the code in question try to re-assign entry that are in normal file
to links based on some condiftion... but when it did that
it would overwrite the content of the array that was passed
instead of adding to it, hence loosing any Unixlink entries.
Change-Id: Ia879dfefa4bf2fc635c05864b014e0712e51812d
Reviewed-on: https://gerrit.libreoffice.org/6574
Reviewed-by: David Tardon <dtardon@redhat.com>
Tested-by: David Tardon <dtardon@redhat.com>
Because NSS libraries are dynamic and OpenSSL static, using NSS saves
1.5 MB in the oox library [even though it's not as 1337 apparently]:
-rwxrwxr-x. 1 ms ms 8889575 2. Nov 13:45 libooxlo.so.nss
-rwxrwxr-x. 1 ms ms 7773576 2. Nov 13:45 libooxlo.so.nss.stripped
-rwxrwxr-x. 1 ms ms 10340276 2. Nov 13:37 libooxlo.so.openssl
-rwxrwxr-x. 1 ms ms 9042216 2. Nov 13:37 libooxlo.so.openssl.stripped
Change-Id: I387496ae364acb1286d753d52f04924631136750
And add Dump() to ScRangeData for debugging convenience and re-enable
previously failed test cases.
Change-Id: I9d8f41a8be4c9c301254ef300c7b7f0c1ea7f393
Also fix incorrect const methods. Methods that populate the document
model should not be marked const even if the compiler allows it.
Change-Id: Ic5d1670ce93c166d0f44ace04494fccab6eac275
Also, those createFoo() methods shouldn't be const since it does
modify the state of the document model.
Change-Id: I6a9267c54710f359506ca39c1e213f82595ebfe3
...where the previous .uno:PrintPreview request has not yet been processed, the
same way it had been made more robust for ScAccessiblePageHeader.java in
5b5c52a3a4 "Try make this more robust in slow
environments" plus b5dd2b4218 "Also cope with
intermittend DisposedExceptions."
Change-Id: I5244e8f8af9560ae986ed3a836bdf9571574420f
Arguments 1 and 2 of MMULT have the same name in the function
wizard, which is both confusing for the user and inconstent
with other argument names for matrix functions.
Patch simply changes the names to match the description of
the arguments of MMULT.
Change-Id: I9fc35907133aa61248182bd3b253a7f90ba7cd2c
Reviewed-on: https://gerrit.libreoffice.org/6526
Reviewed-by: Eike Rathke <erack@redhat.com>
Tested-by: Eike Rathke <erack@redhat.com>
If the scale settings specify that the print ranges must be scaled
across a specific number of pages, the default behaviour is to ignore
breaks to avoid the case where breaks force more pages than specified.
Here, an option under Calc -> Print -> Pages is added so that the user
can specify that manual row and column breaks should be forced.
Change-Id: I445cd7ce9e16e4ec2d0c320f059edad62b40f22d
Reviewed-on: https://gerrit.libreoffice.org/6531
Tested-by: Eike Rathke <erack@redhat.com>
Reviewed-by: Eike Rathke <erack@redhat.com>
If a print range's manual breaks forced it over more pages than specified
by the sheet scale settings, the zoom calculation wasn't able to
converge on a zoom level, so it bottomed out at ZOOM_MIN.
This issue only appears if the Calc/Print/Page/ForceBreaks option is
selected and simply ensures the minimum number of pages is at least
the number required by the breaks in the sheet.
Change-Id: Iba36e850081718b1aa43e5c3db3c883530885853
Reviewed-on: https://gerrit.libreoffice.org/6532
Reviewed-by: Eike Rathke <erack@redhat.com>
Tested-by: Eike Rathke <erack@redhat.com>
This change fixes two bugs:
1) In it's old location at PDFFilter::implExport, the pdf viewer is called
*before* the final file is actually written. It causes a problem under
Windows, because we use ShellExecute function which fails if the file
doesn't exist. Calling a function blocks execution, so we have no chance
to write the file before that function fails. (fdo#68099)
2) Being a part of the saving/filter code means that the pdf viewer will
open even when converting to a pdf in headless mode. It's definitely not
something that we want to happen.
Change-Id: I1a6f70ec76100ac9e008869aa57ad738ced3ab6c
Reviewed-on: https://gerrit.libreoffice.org/6565
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>
We need the EditEngine Modification handler of the inputbar that we are feeding
to be disabled while this dialog is open. Otherwise we end up in a situation
where...
a) this ScFormulaDlg changes the editengine
b) the modify callback gets called
c) which also modifies the editengine
d) on return from that modify handler the editengine attempts to use
old node pointers which were replaced and removed by c
We turn it off in the ctor and back on in the dtor, but if calc has to repaint,
e.g. when switching to another window and back, then in ScMultiTextWnd::Paint a
new editengine will have been created via GetEditView with its default
Modification handler enabled. So ensure its off everytime we will access it via
InputReplaceSelection pScMod->InputEnterHandler();
I wonder if we should really be calling ScMultiTextWnd::GetEditView
(which creates an EditView if one does not already exist) in
ScMultiTextWnd::Paint or just check for pEditView directly, but that
then leading to the need for it to be explicitly created somewhere else.
As it stands in ScMultiTextWnd::Paint(...) the "if ( pView )" can never
be false.
Change-Id: I2f6df9ce0cc7ec59e8be33f4e67d91422796d251
