| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
|
|
| |
Amend fd5f83e612729cebc5395c992bd98628bb9ea25f
calling fetchMore in create_mapping was a bad idea bacause it may lead
to infinite recurtion
Make a special case for hasChildren instead
Task-number: 250023
Reviewed-by: Marius Bugge Monsen
BT: yes
(cherry picked from commit f8fba0d48a1f30540ddf15f0d36f415b192d1d8b)
|
|
|
|
|
|
|
|
|
| |
QSortFilterProxyModel::hasChildren need to construct the mapping. And when it
tries to construct the mapping, it needs to fetch the childs, so there is none.
Task-number: 250023
Reviewed-by: Marius Bugge Monsen
(cherry picked from commit fd5f83e612729cebc5395c992bd98628bb9ea25f)
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
same.
updateBoudingRect update the item only if the boundingRect change
but if we have 123 as an initial text and then we set 321 as the new
text, then nothing happen because the rect is the same.
In case the boundingRect change then we call update 2 times but
the item is already dirty so the second call will just return.
BT:yes
Reviewed-by: Andreas
(cherry picked from commit 36b16a126e24cb8d8e5c34cfd807bc7c51ea49cf)
|
|
|
|
|
|
|
|
|
| |
qt_accHotKey() was pretty buggy; it could both crash or spin forever
in some cases.
Task-number: 221731
Reviewed-by: alexis
(cherry picked from commit b33ebce3de3efd98a45c8ca0a349f78aac09c875)
|
|
|
|
|
|
|
|
|
| |
The submenu would always appear to the side of the menu instead of its
right.
Task-number: 250673
Reviewed-by: ogoffart
(cherry picked from commit 8f223e03effba1aef6be5e50caf7482af1d7dd8a)
|
|
|
|
|
|
| |
This reverts commit 66fe4167b123c644874216af4f1c1a2fb3c478b2.
Oops, missed the bit in the comment about not fixing for a 4.5 patch
release.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Regression caused by optimizations in QGraphicsItem and QGraphicsScene.
The changes in QGraphicsItem fix bugs in QGraphicsItem::mapToParent
functions, which did the translation before applying the transformation,
instead of the other way (transform, then translate). This bug caused
almost all mapToParent and mapRectToParent functions to behave wrongly.
Unfortunately the new helper functions in QGraphicsScene for discovering
items made use of these functions, which introduced a regression. Fixing
these functions also fixes item discovery.
The other part of this change fixes a regression caused by c1909321,
which luckily happened after 4.5.0 and never saw the light of day. The
fix is to also invalidate the cached clip path even if there is no scene,
which is necessary if you build your scene graph outside the scene, and
finish off by adding the root item to the scene.
Task-number: 250680
Reviewed-by: Alexis
(cherry picked from commit 7a0f9ae94ca7a47dc285431d31f6839c07870194)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This change shows a limitation in Graphics View caused by QPen's
default width being 0 (cosmetic), while Graphics View actually
does not support cosmetic pens at all. Because items are at risk of
drawing lines that poke 1 pixel outside their bounding rect,
QGraphicsView must look for items that are up to one pixel
larger than their bounding rect mapped to viewport coordinates.
Furthermore, mapToScene(QRect) forces us to adjust the
input rectangle by (0, 0, 1, 1), because it uses QRect::bottomRight()
(etc) when mapping the rectangle to a polygon (which is _wrong_). Since
this behavior has been there since 4.2, we don't want to fix it in
a 4.5 patch release...
The only _proper_ fix to this problem is for the view to know the item's
"adjust" in device coordinates, allowing items to use cosmetic pens
freely. Fex, we could introduce QGraphicsItem::viewportMargins() or so.
Added an autotest to ensure this doesn't break again.
Reviewed-by: bnilsen
(cherry picked from commit cc18633fe45d599bfeac2a8b2737d155f1dd5564)
|
|
|
|
|
|
| |
RevBy: Andreas
AutoTest: included
(cherry picked from commit d516e5fbed3a7eac20229ead34221c732f85cdb6)
|
|
|
|
|
|
|
|
|
| |
RevBy: bnilsen
AutoTest: Bench
Details : if we come from setPosHelper (so itemChange) we don't need to do all the stuff regarding the size in setGeometry because the size doesn't change.
I remove two calls to fullUpdateHelper and update() because prepareGeometryChange already call updateHelper and setPosHelper call fullUpdaterHelper too so we don't need to call them inside setGeometry.
We can only call prepareGeometryChange only if we don't come from setPos.
(cherry picked from commit d74f1a91b05b943c1a8ae7847de6ee50b2093b89)
|
|
|
|
|
|
|
|
|
|
|
| |
When QString::replace was optimized, this specific overload missed out
on sanity checking of the arguments.
Task-number: 249517
Reviewed-by: Joao
Reviewed-by: hjk
BT: yes
(cherry picked from commit afb3848e0abe0cd2355e7ee7319fd904288e4636)
|
|
|
|
|
| |
Reviewed-by: ogoffart
(cherry picked from commit a94b601866740e483f093233f59f43b022a68735)
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The problem here is that we are filling the sceneEventFilters map
when we install evenfilter but we never remove the references of an
item if it has been removed from the scene or deleted. The deletion can
keep stale pointers into the map and a crash can happen.
BT:yes
Task-number:250272
Reviewed-by: bnilsen
Reviewed-by: andreas
(cherry picked from commit 0bb526f8fab33a42df56c8a60c272e2cca4cc792)
|
|
|
|
|
| |
slower platforms (like WinCE) need some more time to actually update.
(cherry picked from commit 9a41b437be6014c7a671591e2741c5ca7d8c8be8)
|
|
|
|
|
| |
add additional file to deployment.
(cherry picked from commit e2fa120510584327b7e46ff2a4413aa659a5bd65)
|
|
|
|
|
|
|
|
| |
Discovered in Kopete trunk
BT: yes
Reviewed-by: Thierry
(cherry picked from commit 91aeceb7132f4afae0996443a162a5783d8f17aa)
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Zero timers on Windows would continue to fire even after being stopped
as long as a new timer was started that reused the pointer address of
the zero timer. Fix this by only re-firing zero timers if the zero
timer hadn't been stopped (we can check this by looking at the
inTimerEvent flag, which is set to false by registerTimer()).
Task-number: 247401
Reviewed-by: Denis Dzyubenko
Reviewed-by: Prasanth Ullattil
(cherry picked from commit 74f092408d1a870e2c1e2a49182c8e09f952055d)
|
|
|
|
|
| |
Reviewed-by: TrustMe
(cherry picked from commit df461a4a1bc7492729f701d38c81916b1be386b0)
|
|
|
|
|
|
| |
This change was performed by a script that was written by the
previous release manager and reviewed by me. Thus
Reviewed-by: Trust Me
|
|
|
|
|
|
|
|
| |
RevBy: Joerg
AutoTest: mediaobject
Details: Since our wave files on Windows CE are very short (memory) we actually land up in the PausedState
when playback is finished
(cherry picked from commit 0ea43e7d28815c133e8029f3d966871a10fef8e0)
|
|
|
|
|
|
|
| |
Reviewed-by: joerg
QTest::newRow only accepts char* and without Qt3Support there is no implicit cast available.
(cherry picked from commit b8dcae1572651774085024ddf4cb02e4a954bcd7)
|
|
|
|
|
|
|
| |
Reviewed-by: thartman
function declaration was missing arguments as done in the other testcases in 831d2742b7c41924f052acd81620e8bfc58afde7
(cherry picked from commit e517ecc9025b68179c67a383791eefbedfee0543)
|
|
|
|
|
| |
Reviewed-by: joerg
(cherry picked from commit a93551a2e3e590400b09bc076d3a6883c162b75d)
|
|
|
|
| |
(cherry picked from commit 6393b9fa8474b7b9e86319f54477cba9bec65d11)
|
|
|
|
|
| |
Reviewed-by: jbache
(cherry picked from commit 6ce4a3b66e2b31f80d1a2d09a4c4087f88a7a9e8)
|
|
|
|
|
|
|
|
| |
Use qmake -query QT_INSTALL_HEADERS at test startup time instead of
relying on the dead QTDIR environment variable.
Reviewed-by: Bradley T. Hughes <bradley.hughes@nokia.com>
(cherry picked from commit d071b3528f51aaded18029f2f22e8db03e47a2a6)
|
|
|
|
|
|
|
| |
Remove unused QTDIR tests.
Reviewed-by: Bradley T. Hughes <bradley.hughes@nokia.com>
(cherry picked from commit 10ba12c7b73b89761a43a450c2932c8a56bf64db)
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
We have to pass the correct include paths when testing the multiple
inheritance warning, because we want moc to find the QtGui and QObject
headers.
In addition we have to adjust the line numbers and include QObject in
the qobject test to avoid a warning about the unknown QObject
superclass.
Reviewed-by: Bradley T. Hughes <bradley.hughes@nokia.com>
(cherry picked from commit 73cffd26dcfbfee48caf86329d636361d02e5c03)
|
|
|
|
|
|
|
|
|
|
|
|
| |
The main problem is that enterDirectory received a QModelIndex that can
be a proxy index and not a source index if there is a proxy set on the
QFileDialog. This fix basically discover some other crashes in the
completer. The problem was that we didn't apply the proxy to the
completer too. So we basically messed up source and proxy indexes.
Both have a proxy set now and convert the model indexes if needed.
Task-number:250194
Reviewed-by:jasplin
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Since there is no way to find out if gradient stops have been added to a
gradient, a gradient stop at NaN was earlier used in the SVG module to
mark the gradient as empty. This could cause floating point exceptions.
The usage of NaN has now been replaced by a boolean in
QSvgGradientStyle.
Duplicate entry "qxmlpatterns" was removed from auto.pro in the auto
test directory.
Task-number: 250146
Reviewed-by: Samuel
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Commit 4a00810cc394b7da99ddb2d13eb045352ede25d3 ensures
that setting the focus policy of a widget applies to its
proxy (if any) as well. This propagation effect is however
unfortunate for a QCompleter in PopupCompletion mode.
The client widget (typically a QLineEdit or a QTextEdit)
acts as the focus proxy to the popup menu. Internally,
the completer sets the focus policy of the popup menu to
NoFocus, but this should not affect the focus policy of
the editor.
Reviewed-by: janarve
Task-number: 250064
|
|
|
|
|
|
|
| |
Task-number: 244500
Task-number: 244485
Reviewed-by: ossi
Reviewed-by: thiago
|
|
|
|
|
|
| |
... without having to change the parent process's environment.
Reviewed-by: mariusSO
|
|
|
|
|
|
|
|
|
|
| |
And some indentation fixes in qmake itself.
Fixes indentation; TRUE => true; FALSE => false; #includes and #defines
cleanup; removes dead code; comments... and somewhere along the way
marked a function static.
Reviewed-by: mariusSO
|
|
|
|
|
|
|
|
|
| |
Qt3 support prevented the tests from running on Mac Cocoa.
Also started some Spring cleaning (it's just around the corner!) and
applied some YAGNI :-)
Reviewed-by: NRC
Reviewed-by: mariusSO
|
| |
|
|
|
|
|
|
| |
... but test against the test server.
Reviewed-by: ogoffart
|
|
|
|
|
|
| |
Make db autotest data set generation more consistent, slightly faster.
Also, add ability to easily generate a set of a single databases
entries worth of criteria to test.
|
|
|
|
|
|
|
|
| |
Handle cases where QProcess::start() and startDetached() are passed a
null, empty, or "invalid" program string instead of crashing.
Task-number: 247865
Reviewed-by: Andreas Aardal Hanssen
|
|
|
|
|
|
|
|
|
| |
offsets not handled correctly
Now also removes negative timestamps.
Task-number: 249059
Reviewed-by: Justin McPherson
|
|
|
|
| |
Signed-off-by: Peter Hartmann <peter.hartmann@trolltech.com>
|
|
|
|
|
|
| |
Signoff-by: Thiago Macieira <thiago.macieira@nokia.com>
Signed-off-by: Peter Hartmann <peter.hartmann@trolltech.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Test if we're not past the end of the array before we attempt to read it.
Dates without timezone information are taken to be UTC, so merge the code.
If the date parsing failed, don't add a malformed cookie. Better to
have no cookie than to have it for past its expiration date.
Also update the autotests, since we now can cope with some of the
invalid cookies seen on the net.
Signed-off-by: Peter Hartmann <peter.hartmann@trolltech.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
According to our documentation we should ignore leading and trailing
whitespaces when converting a string to number with QLocale::toInt and
similar functions. However that didn't work for some locales - for
those ones that declare groupseparator as 0xa0 (which looks similar to
space) since we provide a workaround to accept space as a group
separator for those locales. And since the workaround was there for a
long time it doesn't make sense to change the behavior and the fix is
to explicitely remove leading and trailing whitespaces before doing
any conversion.
Reviewed-by: mariusSO
|
|
|
|
|
|
|
| |
Call isEmpty() on the normalized rect instead of the original rect.
Task-number: 247505
Reviewed-by:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This may happen if you only have something like
QAbstractScrollArea:focus { background: foo; }
and you do not have a background set for the general case
and the viewport has a different background role (such as QTextEdit,
QAbstractItemView, ....)
This is a regression from 4.4 since in 4.4 the :focus has no effect if
you didn't have a background for the general case.
Reviewed-by: bnilsen
Task-number: 188195
|
|
|
|
|
|
|
|
|
|
| |
When drawing ellipse without stroke we need to use the same width/height
for the fill as we would use for the stroke for the stroke and fill to
match. Filling first and then stroking should produce the same result as
filling and stroking in one go.
Task-number: 249490
Reviewed-by: Trond
|
|
|
|
|
|
|
|
|
|
|
| |
qscriptvalue_cast implementation was changed from 4.4 to 4.5 because
of the introduction of QScriptValue constructors that don't take an
engine pointer. However, when the old constructors are used, the
behavior of qscriptvalue_cast should be as before, which this patch
ensures. In short: If we have an engine pointer, use it.
Task-number: 248802
Reviewed-by: Ariya Hidayat
|
|
|
|
|
|
|
|
|
| |
The pieceTable member was only initialized by QTextDocumentPrivate
_after_ the call to createObject. This patch initializes it at
construction time.
Task-number: 246138
Reviewed-by: Simon Hausmann
|
|
|
|
|
|
|
|
|
|
|
|
| |
The testlib subtests are conceptually not standalone unit tests, rather
they are helper programs for the real unit test, tst_selftests.
So, as is done for other similar tests (e.g. qprocess), don't name them
with `tst_'.
This makes it easier for automated tools to figure out that they
shouldn't directly run these tests.
Reviewed-by: Shane Bradley
|