... creating new one in RecreateStartTextFrames().
This was causing a problem with duplicate call of AddToListRLHidden() on
a node that was already added by the existing MergedPara; logically the
add would need to be reverted when the existing MergedPara dies, but
there's no count on it, so the safe way to fix this is to delete
existing MergedPara first.
This was on node 1083 in the bugdoc.
Change-Id: I1fec613829b3b12e3846fc4b195559d9a6079560
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/99523
Tested-by: Jenkins
Reviewed-by: Michael Stahl <michael.stahl@cib.de>
This reverts commit ad0351b849.
It turns out that this conflicts with commit
14e87a4b15 "tdf#133967 sw_redlinehide: fix
redline deletion of table containing redlines"
but oddly enough that fix fixes tdf#132944 as well, but both commmits
combined don't work e.g. on node 1083/1084 in the tdf#133967 bugdoc we
got no text frame because SwUndoDelete has deleted it, but since it's
not merged it would be left alone by the new fix anyway.
This fix was quite hacky anyway, so just revert it.
Change-Id: Id17f3575c742b8761a8458914a28ae8047360fff
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/99522
Tested-by: Jenkins
Reviewed-by: Michael Stahl <michael.stahl@cib.de>
This reverts commit 51d5de51b21b13e2fac92ec53522a48896fbd3da.
Because fix for tdf#132944 will be reverted too.
Change-Id: I3b4d5031be0d39435aa39867239db549d2f7b71e
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/99521
Tested-by: Michael Stahl <michael.stahl@cib.de>
Reviewed-by: Michael Stahl <michael.stahl@cib.de>
...split from -fuse-ld with <https://github.com/llvm/llvm-project/commit/
1bc5c84710a8c73ef21295e63c19d10a8c71f2f5> "[Driver] Add --ld-path= and deprecate
-fuse-ld=/abs/path and -fuse-ld=rel/path", and now causing warnings (or even
errors with -Werror) like
'-fuse-ld=' taking a path is deprecated. Use '--ld-path=' instead
when --enable-ld is configured as a full path. (--enable-ld was vague whether
it supports full paths, but it appeared to work reasonably well at least with
old versions of Clang.)
Change-Id: I5a7dfd992b56aba78dd3646045fb9a827dc40321
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/99569
Tested-by: Jenkins
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
Just store the last used menu id, and consistently get all the
information (command name, image id, target frame etc.) from the menu
instance. Also implement the XSubToolbarController interface, so we get
a chance to update the image on icon size or theme change.
The initial action and icon of the button are based on the url passed
by the .uno:AddDirect status update, given that there is a corresponding
menu item. The only defined fallback now is to .uno:AddDirect itself for
both executing and getting the icon, but this shouldn't happen normally,
as I fixed all the affected cases I found so far, namely Base (including
windows like table design etc.) and the Basic IDE.
Actually these cases were partially broken anyway, and as a result now
completely fixed:
- In Base main app window, the button used to default to a new Writer
document, although its tooltip advertises Ctrl+N as the kb shortcut, and
pressing that shortcut actually opens the database creation wizard (as
expected).
- In the Basic IDE, the button was working, but not the Ctrl+N shortcut.
Change-Id: I6537b3f38e7d7001ec1b39d78285445464fd8cbe
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/99514
Tested-by: Jenkins
Reviewed-by: Maxim Monastirsky <momonasmon@gmail.com>
Regression from commit a2f85c062a
(tdf#130362 sw: fix anchoring of inline math objects, 2020-02-04), the
problem was that the SwFlyFrameAttrMgr ctor wanted to set the anchor
type to at-char, and then later
DocumentContentOperationsManager::InsertEmbObject() wanted to undo that
for math objects, but this did not play nicely with objects imported
from DOCX.
So don't set and clear the anchor type, rather set it conditionally in
the first place. This allows setting the anchor type in writerfilter/
before insertion, and then all of 1) docx import (depends) 2) insert of
Math objects (as-char) 3) insert of images (at-char) are working.
Change-Id: I94d82c12f30d069426db1bab70c456cadf1c91ef
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/99559
Reviewed-by: Miklos Vajna <vmiklos@collabora.com>
Tested-by: Jenkins
not its width, but its width and the the width of the corner square
or just take its position if visible
Change-Id: Ideaea789046fb03067501798b5a541985ef75e0e
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/99485
Tested-by: Jenkins
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
IsFootnoteAllowed() removes the footnote marker from
the repeated table headers, but it didn't remove the
view menu -> field shadings background.
NewExtraPortion() is the only function that calls
NewFootnotePortion(), and it will create a blank
FieldPortion is nullptr is returned, so return a
null FoonotePortion instead so that we can test
for is-footnote and then prevent the drawing
of the fieldshadings.
FieldShading is an on-screen item only and never printed,
so no unit test is possible.
Change-Id: I940eaf2dc3f4ad808b333d46f0964df31a9345b1
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/99423
Tested-by: Jenkins
Reviewed-by: Michael Stahl <michael.stahl@cib.de>
Reviewed-by: Justin Luth <justin_luth@sil.org>
The change done for tdf#134237 was meant to avoid caching of very
large images when zooming in. But it turns out that it happens
quite often that we get asked to draw something larger than
the drawing area, such as when scrolling a document with a larger
image. So tweak the check to hopefully avoid only huge images
or when upscaling a lot.
Change-Id: Ifa76e0ef2df78085afc21fd064b77d5b1c9513b5
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/99482
Tested-by: Jenkins
Reviewed-by: Luboš Luňák <l.lunak@collabora.com>
* Update translations from branch 'master'
to d0e75e01d08cb94b06a00110a0dd16404bec760c
- update translations for master
and force-fix errors using pocheck
Change-Id: Iba2b940b559a295467b1a9e77ef87cadb24633e7
Since SkiaSalBitmap is required to support bitmap formats that
Skia itself doesn't support (1bpp, 4bpp, 24bpp,etc.), it needs
to store its own copy of the bitmap data and convert to SkImage
for use. With huge images this can lead to large duplicated
memory usage. Since the code can actually drop the buffer and
create it if needed (which it normally is not), just drop large
buffers after converting to SkImage.
Ideally SalBitmap should be able to say which bitmap formats it
supports and VCL code should oblige, which would allow reusing
the same data.
Change-Id: I98af62a51dde5d738cc8afcdd2fcdc25ff89952f
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/99476
Tested-by: Jenkins
Reviewed-by: Luboš Luňák <l.lunak@collabora.com>
The timeout used is still the original 90ms, not the attempt to use 22ms so
this doesn't achieve anything. Making 22ms stick is way too fast, so leave
this at the value that is currently actually used.
Change-Id: I7207dd9adebf4e0f7f4e486a71f388bc6760d939
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/99470
Tested-by: Jenkins
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
OOXML formatting code support with Hungarian "félkövér" (semibold)
and "dőlt" (italic), similar to German, according to formatting
code standard of OOXML 18.3.1.46 headerFooter (Header Footer
Settings): "Both font-name and font-type can be localized values."
Co-authored-by: Attila Szűcs (NISZ)
Change-Id: I95de548fe27d4a6c4e9504dc1f746c8d4419222b
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/99016
Tested-by: László Németh <nemeth@numbertext.org>
Reviewed-by: László Németh <nemeth@numbertext.org>
Before, when align function was used the textboxes
(shape+frame) went apart, now the textframe follows
the shape.
Co-dev: Attila Bánhegyi (NISZ)
Change-Id: I4d69a2f7f4e5ef50d17bb0871c501d6e0026d0e2
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/99048
Reviewed-by: Attila Bakos <bakos.attilakaroly@nisz.hu>
Reviewed-by: László Németh <nemeth@numbertext.org>
Tested-by: Jenkins
Tested-by: László Németh <nemeth@numbertext.org>
A delete redline is created from node 371 to node 1625; there are
existing redlines in the range, so AppendRedline() splits the new
redline into pieces between and on the existing redlines.
Even worse, because some of the existing redlines are in a table,
InsertWithValidRanges() then splits the pieces that overlap the table
into even smaller pieces, such that there are now the table structural
nodes that were in the initial new delete redline's range but are now
covered by none of the pieces.
This means that multiple merged text frames are created, with non-merged
frames between them, and thus the functions
UpdateFramesForAddDeleteRedline() and
UpdateFramesForRemoveDeleteRedline() [for Undo] have to deal with this
unexpected complication.
(regression from sw_redlinehide)
Change-Id: Ibf71c1deb2bc3fa2af16035954d5ef71b6827be5
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/99411
Tested-by: Jenkins
Reviewed-by: Michael Stahl <michael.stahl@cib.de>
Don't delete frames on start node in SwUndoDelete::UndoImpl; happens
on SwUiWriterTest::testTdf83260().
(regression from ad0351b849)
Change-Id: I271d8bcbe4ab266e9e922e144450f05a24d183cb
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/99410
Tested-by: Jenkins
Reviewed-by: Michael Stahl <michael.stahl@cib.de>
This happens on tdf#133967 bugdoc with the Delete/Delete case first, in
line 1445 - the start positions are equal, so the new redline is empty
=> DocumentRedlineManager.cxx:92: redline table corrupted: empty redline
Similar bugs exist for Delete/Insert case in line 1725 and Format/? case
in line 1892, the latter even checks a nonsensical condition
*pEnd != *pRStt that is always true for Outside case.
It's like that since inital CVS import.
Change-Id: I7ade25380a5a43b14e87db37da8fc84267e89dd2
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/99389
Tested-by: Jenkins
Reviewed-by: Michael Stahl <michael.stahl@cib.de>
So I tested a lot of stuff, but missed the missing focus for new
dialog windows :-( This is IMHO a far worse problem, then the
popup, so just revert the fix.
This reverts commit 06fb3e7b7f.
Change-Id: I391ad91ded90e4518ab024572d3f182769b0bcda
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/99451
Tested-by: Jenkins
Reviewed-by: Michael Weghorn <m.weghorn@posteo.de>
To save PDF annotations, we need to allow a .uno:Save if the
document was opened from a PDF file. When we get a .uno:Save
command, we need to divert that to SaveAs into the same file as
the current document was opened with.
Change-Id: I0c511c4e5501de03ea8b0efb5aaf37cd09936e6e
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/99463
Tested-by: Jenkins
Reviewed-by: Tomaž Vajngerl <quikee@gmail.com>
This reverts commit a26a5bb71c.
Reason for revert: apparently compilation with installation sets wasn't tested at all/the change is incomplete and breaks because of missing files.
Change-Id: Ied6973659bd0c25f613d615a4cae699b04ae2a43
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/99398
Tested-by: Christian Lohmaier <lohmaier+LibreOffice@googlemail.com>
Reviewed-by: Christian Lohmaier <lohmaier+LibreOffice@googlemail.com>
See tdf#74608 for motivation.
And adjust the sfx2 model factory code to cope with UNO
constructor functions.
Change-Id: I7e57fb3136ad0b3caadd511ea63cf98d3c03ab3d
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/99446
Tested-by: Jenkins
Reviewed-by: Noel Grandin <noel.grandin@collabora.co.uk>