Fixed a misguided condition in the check for bogus texts in the sscanf
branch of the decoder; it checked for 'e' but neglected 'E', which is
just as valid.
Change-Id: I9236c76faea000c92df641930e401bce445e06c8
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>
Revised some toFloat()s to be consistent with the matching
toDouble()s; previously, they would return infinity if toDouble() did
but return 0 if toDouble() got a finite value outside float's range.
That also applied to values that underflowed float's range, succeeding
and returning 0 as long as they were within double's range but failing
if toDouble() underflowed. Now float-underflow also fails. Amended
their documentation to reflect this more consistent reality.
Added some tests of out-of-range values, infinities and NaNs.
[ChangeLog][QtCore][toFloat] QString, QByteArray and QLocale returned
an infinity on double-overflow (since 5.7) but returned 0 on a finite
double outside float's range, while setting ok to false; this was at
odds with their documented behavior of returning 0 on any failure.
They also succeeded, returning zero, on underflow of float's range,
unless double underflowed, where they failed. Changed the handling of
values outside float's range to match that of values outside double's
range: fail, returning an infinity on overflow or zero on underflow.
The documentation now reflects the revised behavior, which matches
toDouble().
Change-Id: Ia168bcacf7def0df924840d45d8edc5f850449d6
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>
They actually return infinity if conversion overflows, while still
setting ok to false; they were documented to return 0 on failure, with
no mention of this special handling of overflow. Documented reality
rather than changing the behavior. Gave underflow as an example of
failure other than overflow (toDouble()s do indeed fail on it).
Added some tests of out-of-range values, infinities and NaNs.
[ChangeLog][QtCore][toDouble] QString, QByteArray and QLocale return
an infinity on overflow (since 5.7), while setting ok to false; this
was at odds with their documented behavior of returning 0 on failure.
The documentation now reflects the actual behavior.
Fixes: QTBUG-71256
Change-Id: I8d7e80ba1f06091cf0f1480c341553381103703b
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>
A couple of QLocale tests were using setlocale twice to provide a
transient locale tweak in tests; however, if the test in between
fails, that can leave the program running in the "transient" locale
after. So implement a proper class whose destructor ensures the
transient is tidied away. Also change the locale in use by one of
these transient changes: it purported to be checking things didn't
depend on locale, but was using the same local as most of the
test-cases for its test.
Change-Id: I0d954edcc96019a8c2eb12b7a7c568e8b87a41d5
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>
When calling setDocument (directly or through the constructor) a delayed
rehighlight is initiated. Previously, if any text was changed before
this rehighlight could run it would cancel the rehighlight, even if the
changed text only caused a new block of text to be highlighted.
Fixes: QTBUG-71307
Change-Id: Ib09b664d90906f5b4427105f0e45469806f3a779
Reviewed-by: Jędrzej Nowacki <jedrzej.nowacki@qt.io>
When a time-zone does a spring-forward, skipping an hour (either to
start DST or to move its standard time), there's an hour that doesn't
exist on the day in question. That hour can be the first hour of the
day, in which case using 0:0 as the default time is broken. So catch
this case and use the first time that day that makes sense.
Fixes: QTBUG-70823
Change-Id: I23dae9320a3cdd2c988841a7db1b111edb945730
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
QUrl::RemoveAuthority is RemoveUserInfo | RemovePort | 0x10, so the
condition
if (options & QUrl::RemoveAuthority)
would match if any of the other bits for the username, password or port
were set, which meant we would skip the host comparison. Ditto for
username and RemovePassword.
[ChangeLog][QtCore][QUrl] Fixed a bug that caused QUrl::matches to
incorrectly compare two URLs with different hostnames or different
usernames as equal, if certain QUrl::RemoveXxx options were passed.
Change-Id: I015970a03b874898bba7fffd155856ab9d6cb1be
Fixes: QTBUG-70774
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Before this change we inserted newline only if an option has a
description and ended up with an arbitrary long line with all options.
[ChangeLog][QtCore][QCommandLineParser] Fixed a bug that caused the help
output to show two options or more in the same line if the options didn't
have a description.
Task-number: QTBUG-70174
Change-Id: Id54b9ae13ee596869e4dc14e09301aea19eed2f8
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Change-Id: If030b56ad97e047d89d442629262b4839df306d4
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>
Both sqlite3_open_v2 and sqlite3_close are documented to return an error code:
https://www.sqlite.org/c3ref/open.htmlhttps://sqlite.org/c3ref/close.html
However, those were ignored (other than checking whether the operation
succeeded), causing QSqlError::nativeErrorCode() to always be "-1" when there
was an error while opening/closing the database.
Additionally, the error string needs to be read (via sqlite3_errmsg16) in
qMakeError *before* d->access is set to 0, or the databaseText() will always be
"out of memory" no matter what error actually happened.
Task-number: QTBUG-70506
Change-Id: I75cbf178c9711442e640afd26c4502214d20c598
Reviewed-by: Andy Shaw <andy.shaw@qt.io>
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
We have #if-ery on Q_OS_DARWIN controlling an expectation of gettign
"GMT+1" and "GMT+2" instead of "CET" and "CEST" in two tests; this
turns out to not be a deficiency of macOS so much as of how we
configure Coin's VMs. While we fix that, we need to ignore failures
in these tests, so that we can pull the #if-ery out and clear the
blacklist once the VMs are set up properly.
Task-number: QTBUG-70149
Change-Id: If3577200cf980b3329161ab3eea7bd2e9d0124e0
Reviewed-by: Tony Sarajärvi <tony.sarajarvi@qt.io>
Under certain circumstances, if you had a widget with a QOpenGLPaintEngine,
and drew QStaticText into this, and then later had Qt Quick access the same
cache and try to resize it, we would get a crash because the resize function
would have a pointer to the paint engine and try to access its shader manager
(which would now be null, since this is outside the begin()/end() phase of the
paint engine.
The solution is to reset the paint engine pointer to null on the cache once it
has been populated and it is no longer needed.
[ChangeLog][QtGui][Text] Fixed a possible crash when combining QStaticText,
QOpenGLWidget and Qt Quick in the same application.
Task-number: QTBUG-70096
Change-Id: I7383ad7456d1a72499cfcd2da09a5a808d4b3eff
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
CLDR up to somewhere between v29 (used by 5.9) and v31.0.1 (used by
5.10 and later) claimed Costa Ricans don't include fractions in their
currency; now it claims they expec two digits. Apparently one of them
does expect those digits, so this is the regression test I'll be
cherry-picking back to LTS, to accompany the CLDR updates they need.
Task-number: QTBUG-70093
Change-Id: I138772cc6013fa74de4f7c54b836cac83421eab2
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
_q_interpolate<T> subtracts 2 arguments of type T,
for unsigned types this can cause wrapping around
Task-number: QTBUG-57925
Change-Id: Iffa59f413579a3d5de8cb728fe71443d8e8a04aa
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
QAccessible::StaticText should be mapped to UIA_TextControlTypeId instead
of UIA_EditControlTypeId.
Task-number: QTBUG-69894
Change-Id: If2f8f55d2be492c02a3af5b1813ca12cf774a33a
Reviewed-by: Oliver Wolff <oliver.wolff@qt.io>
Before this change, if you try to insert a row at the bottom of QSortFilterProxyModel
the row will be inserted in the source model at position proxy->rowCount rather
than at the bottom. This causes insert at apparently random positions in the source.
[ChangeLog][QtCore][QSortFilterProxyModel]
QSortFilterProxyModel::insertRows(row,count,parent) with
row == QSortFilterProxyModel::rowCount will insert at the bottom of the source model
rather than at the row QSortFilterProxyModel::rowCount of the source model
Task-number: QTBUG-58499
Task-number: QTBUG-69158
Change-Id: Ie78416c8fbc429303b8c9c98375630e3e4d85f6d
Reviewed-by: David Faure <david.faure@kdab.com>
qt_nameprep() already reset the string to its original length to
indicate failure, but we didn't handle that in qt_ACE_do(). So make it
have a return value whcih makes it easier to handle that case and do
handle it.
[ChangeLog][QtCore][QUrl] Fixed a bug that caused URLs whose hostnames
contained unassigned or prohibited Unicode codepoints to report
isValid() = true, despite clearing the hostname.
Change-Id: I41e7b3bced5944239f41fffd1545b7274c4b419d
Reviewed-by: David Faure <david.faure@kdab.com>
Problem:
CMake Error in CMakeLists.txt:
A logical block opening on the line
.../qtbase/tests/auto/cmake/CMakeLists.txt:149 (if)
is not closed.
Broken by change 02ed1b36da
Change-Id: I6c04721edbccaa9fcdb53af92d33dfa87eeaebb8
Reviewed-by: Rolf Eike Beer <eb@emlix.com>
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
Reviewed-by: Kai Koehne <kai.koehne@qt.io>
QMessageBox window ignores QCloseEvent if it was created
with Ok button and the detailed text was set. But it can
be closed if it contains only one button.
Make it closable if there are two buttons and one of them
is the "Show Details..." button.
[ChangeLog][QtWidgets][QMessageBox] A message box with two
buttons, one of which is the "Show Details..." button,
can be closed by clicking the X button on the window's
title bar.
Task-number: QTBUG-69526
Change-Id: Iba09e38561eb3898dc2aecfd38d8519d512a71c1
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@qt.io>
When showing line and paragraph separators at an offset from the start
of the string, the end of string pointer would be incorrectly set, and
we would read past the end of the string. If any part of this memory
happened to match the line or paragraph separator, then we would
overwrite it and have a crash.
I couldn't find any reliable way to test this, since the crash depends on
the contents of the memory after the string allocated by the algorithm.
But with an overflow of 100 000 characters, I found that it crashed every
time I ran the test.
[ChangeLog][QtGui][Text] Fixed potential crash when using
QTextOption::ShowLineAndParagraphSeparators.
Task-number: QTBUG-69661
Change-Id: I17d1996b883560bacdc7ce114c8aeb2b0108faea
Reviewed-by: JiDe Zhang <zccrs@live.com>
Reviewed-by: Michal Lazo <xlazom00@gmail.com>
Reviewed-by: Konstantin Ritt <ritt.ks@gmail.com>
[ChangeLog][QtCore][QObject] Fixed a bug in setProperty() that caused a
property change not to take effect if the old value compared equal using
QVariant's equality operator, but the values were not strictly equal.
Task-number: QTBUG-69744
Change-Id: I00e04a465fcf4fc1a462fffd1547885861a07a64
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
Reviewed-by: Olivier Goffart (Woboq GmbH) <ogoffart@woboq.com>
A typo meant the color-table was not fixed. For safety fallback colors
are also made opaque.
Change-Id: I3e609882177604910c4343c86f00221a89af9078
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
Do not overwrite stdout/stderr by default, but only if requested.
This restores the behavior of QProcess::startDetached of Qt 5.9.
Task-number: QTBUG-67905
Change-Id: Idccf7b0da7bd80f88a0624286ddf2851bc974fb1
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
When the connection has been encrypted we will,
in QHttpNetworkConnectionChannel::_q_encrypted, emit 'reply->encrypted'
in which user slots can be called.
In the event that the user calls abort it will, however, not abort until
the next time it goes back to the event loop (which might not happen
until after the request has already been sent).
Task-number: QTBUG-65960
Change-Id: I96865f83c47f89deb9f644c86a71948dbb0ec0d0
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Timur Pocheptsov <timur.pocheptsov@qt.io>
A default-constructed QJsonObject has no data payload, it is only a pair of null
pointers. So, when it becomes necessary to 'materialize' such an object, a
special global emptyObject constant is used as the substitute payload. There is
a small problem with this global constant though, namely that it's is_object
flag is unset. In other words, the emptyObject is not an object, but an array.
Fix by setting the is_object flag on emptyObject.
The example code in the bug report
QJsonObject parent;
QJsonObject child;
parent["child"] = child; // 1
child = parent["child"].toObject(); // 2
child["test"] = "test"; // 3
runs into this problem on line 1. Inserting the default-constructed child means
inserting a copy of emptyObject. On line 2 a pointer to this copy of emptyObject
is retrieved and cast to an object. But it's not an object, it's an array, so
things go wrong hereafter.
Specifically, on line 3, two inserts are performed, one from operator[] and one
from operator=. Each insert increments a compaction counter. The second insert
triggers compaction (QJsonObject::insert calls Value::requiredStorage calls
Data::compact) and compaction branches based on the is_object flag. Replacing
line 3 with
child.insert("test", "test");
causes the example to appear to work since compaction is not triggered and the
JSON serializer does not look at the is_object flag. Still, any further insert()
calls would trigger compaction and memory corruption.
Task-number: QTBUG-69626
Change-Id: I8bd5174dce95998bac479c4b4ffea70bca1a4d04
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
I'm not sure why QSFPM purposefully emits dataChanged for a source
dataChanged that triggers a layoutChanged (i.e. due to sorting, multiple
rows are moving around). (This predates the git import in Qt 4.5.)
Surely whoever is listening will not gain much from the "small" dataChanged
after the "big" layoutChanged... anyhow, this documents the current behavior,
at least.
It also proves that the bug I saw long ago (changing a filtered-out
value used to emit dataChanged(invalid, invalid), IIRC) is no longer present.
Change-Id: I8975c549db88226b2b3393de9f8dca4f4109df15
Reviewed-by: Christian Ehrlicher <ch.ehrlicher@gmx.de>
Reviewed-by: Thorbjørn Lund Martsum <tmartsum@gmail.com>
The copy-assign operator tests against other.d being NULL but the
copy-constructor didn't. This can only matter if the value being
copied has been moved from, so we could probably replace with an
assertion in practice, but we should at least be consistent.
Amended test to check this case too; and verified new test crashes
without this fix.
Change-Id: I46872a677775944bbdf6a9112e719873e574ae60
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
All test cases use QPluginLoader which will not work for static builds of
Qt.
Change-Id: I7dcddcd5213681bd3ae4cd85e49ee0bb5748f687
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
The code I introduced in 4ee7425794 only
dealt with systems that reasonably used a 64-bit off_t parameter. Turns
out that we don't turn on largefile support on 32-bit Android, which
meant that the fix caused a regression.
[ChangeLog][QtCore][QFile] Fixed a regression that caused QFile::map()
to succeed or produce incorrect results when trying to map a file at an
offset beyond 4 GB on 32-bit Android systems and on some special Linux
configurations.
Task-number: QTBUG-69148
Change-Id: I2c133120577fa12a32d444488bac3e341966f8d7
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Using QFileSystemEntry::isAbsolute() broke handling of resource paths.
Extended QDir::absoluteFilePath() tests to cover absolute resource path
and some UNC variants also resolved in the same fix.
Amend existing filePath tests to use drives where needed.
Task-number: QTBUG-68337
Change-Id: I4f02cf67828ad93e562857118f8442037f18bab7
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Eight bytes into the Binary JSON header there's a 32-bit little-endian
size, which qJsonFromRawLibraryMetaData uses to determine the size of
the stored metadata. That value is passed as a size to QByteArray, which
means certain values could cause crashes due to being too big or via
sign-extension in 64-bit.
[ChangeLog][QtCore][QPluginLoader] Fixed an issue that could cause a
crash when certain damaged or corrupt plugin files were scanned.
Change-Id: I117816bf0f5e469b8d34fffd153dc5425cec39a7
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
CSS geometry constraints such as "min-height" or "min-width"
will set size constraints on the widget.
Removing the stylesheet should remove these constraints.
Task-Id: QTBUG-69418
Change-Id: I1008e4390281c90112303d72dd7d59a8acddfcd9
Reviewed-by: David Faure <david.faure@kdab.com>
Reviewed-by: Olivier Goffart (Woboq GmbH) <ogoffart@woboq.com>
When using actual memory allocation, limit to 64 MB, not the full file
size. On most systems, the memory map technique will work, so this won't
even be tried. In any case, we don't need the fix for the OOM situation
that was applied in commit e211ab76d7.
As for the memory mapping technique, this commit limits the allocation
to reasonable values given the virtual memory addressing space. Half a
gigabyte is probably acceptable on 32-bit systems, where there should be
a contiguous space for the OS to allocate the file in. This commit also
fixes an overflow when converting from qint64 of the file size to ulong
(32-bit on 32-bit platforms and on Windows).
For 64-bit systems, we currently limit to 1 TB.
Change-Id: I117816bf0f5e469b8d34fffd153dc1705a8eedc4
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
When a screen is positioned to be above another, then the toolbutton
menu could be shown on the wrong screen if it was deemed to not fit
below the button the same screen. This ensures that it shows it on the
right screen when there is sufficient space to do so.
Change-Id: Ia626b28f74c1931904ff5b30cca17e63914d3c79
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@qt.io>
Handle hard zero independently in each coordinate, otherwise hard zero
is never equal to anything but itself.
Task-number: QTBUG-69368
Change-Id: I8b1131472bb92efc706a04e0b067e2211a5ccb0c
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
QSKIP() discards the whole test it appears in; so is not the right way
to announce that (and why) the test has just skipped a few sub-tests.
This was concealing a later failure on macOS, here fixed.
Change-Id: I9b07208413b9e101569a22505ad41f07ade4062b
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
QSKIP() discards the whole test it appears in; so is not the right way
to announce that (and why) the test has just skipped a few sub-tests.
This was concealing a later failure on macOS, here fixed.
This matches an earlier fix for tst_QDateTime.
Change-Id: Idaf34a9d60d84202fd41d15455209457cc281f60
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
