In CppunitTest_sw_ooxmlfieldexport testFdo81492 a TextFieldmark is
inserted at a position where already another TextFieldmark starts. The
defensively programmed lcl_AssureFieldMarksSet notices there is aleady a
dummy character at the start position, and does not insert another one,
but then the dummy character for the end position is inserted, moving
the start position index, which puts the start position behind another
bookmark.
So we end up with a field mark that has a end character but not a start
character and an un-sorted m_vAllMarks.
Change-Id: Icd15e83471e18f607eb41b2f7b0c2ce61c94ff9f
The recently added tdf90697.rtf caused a libstdc++ assertion about
non-sorted m_vAllMarks. This is because a SwXParagraph was disposed, a
bookmark on it was moved by correctMarksAbsolute onto the first
paragraph in a table, and there was a SwXTextRange with a UnoMark
pointing to the SwTableNode, so the sort order was not maintained.
So if at least one mark was moved, re-sort the array.
Change-Id: I0efc7827c826e696f29f8480bab1044b31ccfe39
...from configuration (see issue for details). Also, remove mention of that
configuration tree from the documentation of the css.util.PathSubstitution UNO
service.
Change-Id: Ia0373bc7d19e650b7954b7dbb934a578c96fcab5
Multiple callers involved. Most significantly ScDocument::Broadcast()
calls ScDocument::TrackFormulas() individually. Track/collect pending
formula cells at the end of the bulk broadcast instead, which gives an
instructions read speedup by factor 6 for the broadcast, and an overall
speedup in the scenario for inserting the rows by factor ~2 wall clock
time.
ScDocument::InsertRows()
Before, Ir Incl: 282,227,033,656
After, Ir Incl: 66,307,994,805
With cycle detection:
ScDocument::TrackFormulas()
Before:
Ir Incl Ir Self
66,981,644,959 11,913,444,899
After:
Ir Incl Ir Self
10,819,556,073 1,973,232,494
Change-Id: I85fe8b03ecb52cffaa6fa14354b3cc3467ecc111
the problem being that the attempt to restore the focus in the document in the
floatingwindow tear down is blocked because the menus have set that window as
modal so the focus isn't set.
The attempt to set the focus in the floatingwindow teardown causes the
SavedFocus in the menufloatingwindow to be dropped, so the extra layer of code
to restore focus after modality is removed doesn't do anything
this is fallout from...
commit dd46727b99
Author: Caolán McNamara <caolanm@redhat.com>
Date: Tue Apr 5 15:27:38 2016 +0100
Resolves: tdf#87120 no keyboard navigation inside floating windows
which allows the focus to "truly" enter the menus, triggering the
floatingwindow attempt, which fails but blocks the menufloatingwindow attempt
easiest thing seems to make the restoring modality, and then restore focus,
something that MenuFloatingWindow does before it finishes
Change-Id: I97a4ac71419dcb04709e4667586d8343e89ddbeb
commit 9b7dfb1064
Author: Caolán McNamara <caolanm@redhat.com>
Date: Thu Jun 9 16:52:21 2016 +0100
EndSaveFocus does nothing on the !bRestore case
removed the code that made it look like it made sense
Change-Id: I2e7280fd6643e86d65cba04c8302be099d6f3d0e
A comment text edit (provided by editeng) works with relative twip
coordinates, conversion of the cursor position to absolute twips happens
in ImpEditView::ShowCursor(), provided that the pOutWin member has a map
mode origin that respresents the offset correctly.
This is not true during SwSidebarWin::DeactivatePostIt(), the map mode
origin is already reset back to 0, so just don't emit callbacks during
shutdown, the sw body text cursor will be shown later anyway.
Change-Id: I02c15bb9fad99db8e43fd2f37df770dd165be788
Since commit 32bc8ddbf3 the jvmfwk rejects
Apple JDK and while the Java code builds fine with it, various tests such as
CppunitTest_dbaccess_hsqldb_test fail if it is the only installed JRE.
The simplest way to avoid the problem is to not allow building with
Apple JDK.
Change-Id: I5701e38cffded4596ac94608867a038ff76f75ec
causes mathtype to display a warning dialog, so
try restoring them back to loaded after loading
them in order to get their preferred size
Change-Id: Idff714efa228a739f380dbae429d852a8f8c5298
Reviewed-on: https://gerrit.libreoffice.org/29234
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
...otherwise, if $(username) happens to be "user", it will endlessly re-
substitute inside a (already partly re-substituted) URL of the form
$(user)/...
Regression introduced with 16fb0d3d0f "tdf#98407
PathSubstitution: Add substitution for $(username)".
Change-Id: I1c8b64f383fdfd97fa5edc192e9ca4b46944d6f1
