_bitmap() to GDK. Should be applied.
Optimization for g_node_first_child. Looks good to apply.
Adds "shadow type" to scrolled-window, removing the need for
widgets like GtkViewport and GtkClist to draw their own shadows.
Also allows adding shadows to the Scrolled-Window + Layout
combination nicely. Should go in 1.3.
Make GtkText use IBeam rather than arrow for the mouse cursor.
Changes the repeat on the scrollbar arrows to have an initial
Spelling fixes for comments and docs.
Updates for nl.po file. Should be applied.
Fixes requiring further evaluation
Adds -R to the output of glib-config; I think one of the GNU info
pages has some manifesto as to why -R is evil... but otherwise
might be OK. (Would require the corresponding change to gtk
to be useful)
Patch to remove some warnings on HP/UX. The part of this
which get rid of implicit casts between gpointer and
function pointers should definitely be applied. #warning
is no longer found in glib. The #include <time.h> part also should
be unecessary since we simply include a forward declaration
for struct tm now.
Changes the way GtkViewport does size allocation by removing
some padding in most cases. Might be good to apply? Needs
Some changes to sending focus in events to deal with oddities
in XIM. Not clear if it is "correct" or not, though it obviously
does solve problems.
Changes ordering so has_selection is already set during the
"delete_text" callback. Neither this nor what is in GTK+
now look correct to me, since the delete_text callback
could refuse to delete the text.
Warnings fixups on HP/UX. May have some valid patches inside
of it, though a lot of the stuff there is a bit questionable.
Changes xoffset and yoffset in GtkLayout to be gint instead
of guint. Does it make sense for them to be negative?
Claim in README of bugs from this, but no information about
how to reproduce these bugs.
Makes exiting from a GTK+ program thread safe.
Fixes to GtkText.
Reasonable looking feature additions
Copy function for GNode structures. Could be useful. Very small
Patches to add a GIntSpan data type to GLib. These are compactly
represented integer sets, much like the GNUS range data
type. This might be useful for people...
Splay tree implementation. Require someone deciding the "what binary
trees do we want" questions. Do we want a generic binary
tree base-class with implementations?
Implements g_tree_lookup_range. Basically returns a GSlist of
all nodes in a tree in a certain range. Might be useful.
Adds "operation" when setting the shapes for a widget so you
can OR, etc, the shape with the existing shape. Might be
OK to apply though it should use GdkFunction enumeration.
Adds blinking cursor to GtkEntry.
Changes scrolled windows and Clist to scroll '1 page - 1 line'
instead of 0.5 page.
Adds a tail pointer to GtkCList to make appends linear. It
might be better to just convert the linear list to an array?
Considers all tooltips within a single Tooltips to be in a
single "group" and removes the delay when popping up a tooltip for
another widget in the same group immediately after exiting another
widget in the group.
This is incompatible with existing practice where people
just use a single tooltips object for everything, but probably
is a desirable UI enhancement.
Changes the way the focus indicator is drawn for buttons to be
windowsesque. I don't see applying this unless we make the
focus rectangle be a dotted rectangle for everything, and also
change the default to be drawn in the windows fashion
But that would break all the themes.Non-context diff!
Adds a MRU list to GtkFileSelection.
Adds gtk_clist_get_selection(), gtk_clist_get_nth_selection().
I don't think the latter function is useful, but the former
addition would remove a common question.
Mixes together gtk-rdm-990711 plus:
Changes GDK_RELIEF_HALF to display relief only on prelight
Changes GDK_RELIEF_NONE to never display relief.
This change is almost certainly correct, though it might
break existing programs.
Rationalizes GTK+ options to all accept either --value=option
or --value option. The gdk_arg_* stuff in 1.3 also fixes this,
so this is probably obsolete.
Fix segfault with GtkItemFactory. Was fixed in CVS some time
between 1.2.6 and submission of this patch.
cs.po. Looks like we have this in place now.
Makes window/drawable subsitutions. No README. I think all
this stuff is now properly changed in the 1.3 branch.
Adds some warnings for questionable/invalid allocations. We already
now do that now, though in a somewhat different way. (This patch
changes GtkAllocation to have signed coordinates)
Patches probably not suitable to be applied
Minor modifications to g_slist_reverse(). Modifying parameters
in a function is definitely acceptable. Also has suggestion about
removing an extra temporary, but would make no significant differnence.
Some fiddling with the Julian data formula in GLib. Seems to
be no advantage to applying this patch ... since it would
diverge the (complicated) formula from the referenced source
without any compelling performance advantage.
Patches to add some basic-like string functions to GLib. Mostly,
these are just renames of existing stdc functions...
not exactly within the normal domain of GLib.
Color->alpha function for pixmaps. I don't think this belongs in GDK;
you really should be using GdkPixbuf and doing this _before_ you
create your pixmap on the display rather than pulling the information
back from the X server. The implementation here doesn't even work
properly for pseudo-color displays.
Allows GTK+ applications to access X resources. I don't think
this belongs in GTK+. If you want X resources, use XLib within
your GTK+ application.
Adds a blinking cursor to GtkEntry. Also makes insertion point
cursor a "I-beam" Non-context diff! (Blinking part duplicate of
gtk-enf-990513-1, having the insertion cursor be a I-beam seems
cluttered to me; Motif does it that way, Windows doesn't)
Blippiness ... scheme to have menus and toolbars vanish when
window looses focus. Maybe could be done as a module?
GtkMagic ... fancy scheme for theme server/client mechanism.
Does not belong in GTK+, but patches to make it possible to
do as a module would be accepted.
Tries to make GtkLayout cross platform by adding fallback
implementation on windows in terms of GDK primitive. Not the
correct thing to do, since GtkLayout is just there to fix problems
in X that don't exist in windows. We'll have some better
fix for making GtkLayout cross-platform for this at the GDK
level in 1.3.
Major additions to the texi reference materials (!). Need to redirect
the author at the RDP and see if he is interested in integrating
any of his work into that.
Update to the pl.po file. Looks like we now have a complete
translation, thought some of the strings we now have are a bit
different from the ones here.
] [Thread Prev