The original OUString is destroyed along with its containing
object part way through the call to PaletteManager::PopupColorPicker(),
so a copy must be taken
Change-Id: I8417ff23a17003e80a4f65e0cf1ad2a8dda5386c
Reviewed-on: https://gerrit.libreoffice.org/12031
Reviewed-by: Noel Grandin <noelgrandin@gmail.com>
Tested-by: Noel Grandin <noelgrandin@gmail.com>
This reverts commit 53c836701e.
because it means that all disabled buttons get their "click" delivered
so disabled buttons are only visually disabled.
e.g. writer: insert->table->autoformat, click on the disabled "rename" or "add"
or "remove" and the sub dialog appears. Or in calc insert->object->chart and
click on the disabled "back" -> crash
The VAxisBase::m_xNumberFormatsSupplier refers to the ChartModel itself,
and apparently that is a cyclic reference. Naively using the
ChartModel's m_xNumberFormatsSupplier in
ChartView::impl_createDiagramAndContent() because it will later be passed
to AxisHelper::getExplicitNumberFormatKeyForAxis(), which expects to be
able to convert it to a ChartModel.
Since passing around the ChartModel as an XNumberFormattingSupplier is
sort of un-intuitive anyway, refactor some methods to use XChartDocument
instead, and only create the VPolarAxis / VCartesianAxis with the
ChartModel's m_xNumberFormatsSupplier.
The drawback is that if ChartModel::attachNumberFormatsSupplier()
is called after ChartView::update() has created the axes, it may not
have an effect on them; not sure if that is a real or hypothetical
problem.
Change-Id: Ib5f0d5882b85adaf44f80e086f19178b3e64882f
Initially this didn't help at all, but with some tweaks it now finds
the leaked ChartModel as the #3 and #4 printed stacks, with a very bad
score of 40.
Change-Id: I918bcb64bad0b9432bf2b670b03915836fb34a11
Also:
* Something I missed from the HIG: standardize Help button position
Change-Id: Iaac65bd277d87ff58883f287808a5c5b804cc04e
Reviewed-on: https://gerrit.libreoffice.org/12015
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>
...not allowing certain o3tl::is_typed_flags::Wrap conversion operators to be
explicit, thus causing ambiguities here.
Change-Id: I5108885229eb185ceac57042e67405b393cdcf3b
... for now"
This reverts commit 599473a13e3c5ba669d995c913ad415bd674a16c.
moggi says the unspecified crash happened with the tests following the
comment in chart2import.cxx:75, but the tests run fine in valgrind now.
The commit creates a memory leak in the CppunitTests as
SfxBaseModel::dispose() returns early, because mbDisposed is set.
Change-Id: I987dce6a4090784d166a4a3b2e7f9e2c0cddd289
We used to ignore this element with a "false" logical attribute, but
that causes a problem when an inherited style wants to explicitly
disable this element from a parent style.
Change-Id: I2fc4c5b2002987d44679d863ea3f19114b9ab7a4
This reverts commit a2deca9d75: config_map had
been introduced on purpose with bfb978334c
"configmgr: faster / simpler compare for keys."
Change-Id: I3ae0edc4d23977c6537bd2f462f90a702266de6d
