------------------------------------------------------------------------
r932615 | zwabel | 2009-02-27 01:23:15 +0000 (Fri, 27 Feb 2009) | 6 lines
Backport r932614:
Copy the completion-models before iterating over them, and always make sure that the processed model has not been removed as a side-effect.
This should fix a possible crash that could happen when one o t he called function caused the completion to abort completey, thus invalidating the iterator.
BUG: 185661
------------------------------------------------------------------------
r932624 | mpyne | 2009-02-27 04:09:23 +0000 (Fri, 27 Feb 2009) | 5 lines
Backport fix for default buttons in KDialog to KDE 4.2.2.
CCBUG:148972
CCBUG:82014
------------------------------------------------------------------------
r932632 | mpyne | 2009-02-27 04:37:35 +0000 (Fri, 27 Feb 2009) | 4 lines
Backport fix for the "following link to local file" warning in KHTMLPart's default button to KDE 4.2.2.
BUG:82014
------------------------------------------------------------------------
r932676 | scripty | 2009-02-27 08:14:56 +0000 (Fri, 27 Feb 2009) | 1 line
SVN_SILENT made messages (.desktop file)
------------------------------------------------------------------------
r932944 | dfaure | 2009-02-27 15:53:18 +0000 (Fri, 27 Feb 2009) | 3 lines
Add a call to yylex_destroy (called kiotraderlex_destroy here) in order to fix the memory leak
on every trader query. Found by Kevin, confirmed by valgrind on kservicetest.
------------------------------------------------------------------------
r932955 | dfaure | 2009-02-27 16:11:55 +0000 (Fri, 27 Feb 2009) | 2 lines
Fix memory leak in KServiceTypeProfile::deleteServiceTypeProfile().
------------------------------------------------------------------------
r932964 | mueller | 2009-02-27 16:49:10 +0000 (Fri, 27 Feb 2009) | 2 lines
backport poppler detection fix
------------------------------------------------------------------------
r933015 | dfaure | 2009-02-27 20:08:14 +0000 (Fri, 27 Feb 2009) | 4 lines
1) fix KActions being leaked every time the RMB was pressed on a KTextEdit (they were all deleted with the textedit though, but that's a lot of action instances)
2) handle those actions' shortcut using OverrideEvent, since Qt's ambiguous shortcut detection is too stupid to give priority to focused widget, when using actions.
This fixes "Ctrl+R in a textedit reloads the konq page instead of showing the replace dialog", #164321
------------------------------------------------------------------------
r933021 | dfaure | 2009-02-27 20:33:04 +0000 (Fri, 27 Feb 2009) | 2 lines
Backport: fix yet-another-QCoreApplication::postEvent: Unexpected null receiver (due to NULL->deleteLater()).
------------------------------------------------------------------------
r933031 | dfaure | 2009-02-27 20:54:50 +0000 (Fri, 27 Feb 2009) | 3 lines
Fix shortcut clashes when splitting the konq window; the default "window context" is too much, restrict shortcuts to "when this widget or a child of it has focus"
CCBUG: 177332
------------------------------------------------------------------------
r933221 | chehrlic | 2009-02-28 12:29:46 +0000 (Sat, 28 Feb 2009) | 1 line
win32 compile++
------------------------------------------------------------------------
r933877 | orlovich | 2009-03-01 20:43:43 +0000 (Sun, 01 Mar 2009) | 4 lines
If a QLinkedList changes, QLinkedListIterators will operate on older version, which we don't want here
(we want removals from ~HTMLImageElementImpl to be visible); so just don't use an iterator.
Fixes a crash reported by Gof..
------------------------------------------------------------------------
r934014 | scripty | 2009-03-02 07:21:22 +0000 (Mon, 02 Mar 2009) | 1 line
SVN_SILENT made messages (.desktop file)
------------------------------------------------------------------------
r934035 | berendsen | 2009-03-02 08:50:22 +0000 (Mon, 02 Mar 2009) | 1 line
systemSeparatorMarkup -> system-separator-markup
------------------------------------------------------------------------
r934418 | jacopods | 2009-03-03 02:08:31 +0000 (Tue, 03 Mar 2009) | 3 lines
backport - prevents deadlock (deferredRun)
CCBUG: 181057
------------------------------------------------------------------------
r934461 | scripty | 2009-03-03 07:39:34 +0000 (Tue, 03 Mar 2009) | 1 line
SVN_SILENT made messages (.desktop file)
------------------------------------------------------------------------
r934465 | jacopods | 2009-03-03 07:52:25 +0000 (Tue, 03 Mar 2009) | 2 lines
Added a way to flag old searchJobs as stale, make sure that obsoleted matches do not interfere with new ones, corrected a misplaced brace in the previous commit
------------------------------------------------------------------------
r934516 | zwabel | 2009-03-03 11:01:16 +0000 (Tue, 03 Mar 2009) | 3 lines
Backport r934515:
When grouping is disabled, do not iterate over all items trying to expand them. This was very slow.
------------------------------------------------------------------------
r934518 | zwabel | 2009-03-03 11:03:29 +0000 (Tue, 03 Mar 2009) | 4 lines
Backport r934517:
- Reset the model at the beginning of createGroups(..). This should fix a crash with qt-4.5 that happened because of temporary the inconsistent state.
- Also limit the amount of best-match computation done when grouping is disabled.
------------------------------------------------------------------------
r934640 | helio | 2009-03-03 14:21:05 +0000 (Tue, 03 Mar 2009) | 2 lines
- Backport dfaure utf-8 filename fix from trunk.
------------------------------------------------------------------------
r934650 | mlaurent | 2009-03-03 14:30:01 +0000 (Tue, 03 Mar 2009) | 2 lines
Backport: fix select text now spellchecking works
------------------------------------------------------------------------
r934867 | zwabel | 2009-03-03 23:50:17 +0000 (Tue, 03 Mar 2009) | 3 lines
Backport r934866:
The model notification for the "broaden" operation sometimes does not work properly. Just reset the model after performing such an operation.
------------------------------------------------------------------------
r934869 | zwabel | 2009-03-03 23:52:51 +0000 (Tue, 03 Mar 2009) | 4 lines
Backport r934868:
Now with Qt 4.5, there seems to be some bug in Qt that makes the QTreeView show a scrollbar even when the contents fits into the space pixel-perfectly.
To workaround it, force the scroll-bar off when we know that the contents fits in.
------------------------------------------------------------------------
r935261 | orlovich | 2009-03-04 18:21:48 +0000 (Wed, 04 Mar 2009) | 5 lines
Go ahead and make removeChildren hookable, so table
and select can trap it like they do removeChild. Fixes a crash
with textContent = and tables (with wider cases
of some importance going forward).
------------------------------------------------------------------------
r935314 | dfaure | 2009-03-04 20:26:24 +0000 (Wed, 04 Mar 2009) | 3 lines
Support for sockets and char/block devices and named pipes, based on patch by Marcos Dione.
(BUG 176161)
------------------------------------------------------------------------
r935318 | apaku | 2009-03-04 20:30:05 +0000 (Wed, 04 Mar 2009) | 3 lines
Backport r935303:
Properly support default values for Url and UrlList
------------------------------------------------------------------------
r935345 | jacopods | 2009-03-04 21:05:24 +0000 (Wed, 04 Mar 2009) | 3 lines
"And they lived happily ever after with no more duplicate matches =)"
A local copy of the context is now created in the FindMatchesJob ctor, not anymore in AbstractRunner, this prevents old queries to be run with a valid dptr
------------------------------------------------------------------------
r935363 | ltoscano | 2009-03-04 21:48:33 +0000 (Wed, 04 Mar 2009) | 4 lines
Backport r935361:
Use the localized name for the shortcut.
------------------------------------------------------------------------
r935418 | mpyne | 2009-03-05 02:04:27 +0000 (Thu, 05 Mar 2009) | 6 lines
Backport fix for bug 174864 (Can't use file names with ':') to KDE 4.2.2. My
trunk commit mentioned that this helps except for full protocol lines, but file
names can't contain '/' so this is actually sufficient.
BUG:174864
------------------------------------------------------------------------
r935438 | scripty | 2009-03-05 07:48:14 +0000 (Thu, 05 Mar 2009) | 1 line
SVN_SILENT made messages (.desktop file)
------------------------------------------------------------------------
r935486 | rpedersen | 2009-03-05 09:06:33 +0000 (Thu, 05 Mar 2009) | 2 lines
backport 935483
------------------------------------------------------------------------
r935626 | aseigo | 2009-03-05 20:37:52 +0000 (Thu, 05 Mar 2009) | 2 lines
no to be shown
------------------------------------------------------------------------
r935641 | zwabel | 2009-03-05 21:18:39 +0000 (Thu, 05 Mar 2009) | 3 lines
Backport r935640:
- Do not notify the model when doing the "broaden" operation, because we use reset() to notify it in the end, and the local notifications don't work properly. This could lead to massive QTreeView internal warnings and GUI lockup.
------------------------------------------------------------------------
r935720 | zwabel | 2009-03-06 00:28:07 +0000 (Fri, 06 Mar 2009) | 3 lines
Backport r935719:
Don't forward "end" into the completion list. "home" isn't forwarded as well, so this is consistent.
------------------------------------------------------------------------
r935769 | scripty | 2009-03-06 07:42:34 +0000 (Fri, 06 Mar 2009) | 1 line
SVN_SILENT made messages (.desktop file)
------------------------------------------------------------------------
r935835 | dfaure | 2009-03-06 11:51:38 +0000 (Fri, 06 Mar 2009) | 4 lines
Use getgrouplist() instead of enumerating all groups with setgrent/getgrent, which can be really slow with centralized setups (e.g. winbind)
Based on patch by andreas at conectiva - thanks! (code simplified and made more C++). Fix will be in 4.2.2.
BUG: 89646
------------------------------------------------------------------------
r935837 | dfaure | 2009-03-06 11:54:51 +0000 (Fri, 06 Mar 2009) | 2 lines
Backport memleak fix in ACL code.
------------------------------------------------------------------------
r936036 | dfaure | 2009-03-06 19:31:03 +0000 (Fri, 06 Mar 2009) | 3 lines
backport the bugfix part of r926951: && is logical, & is bitwise.
V13 on irc noticed that this was needed to fix setAutostart(false) support.
------------------------------------------------------------------------
r936070 | fredrik | 2009-03-06 21:50:54 +0000 (Fri, 06 Mar 2009) | 1 line
Backport r936069.
------------------------------------------------------------------------
r936528 | zwabel | 2009-03-07 20:44:36 +0000 (Sat, 07 Mar 2009) | 5 lines
Backport r936527:
- Correctly maximize the height of all columns when computing the row-height for computing the height of the completion list
This fixes a problem that appeared with Qt 4.5 and led to the completion-list sometimes showing a scroll-bar when it shouldn't.
- Better management of the list focus
------------------------------------------------------------------------
r936583 | mpyne | 2009-03-08 01:08:00 +0000 (Sun, 08 Mar 2009) | 2 lines
No good reason not to initialize this job, and it gets rid of one of the many gcc warnings.
------------------------------------------------------------------------
r936976 | orlovich | 2009-03-08 19:10:36 +0000 (Sun, 08 Mar 2009) | 9 lines
Fix hovered disabled & complex toolbuttons in Oxygen appearing sunken with Qt4.5
Qt4.5 now sets activeSubControls for hover, too; so we shouldn't sink on that;
and on inspection, we didn't need to with 4.4, either.
I think Oxygen's r922918 can be reverted now, and doing so will likely restore
the hover effects.
------------------------------------------------------------------------
r937063 | mpyne | 2009-03-08 22:44:28 +0000 (Sun, 08 Mar 2009) | 1 line
Backport data:// icon fix to KDE 4.2.2.
------------------------------------------------------------------------
r937086 | mpyne | 2009-03-08 23:14:06 +0000 (Sun, 08 Mar 2009) | 1 line
Backport fix file file kioslave broken icon to KDE 4.2.2.
------------------------------------------------------------------------
r937096 | mpyne | 2009-03-08 23:36:08 +0000 (Sun, 08 Mar 2009) | 1 line
Backport fix for broken mailto: icon to KDE 4.2.2.
------------------------------------------------------------------------
r937328 | orlovich | 2009-03-09 14:38:22 +0000 (Mon, 09 Mar 2009) | 12 lines
We want to convert null => "" in CSS style manipulation.
Fixes expanding of changelog entries on launchpad bazzar changetracker
thingie.
Leaving the report open for now since I need to figure out why
not fixing this produces these:
QPainter::begin: Widget painting can only begin as a result of a paintEvent
QPainter::translate: Painter not active
(Plus I need to add a testcase, and don't have time right now)
CCBUG:186629
------------------------------------------------------------------------
r937495 | habacker | 2009-03-09 19:22:16 +0000 (Mon, 09 Mar 2009) | 1 line
backported rev 914535, 926526 and 936752 from trunk
------------------------------------------------------------------------
r937518 | zwabel | 2009-03-09 20:52:36 +0000 (Mon, 09 Mar 2009) | 4 lines
Backport r937517:
When checking whether a text-hint should be shown, map to the parent widget coordinate when comparing to geometry(), because that is also expressed in parent coordinates.
This fixes the problem that tooltips didn't work close to the left border.
------------------------------------------------------------------------
r937799 | mlaurent | 2009-03-10 13:24:16 +0000 (Tue, 10 Mar 2009) | 2 lines
Backport: fix enable/disable "find next" action
------------------------------------------------------------------------
r937807 | mlaurent | 2009-03-10 13:32:41 +0000 (Tue, 10 Mar 2009) | 2 lines
Backport: don't "find next" when document is empty
------------------------------------------------------------------------
r937917 | smartins | 2009-03-10 19:28:51 +0000 (Tue, 10 Mar 2009) | 4 lines
Backport r937907 by smartins from trunk to the 4.2 branch:
Support for disabled KGuiItems in KPushButton::setGuiItem().
------------------------------------------------------------------------
r937996 | dfaure | 2009-03-10 23:01:16 +0000 (Tue, 10 Mar 2009) | 2 lines
Backport fix for infinite loop when typing "." in KDirSelectDialog. I wonder if there are bugzilla reports for it.
------------------------------------------------------------------------
r938005 | dfaure | 2009-03-10 23:14:24 +0000 (Tue, 10 Mar 2009) | 8 lines
Backport:
* Escape things properly so that help: doesn't show a msgbox ;)
* Ensure help:../../../etc/passwd isn't allowed.
* Fix assert on "help:" without a path.
None of these are really security issues since redirecting to help: issues a warning,
and scripts run in the null domain, but anyway we might as well make this a bit more robust.
------------------------------------------------------------------------
r938007 | dfaure | 2009-03-10 23:30:07 +0000 (Tue, 10 Mar 2009) | 2 lines
SVN_SILENT silence an error that isn't one, happens all the time with shortlived .new and .tmp files
------------------------------------------------------------------------
r938011 | dfaure | 2009-03-10 23:43:29 +0000 (Tue, 10 Mar 2009) | 3 lines
Backport r938010: focus the lineedit by default (for renaming), rather than the Cancel button.
BUG: 182363
------------------------------------------------------------------------
r938121 | dfaure | 2009-03-11 11:38:35 +0000 (Wed, 11 Mar 2009) | 3 lines
Fix kde3 migration bug, my kmail_part.rc said:
but Qt4 wants Alt++ and Ctrl++.
------------------------------------------------------------------------
r938129 | dfaure | 2009-03-11 12:09:28 +0000 (Wed, 11 Mar 2009) | 2 lines
Backport SVN commit 938020 by sreich, fixes context menu.
------------------------------------------------------------------------
r938130 | dfaure | 2009-03-11 12:12:18 +0000 (Wed, 11 Mar 2009) | 2 lines
disable very-verbose StorageDebug by default.
------------------------------------------------------------------------
r938338 | orlovich | 2009-03-11 16:13:30 +0000 (Wed, 11 Mar 2009) | 8 lines
Make sure we don't suspend the tokenizer for a yield when we're handling things like innerHTML.
Fixes us dropping parts of large innerHTML updates, especially on debug builds and on slower machines.
Yeah, seems like the tokenizer & parser need a good profile run.. I also do have
some performance improvements for the tokenizer's element ID resolution pending...
Also fix the typo: yeld -> yield while I am at it.
------------------------------------------------------------------------
r938348 | orlovich | 2009-03-11 16:37:50 +0000 (Wed, 11 Mar 2009) | 3 lines
Don't fall back to latin1 for no good reason... Fixes us not using utf-8 with XHR
in some cases.
------------------------------------------------------------------------
r938370 | dfaure | 2009-03-11 18:08:04 +0000 (Wed, 11 Mar 2009) | 4 lines
Backport the encoding fix for Qt-4.5.0 (r934710), adding a #if QT_VERSION >= 0x040500.
This breaks compilation with 4.5-rc1, but you're not supposed to use that anyway.
BUG: 186038
------------------------------------------------------------------------
r938541 | scripty | 2009-03-12 07:44:53 +0000 (Thu, 12 Mar 2009) | 1 line
SVN_SILENT made messages (.desktop file)
------------------------------------------------------------------------
r938636 | jacopods | 2009-03-12 14:58:28 +0000 (Thu, 12 Mar 2009) | 4 lines
Now we are able to distinguish when text is changed by autocompletion or by the user
+ Added signal userTextChanged, emitted when text is changed NOT by autocompletion
+ Added function userText, to retrieve text without autocompletion
------------------------------------------------------------------------
r938640 | orlovich | 2009-03-12 15:17:03 +0000 (Thu, 12 Mar 2009) | 2 lines
First attempt at Qt4.5 support in testregression; ahem still looks funny.
------------------------------------------------------------------------
r938649 | orlovich | 2009-03-12 15:28:27 +0000 (Thu, 12 Mar 2009) | 3 lines
We can key the getElementById and underDocNamedCache with DOMString these days,
and make the code cleaner and more efficient...
------------------------------------------------------------------------
r938657 | dfaure | 2009-03-12 15:45:24 +0000 (Thu, 12 Mar 2009) | 3 lines
Fix encoding bug, "kdialog --sorry ?\195?\169" or "kmail --subject ?\195?\169" would double-encode the chars, due to a bad QByteArray-to-QString conversion (regression from r931769). Fix will be in 4.2.2.
CCBUG: 186038
------------------------------------------------------------------------
r938788 | dfaure | 2009-03-13 00:10:05 +0000 (Fri, 13 Mar 2009) | 2 lines
Backport percent() fix for storedput.
------------------------------------------------------------------------
r939243 | orlovich | 2009-03-14 13:50:37 +0000 (Sat, 14 Mar 2009) | 3 lines
Backport vtokarev's DOMStringImpl shallow copy mode changes +
the getElementById optimization.
------------------------------------------------------------------------
r939328 | qbast | 2009-03-14 17:08:40 +0000 (Sat, 14 Mar 2009) | 5 lines
Backport of commit 939233:
Cache tag should be created from "etag" header, not "date".
Probably botched copy&paste.
------------------------------------------------------------------------
r939348 | orlovich | 2009-03-14 17:35:13 +0000 (Sat, 14 Mar 2009) | 9 lines
Some significant name lookup optimizations for IDString ---
doing case normalization as we do the lookup, avoiding the need
for explicit lowercasing. Also tweak splitPrefixAndLocalname
to avoid needless copiesin the normal, :-less case.
Gives about a 10-15% speedup on my attach-less (display:none)
parsing benchmark, and about 1.4x or so speedup on getAttribute &
similar microbenchmarks.
------------------------------------------------------------------------
r939363 | orlovich | 2009-03-14 18:11:38 +0000 (Sat, 14 Mar 2009) | 3 lines
Use the new API here too... Makes it cleaner, and maybe a bit faster
------------------------------------------------------------------------
r939380 | aacid | 2009-03-14 18:54:23 +0000 (Sat, 14 Mar 2009) | 6 lines
Backport r939378 | aacid | 2009-03-14 19:52:14 +0100 (Sat, 14 Mar 2009) | 4 lines
Commit my "probably not real solution" patch to fix Ctrl+Delete losing cursor position
Acked by dhaumann
BUG: 178379
------------------------------------------------------------------------
r939470 | ggarand | 2009-03-14 22:42:26 +0000 (Sat, 14 Mar 2009) | 13 lines
automatically merged revision 939169:
support CSS3's background-attachment: local
cf. http://www.w3.org/TR/css3-background/#background-attachment
Scrolling the background with the element content is actually the
behaviour we were using for background-attachment: scroll, in order to
emulate MSIE6/7, but that was a violation of CSS 2.1 (#185631),
so fix that too.
(also fix bogus borderTop addition in isPointInScrollbar, noticed while
playing with thick bordered background-attachment testcases)
BUG: 185631
------------------------------------------------------------------------
r939471 | ggarand | 2009-03-14 22:43:35 +0000 (Sat, 14 Mar 2009) | 6 lines
automatically merged revision 939170:
from recent www-style discussion outcome,
input type=hidden should not match :disabled/:enabled
pseudo classes.
(also fix some decadent spelling)
------------------------------------------------------------------------
r939473 | ggarand | 2009-03-14 22:44:15 +0000 (Sat, 14 Mar 2009) | 4 lines
automatically merged revision 939172:
merge two Apple one liners resulting from spec clarification on www-style
(text-bottom alignment, behaviour of borders of inline flows when
interrupted by block-level elements)
------------------------------------------------------------------------
r939476 | ggarand | 2009-03-14 22:48:18 +0000 (Sat, 14 Mar 2009) | 6 lines
automatically merged revision 939171:
rework the algorithm that avoids endless scrollbar cycles.
there were several mistakes made during Qt4 porting, which in turn made
direct layer translation often trigger pathological scrollbar flicker
(e.g. when moving some dhtml positioned boxes)
------------------------------------------------------------------------
r939477 | ggarand | 2009-03-14 22:48:43 +0000 (Sat, 14 Mar 2009) | 7 lines
automatically merged revision 939173:
also avoid setting the canvas size to the new viewport size before it has
actually changed.
That could confuse the layout code into thinking it doesn't need
a deep relayout.
(fix failing regression test positioning/fixed-001.html)
------------------------------------------------------------------------
r939480 | ggarand | 2009-03-14 22:54:15 +0000 (Sat, 14 Mar 2009) | 10 lines
automatically merged revision 939174:
queue DOM resize events sent to subframes until the toplevel frame
has finished resizing.
This avoids infinite recursion scenarios where javascript would run
during the layout of the toplevel frame.
Problem acurately diagnosed by SadEagle.
BUG: 185790, 182027
------------------------------------------------------------------------
r939488 | ggarand | 2009-03-14 23:20:27 +0000 (Sat, 14 Mar 2009) | 17 lines
automatically merged revision 939487:
Qt < 4.5 had a bug in the layout of QAbstractScrollArea's scrollbars,
that made impossible to properly compute the size of the viewport
for styles using SH_ScrollView_FrameOnlyAroundContents (e.g. Oxygen).
This in turn caused all sorts of off-by-two artifacts.
We once had a symptomatic turnaround made with trial and errors,
but it was wrong and ceased to work with recent Oxygen changes.
Here is the correct fix.
Please test with Qt < 4.4 on branch if you can.
This is for backporting purposes only, and will be removed from trunk
in a moment.
BUG: 179359
------------------------------------------------------------------------
r939531 | scripty | 2009-03-15 07:45:33 +0000 (Sun, 15 Mar 2009) | 1 line
SVN_SILENT made messages (.desktop file)
------------------------------------------------------------------------
r939854 | ggarand | 2009-03-15 21:05:23 +0000 (Sun, 15 Mar 2009) | 7 lines
automatically merged revision 939175:
merge continuations back into a single inline when the block
that created them is removed from the tree.
this prevents unwanted line breaks caused by the stray anonymous blocks.
CCBUG: 176681
------------------------------------------------------------------------
r939958 | scripty | 2009-03-16 07:50:58 +0000 (Mon, 16 Mar 2009) | 1 line
SVN_SILENT made messages (.desktop file)
------------------------------------------------------------------------
r939987 | coolo | 2009-03-16 09:09:27 +0000 (Mon, 16 Mar 2009) | 3 lines
always pass a mode if the flag is O_CREAT
BUG: 187260
------------------------------------------------------------------------
r940021 | coolo | 2009-03-16 11:49:14 +0000 (Mon, 16 Mar 2009) | 2 lines
remember fd
------------------------------------------------------------------------
r940122 | rpedersen | 2009-03-16 16:32:41 +0000 (Mon, 16 Mar 2009) | 7 lines
backport 940117: Fix ruby intenter for block comments and heredocs
Added new functions to the scripting API to make more information
about syntax highlight attributes available to the intent script.
BUG: 187199
------------------------------------------------------------------------
r940172 | chehrlic | 2009-03-16 19:04:26 +0000 (Mon, 16 Mar 2009) | 3 lines
backport r940164 from trunk
CCBUG: 184462
------------------------------------------------------------------------
r940185 | aacid | 2009-03-16 19:39:09 +0000 (Mon, 16 Mar 2009) | 4 lines
Backport r940184 | aacid | 2009-03-16 20:37:45 +0100 (Mon, 16 Mar 2009) | 2 lines
This is causing accelerators not to be properly detected on KTabWidget, remove, semi.acked by ridell
------------------------------------------------------------------------
r940244 | jlayt | 2009-03-16 23:37:15 +0000 (Mon, 16 Mar 2009) | 5 lines
Fix Cups support to translate job print time entered by user in local time to
UTC as expected by Cups. Backport from 4.3.
BUG: 186954
------------------------------------------------------------------------
r940709 | ggarand | 2009-03-18 05:05:05 +0000 (Wed, 18 Mar 2009) | 20 lines
automatically merged revision 939176:
Rephrase the line-breaking algorithm, so that inline flows are no longer
skipped and accounted for retroactively as part of text objects.
Inline flows have to stand on their own. They might have fancy
border/padding/margin, that can be different on each end.
We will still skip them eventually, but only after their endings
have been gauged.
Ditto when deciding if a given render object requires a line box or
can be skipped as part of initial whitespace.
you could have nested inline flows with different b/p/m at each end,
so both ends must be walked.
Initial inline flows will also trigger the ignoringSpaces state as needed -
indeed their b/p/m is not supposed to affect the treatment of initial
whitespace.
BUG: 176681
------------------------------------------------------------------------
r940710 | ggarand | 2009-03-18 05:06:02 +0000 (Wed, 18 Mar 2009) | 1 line
automatically merged revision 939526:
------------------------------------------------------------------------
r940712 | ggarand | 2009-03-18 05:07:42 +0000 (Wed, 18 Mar 2009) | 10 lines
automatically merged revision 940706:
fix tricky regression noticed by Vyacheslav Tokarev on Google Reader.
the midpoint flip/flop machinery won't work properly when served skippable
InlineFlows, so have addMidpoint advance to the next object in that case.
also remove the adjustEmbedding global bool, now expressed by the
passing (or not) of a BidiState*.
CCBUG: 176681
------------------------------------------------------------------------
r940713 | ggarand | 2009-03-18 05:08:21 +0000 (Wed, 18 Mar 2009) | 7 lines
automatically merged revision 940704:
use a stricter grammar for the parsing of CSS selectors:
require at least a whitespace for the 'Descendant' combinator.
we used to accept 'html*' as 'html *'
CCBUG: 156947
------------------------------------------------------------------------
r940714 | ggarand | 2009-03-18 05:09:28 +0000 (Wed, 18 Mar 2009) | 7 lines
automatically merged revision 940705:
don't discard iframe children.
changed proposed by Maksim a long time ago.
looks right to me too.
CCBUG: 156947
------------------------------------------------------------------------
r940735 | scripty | 2009-03-18 08:17:20 +0000 (Wed, 18 Mar 2009) | 1 line
SVN_SILENT made messages (.desktop file)
------------------------------------------------------------------------
r940866 | mlaurent | 2009-03-18 12:43:32 +0000 (Wed, 18 Mar 2009) | 3 lines
Backport:
don't continue to search when we don't have search string
------------------------------------------------------------------------
r940869 | mlaurent | 2009-03-18 12:49:50 +0000 (Wed, 18 Mar 2009) | 4 lines
Backport:
When find string is empty delete d->find => "find next" action is disable after.
------------------------------------------------------------------------
r941024 | ogoffart | 2009-03-18 18:07:57 +0000 (Wed, 18 Mar 2009) | 6 lines
Backport 941021
Plasma::Svg: Round the drawing size to the pixmap size, otherwise we can get bad arctefacts.
(noticable on the eyes applet)
------------------------------------------------------------------------
r941184 | smartins | 2009-03-19 01:23:32 +0000 (Thu, 19 Mar 2009) | 6 lines
Backport r941182 by smartins from trunk to the 4.2 branch:
Don't crash while opening non well-formed SVG files.
CCBUG: 160421
------------------------------------------------------------------------
r941586 | cfeck | 2009-03-20 01:14:51 +0000 (Fri, 20 Mar 2009) | 2 lines
backported r941548 for SeaJey :)
------------------------------------------------------------------------
r941599 | dfaure | 2009-03-20 01:38:45 +0000 (Fri, 20 Mar 2009) | 5 lines
Backport speed improvement from r941585, but things are still slow in this branch due to Qt-4.4.
An optimization was added in 4.5's QItemSelectionModelPrivate::_q_layoutChanged:
"special case for when all indexes are selected".
CCBUG: 185267
------------------------------------------------------------------------
r941621 | cfeck | 2009-03-20 03:04:57 +0000 (Fri, 20 Mar 2009) | 4 lines
Fix broken menus in status bar widgets (backport 941619)
CCBUG: 177328
------------------------------------------------------------------------
r941644 | scripty | 2009-03-20 08:13:28 +0000 (Fri, 20 Mar 2009) | 1 line
SVN_SILENT made messages (.desktop file)
------------------------------------------------------------------------
r941801 | chehrlic | 2009-03-20 14:23:26 +0000 (Fri, 20 Mar 2009) | 1 line
backport r941788 from trunk
------------------------------------------------------------------------
r942041 | dfaure | 2009-03-20 21:34:34 +0000 (Fri, 20 Mar 2009) | 2 lines
Backport 928748+928749+928799: support for generic-icon fallbacks from shared-mime-info when loading mimetype icons.
------------------------------------------------------------------------
r942132 | mlaurent | 2009-03-21 07:54:53 +0000 (Sat, 21 Mar 2009) | 3 lines
backport: svn #941893
Fix enabled/disabled search button
------------------------------------------------------------------------
r942136 | scripty | 2009-03-21 08:16:07 +0000 (Sat, 21 Mar 2009) | 1 line
SVN_SILENT made messages (.desktop file)
------------------------------------------------------------------------
r942214 | terwou | 2009-03-21 12:55:07 +0000 (Sat, 21 Mar 2009) | 4 lines
backport r942212
Fix uninitialized variable fixes default value of checkAtEntering
insertStringList/insertItem match documentation: append when index is negative
------------------------------------------------------------------------
r942329 | orlovich | 2009-03-21 15:16:09 +0000 (Sat, 21 Mar 2009) | 4 lines
Merged revision 926297:r926297 | sverberkt | 2009-02-14 19:28:49 -0500 (Sat, 14 Feb 2009) | 3 lines
Don't let labels fire click events on disabled objects.
BUG:159682
------------------------------------------------------------------------
r942331 | orlovich | 2009-03-21 15:18:23 +0000 (Sat, 21 Mar 2009) | 4 lines
Merged revision:r926642 | sverberkt | 2009-02-15 17:03:20 -0500 (Sun, 15 Feb 2009) | 3 lines
Submit a form when return is hit on a checkbox or radio button
BUG:152035
------------------------------------------------------------------------
r942334 | orlovich | 2009-03-21 15:20:49 +0000 (Sat, 21 Mar 2009) | 5 lines
Merged revision:r932486 | sverberkt | 2009-02-26 13:58:42 -0500 (Thu, 26 Feb 2009) | 4 lines
Allow empty text in a JavaScript window.prompt
BUG:145213
------------------------------------------------------------------------
r942337 | orlovich | 2009-03-21 15:23:42 +0000 (Sat, 21 Mar 2009) | 5 lines
Merged revision:r932558 | sverberkt | 2009-02-26 16:55:05 -0500 (Thu, 26 Feb 2009) | 4 lines
Don't allow multiple radiobuttons to be checked on default.
BUG:161272
------------------------------------------------------------------------
r942340 | orlovich | 2009-03-21 15:26:16 +0000 (Sat, 21 Mar 2009) | 5 lines
Merged revision:r936439 | sverberkt | 2009-03-07 11:32:22 -0500 (Sat, 07 Mar 2009) | 4 lines
Also fire onChange events when using auto-completion.
BUG:113776
------------------------------------------------------------------------
r942354 | orlovich | 2009-03-21 15:33:37 +0000 (Sat, 21 Mar 2009) | 3 lines
Merged revision:r937433 | vtokarev | 2009-03-09 13:33:13 -0400 (Mon, 09 Mar 2009) | 1 line
fix typo, create things more lazily to avoid unnecessary allocations
------------------------------------------------------------------------
r942355 | orlovich | 2009-03-21 15:34:48 +0000 (Sat, 21 Mar 2009) | 3 lines
Merged revision:r938362 | orlovich | 2009-03-11 13:30:14 -0400 (Wed, 11 Mar 2009) | 2 lines
Put vtokarev's commit back in now that I know what's wrong with it.
------------------------------------------------------------------------
r942357 | orlovich | 2009-03-21 15:36:01 +0000 (Sat, 21 Mar 2009) | 4 lines
Merged revision:r938365 | orlovich | 2009-03-11 13:32:32 -0400 (Wed, 11 Mar 2009) | 3 lines
Fix off-by-one that caused an infinite loop with explosive memory use.
Yey for fancy version control systems that exceed human ability to keep track of stuff!
------------------------------------------------------------------------
r942360 | orlovich | 2009-03-21 15:44:22 +0000 (Sat, 21 Mar 2009) | 5 lines
Merged revision:r937435 | vtokarev | 2009-03-09 13:35:16 -0400 (Mon, 09 Mar 2009) | 3 lines
Another round of css selector optimizations:
- share same selectors cache between different collect() calls
- create properties once in a big chunk rather than allocate them separetely
------------------------------------------------------------------------
r942361 | orlovich | 2009-03-21 15:45:56 +0000 (Sat, 21 Mar 2009) | 4 lines
Merged revision:r938371 | vtokarev | 2009-03-11 14:08:21 -0400 (Wed, 11 Mar 2009) | 2 lines
don't trigger second look up in the cache if there is no key
nice HashMap api
------------------------------------------------------------------------
r942363 | orlovich | 2009-03-21 15:47:05 +0000 (Sat, 21 Mar 2009) | 8 lines
Merged revision:r938372 | vtokarev | 2009-03-11 14:08:25 -0400 (Wed, 11 Mar 2009) | 6 lines
Remove unneeded properties array, also replace manually allocated properties
with vectors - these stuff isn't really hot, but it leads to much cleaner code
Conflicts:
khtml/css/cssstyleselector.cpp
------------------------------------------------------------------------
r942364 | orlovich | 2009-03-21 15:50:31 +0000 (Sat, 21 Mar 2009) | 5 lines
Merged revision:r938373 | vtokarev | 2009-03-11 14:08:30 -0400 (Wed, 11 Mar 2009) | 3 lines
Instead of using HashMap with Vector values to get multi maps
link similar selectors in a list by remembering next similar selector
now store only first selector index in map, others could be produced by iterating
------------------------------------------------------------------------
r942365 | orlovich | 2009-03-21 15:52:47 +0000 (Sat, 21 Mar 2009) | 3 lines
Merged revision:r938374 | vtokarev | 2009-03-11 14:08:35 -0400 (Wed, 11 Mar 2009) | 1 line
remove redundant methods
------------------------------------------------------------------------
r942366 | orlovich | 2009-03-21 15:58:22 +0000 (Sat, 21 Mar 2009) | 3 lines
Merged revision:r938716 | vtokarev | 2009-03-12 15:35:43 -0400 (Thu, 12 Mar 2009) | 1 line
optimize isWhitespace/isClassWhitespace methods, take faster path via .unicode() call
------------------------------------------------------------------------
r942367 | orlovich | 2009-03-21 15:59:44 +0000 (Sat, 21 Mar 2009) | 4 lines
Merged revision:r938809 | vtokarev | 2009-03-12 22:21:46 -0400 (Thu, 12 Mar 2009) | 2 lines
Sort only matched selectors before collecting properties
should save lots of CPU cycles
------------------------------------------------------------------------
r942368 | orlovich | 2009-03-21 16:02:10 +0000 (Sat, 21 Mar 2009) | 3 lines
Merged revision:r939354 | vtokarev | 2009-03-14 13:48:03 -0400 (Sat, 14 Mar 2009) | 1 line
Use QLatin1Stringa and QLatin1Char to avoid redundant casts. Also try to optimize methods a bit
------------------------------------------------------------------------
r942369 | orlovich | 2009-03-21 16:02:54 +0000 (Sat, 21 Mar 2009) | 3 lines
Merged revision:r939355 | vtokarev | 2009-03-14 13:48:18 -0400 (Sat, 14 Mar 2009) | 1 line
don't use indexOf to compare 2 strings
------------------------------------------------------------------------
r942370 | orlovich | 2009-03-21 16:03:36 +0000 (Sat, 21 Mar 2009) | 8 lines
Merged revision:r939356 | vtokarev | 2009-03-14 13:48:22 -0400 (Sat, 14 Mar 2009) | 6 lines
Avoid usage of QVector as temporary buffers. .clear() on it triggers
internal buffer deletion and shrinking capacity to 0. Which is not good
if you want to append something to it afterwards.
Replacing with saner WTF::Vector and QVarLengthArray, which gives significant
improvement for ProspectiveTokenizer. And less allocations too.
don't ever use QVector if you want to clear it a lot
------------------------------------------------------------------------
r942373 | orlovich | 2009-03-21 16:06:33 +0000 (Sat, 21 Mar 2009) | 4 lines
Merged revision:r941604 | vtokarev | 2009-03-19 21:49:24 -0400 (Thu, 19 Mar 2009) | 2 lines
Use optimized version of getAttribute, store DOMString for contentLanguage
in the document, to avoid conversions
------------------------------------------------------------------------
r942375 | orlovich | 2009-03-21 16:07:42 +0000 (Sat, 21 Mar 2009) | 4 lines
Merged revision:r941605 | vtokarev | 2009-03-19 21:49:54 -0400 (Thu, 19 Mar 2009) | 2 lines
Add caseSensitivity parameter, for startsWith/endsWith methods in
DOMStrngImpl, which allows to avoid some more conversions in checkSimpleSelector
------------------------------------------------------------------------
r942377 | orlovich | 2009-03-21 16:09:35 +0000 (Sat, 21 Mar 2009) | 3 lines
Merged revision:r941880 | vtokarev | 2009-03-20 12:21:01 -0400 (Fri, 20 Mar 2009) | 1 line
misunderstood lang selector code and made it wrong with previous commit, fix it now
------------------------------------------------------------------------
r942378 | orlovich | 2009-03-21 16:10:16 +0000 (Sat, 21 Mar 2009) | 5 lines
Merged revision:r941935 | vtokarev | 2009-03-20 14:35:32 -0400 (Fri, 20 Mar 2009) | 3 lines
Nice performance improvement by Maksim Orlovich in IDString
and my utilization of that in prospective tokenizer, which makes
processAttribute/emitTag timing sane
------------------------------------------------------------------------
r942386 | orlovich | 2009-03-21 16:21:48 +0000 (Sat, 21 Mar 2009) | 3 lines
Merged revision:r917965 | gateau | 2009-01-28 16:22:26 -0500 (Wed, 28 Jan 2009) | 2 lines
Fix error handling in executeScript(const DOM::Node &n, const QString &script)
------------------------------------------------------------------------
r942393 | orlovich | 2009-03-21 16:30:57 +0000 (Sat, 21 Mar 2009) | 4 lines
Merged revision:r933834 | fredrik | 2009-03-01 13:00:21 -0500 (Sun, 01 Mar 2009) | 3 lines
Fix an image corruption issue with the raster graphics system,
and a possible performance issue.
------------------------------------------------------------------------
r942396 | ggarand | 2009-03-21 16:32:23 +0000 (Sat, 21 Mar 2009) | 10 lines
automatically merged revision 942116:
my previous attempt at merging continuations back into inlines
in removeChild was way too shallow.
this ought to work for nested inlines too, so lets walk the chain
all the way back to the root anonymous block while merging.
this should cover most cases. To get a 100% correct behaviour, we will
have to get rid of continuations and use some sort of implicit inline
containers for hosting blocks in inlines.
------------------------------------------------------------------------
r942412 | orlovich | 2009-03-21 17:08:51 +0000 (Sat, 21 Mar 2009) | 4 lines
Fix an another regression in :lang --- should be case-insensitive
(css/lang01.html)
------------------------------------------------------------------------
r942416 | orlovich | 2009-03-21 17:14:13 +0000 (Sat, 21 Mar 2009) | 2 lines
Woops, make this work with Qt4.4 too --- thanks fredrikh
------------------------------------------------------------------------
r942422 | ggarand | 2009-03-21 17:29:56 +0000 (Sat, 21 Mar 2009) | 5 lines
automatically merged revision 942117:
More regression fixing in bidi hell..
Avoid testing for line break if the lBreak iterator contains an inlineflow.
InlineFlows are never a suitable line-breaking opportunity.
------------------------------------------------------------------------
r942456 | terwou | 2009-03-21 19:14:28 +0000 (Sat, 21 Mar 2009) | 3 lines
backport r942455
Fix crash when clicking "Not Defined" qwhatsthis
------------------------------------------------------------------------
r942586 | wkai | 2009-03-22 06:23:49 +0000 (Sun, 22 Mar 2009) | 1 line
backport a patch from trunk, add a new QByteArray return type function, mark old const char* one as KDE_DEPRECATED, fix memory leak in encoding prober
------------------------------------------------------------------------
r942590 | scripty | 2009-03-22 07:41:46 +0000 (Sun, 22 Mar 2009) | 1 line
SVN_SILENT made messages (.desktop file)
------------------------------------------------------------------------
r942943 | berendsen | 2009-03-22 22:42:21 +0000 (Sun, 22 Mar 2009) | 1 line
merge differently headed/dotted on/off
------------------------------------------------------------------------
r943285 | dfaure | 2009-03-23 16:03:10 +0000 (Mon, 23 Mar 2009) | 3 lines
Backport fix for 173101: do not escape '?' in fragments. Please test if this fixes the facebook issue completely.
CCBUG: 173101
------------------------------------------------------------------------
r943287 | jacopods | 2009-03-23 16:06:22 +0000 (Mon, 23 Mar 2009) | 3 lines
Backporting 93708-9 by aseigo
Set the compression timeout to 50ms to improve consistency
------------------------------------------------------------------------
r943296 | orlovich | 2009-03-23 16:17:27 +0000 (Mon, 23 Mar 2009) | 2 lines
Dead in KDE4
------------------------------------------------------------------------
r943315 | jacopods | 2009-03-23 16:58:02 +0000 (Mon, 23 Mar 2009) | 2 lines
Backporting 938728 - 938752 - 938756 - 938778 by aseigo
------------------------------------------------------------------------
r943467 | freininghaus | 2009-03-23 21:38:17 +0000 (Mon, 23 Mar 2009) | 6 lines
Check if KBookmarkDomBuilder's m_stack is empty before accessing its top element. This fixes a crash in KEditBookmarks when importing invalid bookmark files.
This fix will be in KDE 4.2.2.
CCBUG: 176752
------------------------------------------------------------------------
r943483 | aacid | 2009-03-23 22:25:24 +0000 (Mon, 23 Mar 2009) | 4 lines
Add some i18n to kdelibs
This adds some message that were previously untranslatable
CCMAIL: kde-i18n-doc@kde.org
------------------------------------------------------------------------
r943498 | aacid | 2009-03-23 23:20:36 +0000 (Mon, 23 Mar 2009) | 4 lines
Backport r943497 kpty/trunk/KDE/kdelibs/kpty/kptydevice.cpp:
I18N_NOOP -> i18n
acked by ossi
------------------------------------------------------------------------
r943673 | scripty | 2009-03-24 08:18:12 +0000 (Tue, 24 Mar 2009) | 1 line
SVN_SILENT made messages (.desktop file)
------------------------------------------------------------------------
r943913 | dfaure | 2009-03-24 16:40:00 +0000 (Tue, 24 Mar 2009) | 2 lines
backport fix for "korganizer; korganizer" treats the cwd as a command-line argument
------------------------------------------------------------------------
r943998 | orlovich | 2009-03-24 20:04:11 +0000 (Tue, 24 Mar 2009) | 6 lines
Give the proper dirty rectangle when switching frames in animated images.
The previous code mostly worked with Qt4.4 by some happy happenstance, but failed cleanly with Qt4.5 --- yey to that.
BUG:175858
BUG:186159
------------------------------------------------------------------------
r944114 | reed | 2009-03-25 04:20:29 +0000 (Wed, 25 Mar 2009) | 1 line
don't want to do app-bundle stuff on OSX if we're using Qt/X11
------------------------------------------------------------------------
r944115 | reed | 2009-03-25 04:22:54 +0000 (Wed, 25 Mar 2009) | 1 line
X11 on OSX does not use "normal" DISPLAY syntax, it is a path to a socket file instead; handle / in display nicely
------------------------------------------------------------------------
r944116 | reed | 2009-03-25 04:24:07 +0000 (Wed, 25 Mar 2009) | 1 line
startup info no workie on OSX ;)
------------------------------------------------------------------------
r944117 | reed | 2009-03-25 04:26:26 +0000 (Wed, 25 Mar 2009) | 1 line
bad patch, this was fixed properly in trunk
------------------------------------------------------------------------
r944170 | scripty | 2009-03-25 08:45:21 +0000 (Wed, 25 Mar 2009) | 1 line
SVN_SILENT made messages (.desktop file)
------------------------------------------------------------------------
r944277 | lueck | 2009-03-25 11:46:07 +0000 (Wed, 25 Mar 2009) | 1 line
change Standardsprache to Ausweichsprache
------------------------------------------------------------------------
r944314 | reed | 2009-03-25 12:32:32 +0000 (Wed, 25 Mar 2009) | 1 line
only translate / to _ in DISPLAY on Mac OS X
------------------------------------------------------------------------
r944333 | jacopods | 2009-03-25 13:00:11 +0000 (Wed, 25 Mar 2009) | 2 lines
backporting 934733
------------------------------------------------------------------------
r944578 | fredrik | 2009-03-25 19:07:11 +0000 (Wed, 25 Mar 2009) | 2 lines
Backport r944572.
------------------------------------------------------------------------
r944623 | poberndorfer | 2009-03-25 20:44:38 +0000 (Wed, 25 Mar 2009) | 8 lines
backport r944617
rename encodingNameByteArray() to encoding()
make the old encodingName function return something usable
fix some documentation typos
add @since to encoding()
------------------------------------------------------------------------
r944661 | djarvie | 2009-03-25 22:02:41 +0000 (Wed, 25 Mar 2009) | 2 lines
Don't copy instance to itself in operator=()
------------------------------------------------------------------------
r944730 | mpyne | 2009-03-26 00:54:28 +0000 (Thu, 26 Mar 2009) | 4 lines
Backport fix for bug 186774 (broken slider in KIntNumInput) to KDE 4.2.2.
BUG:186774
------------------------------------------------------------------------
r944743 | ggarand | 2009-03-26 01:21:38 +0000 (Thu, 26 Mar 2009) | 8 lines
automatically merged revision 944736:
adapt to correction in draft cssom-view specification.
position:fixed objects where returning an offsetLeft/Top relative to the
canvas origin instead of relative to their pseudo containing-block - the
viewport.
BUG: 187787
------------------------------------------------------------------------
r944744 | ggarand | 2009-03-26 01:24:41 +0000 (Thu, 26 Mar 2009) | 6 lines
automatically merged revision 944740:
I'll special case 'position:fixed' elements after all,
otherwise the computation for other positioned elements isn't
compatible with Gecko. Better safe than sorry.
CCBUG: 187787
------------------------------------------------------------------------
r944745 | ggarand | 2009-03-26 01:25:25 +0000 (Thu, 26 Mar 2009) | 7 lines
automatically merged revision 944739:
fix regression in sizing of the background image for the root element.
this regressed in r855521, as I didn't quite understood all the subtleties
of the background size for the root case.
BUG: 187285
------------------------------------------------------------------------
r944749 | ggarand | 2009-03-26 01:38:07 +0000 (Thu, 26 Mar 2009) | 15 lines
automatically merged revision 944737:
Having the most standard compliant regular expression library thanks
to the kindness of the PCRE library author, we are now in the position of
having to add hacks for compatibility with other engines that only fixed
what's explicitely tested by the acid 3 test ;-(
Closing brackets that aren't part of a character class should be rejected
if they aren't escaped, but other engines accept that.
So upon R.E. compilation failure, we will detect such patterns, and attempt
another compilation after escaping any such brackets.
OK'ed by Maksim
CCBUG: 172383
------------------------------------------------------------------------
r944751 | ggarand | 2009-03-26 01:48:49 +0000 (Thu, 26 Mar 2009) | 14 lines
automatically merged revision 944738:
.Re-implement search in sub-frames, that was lost when switching to
the embedded findbar.
The previous implementation required a frame/iframe to have the focus
(e.g. by clicking its content) to be able to search it.
This one will walk frames in document/reverse-document order, searching
each in turn.
.Fix bug in the search algorithm that would require one two click twice on
'Previous' to effectively match the previous occurence, if the current
occurence was the last of the document.
BUG: 187475
------------------------------------------------------------------------
r944753 | dfaure | 2009-03-26 01:55:23 +0000 (Thu, 26 Mar 2009) | 6 lines
Backport r944752:
1) Fix porting bug which made the cookie-question-dialog always says "You got 2 cookies" (always 2 instances of the same)
2) Use enum instead of magic int with 0,1,2 values
3) Actually make "Apply to this cookie only" do what it says, rather than accepting any other cookie from the same host that
might have arrived meanwhile. Oops. That one was in kde3 too.
------------------------------------------------------------------------
r944782 | ggarand | 2009-03-26 03:54:24 +0000 (Thu, 26 Mar 2009) | 10 lines
automatically merged revision 944781:
at least commit a something to prevent the crash on yahoo.fr,
I'm too tired to figure what's the proper solution.
not really a regression from other changes anyway,
rather a weakness in the pseudo-element code that's been
uncovered by the re-merge of inline continuations.
I owe dfaure a box of my finest bretonne pastry for spotting
that in time - sorry for the late backport.
------------------------------------------------------------------------
r944807 | scripty | 2009-03-26 08:11:53 +0000 (Thu, 26 Mar 2009) | 1 line
SVN_SILENT made messages (.desktop file)
------------------------------------------------------------------------
r944881 | reed | 2009-03-26 12:15:16 +0000 (Thu, 26 Mar 2009) | 1 line
get rid of crufty __progname_full reference, that is not actually used
------------------------------------------------------------------------
r944889 | reed | 2009-03-26 12:43:06 +0000 (Thu, 26 Mar 2009) | 1 line
dfaure said this was still used for glibc stuff; putting it back and fixing the ifdef to actually trigger
------------------------------------------------------------------------
r944896 | dfaure | 2009-03-26 13:00:10 +0000 (Thu, 26 Mar 2009) | 2 lines
SVN_SILENT remove potentially confusing debug output (it's just a window being unregistered, not the whole module)
------------------------------------------------------------------------
r944984 | dfaure | 2009-03-26 16:14:27 +0000 (Thu, 26 Mar 2009) | 2 lines
Update version number so that other modules can start saying they require 4.2.2 (esp. if they use new api, like kdebase)
------------------------------------------------------------------------
r944994 | dfaure | 2009-03-26 16:28:48 +0000 (Thu, 26 Mar 2009) | 7 lines
Don't make the source dir "disappear" (visually, it was still there on disk) if there was an error while removing one of its subdir (e.g. wrong encoding).
This fix will be in 4.2.2 (just in time).
BUG: 187868
This is a more refined version of the fix for #118593 (retested, still works).
CCBUG: 118593
------------------------------------------------------------------------