Age | Commit message (Collapse) | Author |
|
Change-Id: I67fea6c4749b2c69868132bacd761e90933ed84b
|
|
Change-Id: I78a461609bc85cff3cc44dcdf993b16c0a1e37b1
|
|
Change-Id: I9b771bb9b57ba107e6dadb7cf85d44815f487261
|
|
Change-Id: Ic0ecaad629cfaed7d249852db24fd24d2020686b
|
|
Change-Id: Ib3a7a90174b14c77d01285430151098142815867
|
|
Change-Id: I0e6d87db90a60b300f68d83c5f5d2b1ba98f420c
|
|
Change-Id: I3c2fd7eb1ccf445f7f389f7400784748a80aa929
|
|
Change-Id: If4d2c45a4c819d17043b5dcb89a184f470a8fe57
|
|
Painting code for entry in SvTreeListBox switches between default
and highlight font (which is only the default font with a different
colour). This is done using SetFont which assumes that the font
changes entry dimensions, which then recalculates everything. This
is done for every entry so it is slow if there are a moderate number
of entries. Fix here is to call Control::SetFont - we can do this
because we know that switching between highlight and default font
wont change the font dimensions.
On the other hand - this highlighting entries when selected is
(at least on linux) already supported by the native widget.
Change-Id: I12d32f6f7b43e669442e462ad4dd80cde24b9e74
|
|
By default android will use the title and icon of the the starting
activity in the launcher. This can be overriden using an activity-alias
as the default activity, with its own title and icon.
Change-Id: I794b6738fdc383879b3c056c61f62fecbb268c5a
|
|
Change-Id: I87e5000424bed95fd25951b398bd61c3939a97ae
|
|
$one_cab_file variable was always 0, so I deleted the code dealing
with the non-zero case.
Change-Id: I17288070eb94dec00ce970585046da11136b78eb
|
|
$use_packages_for_cab and $cab_file_per_component variables were
always 0, so I deleted the code dealing with the non-zero case.
Change-Id: I0a678524cb308d834e4d61f2156a92400c02ae99
|
|
Change-Id: Ibcaf04447507926a97eed3d1cea400ac589023c2
|
|
Change-Id: Id4a89bd131f37b6844046803aae9b2f6e081af0d
|
|
Change-Id: I53a415e1ec573d342e06d0cf10bad6f8cb95b478
|
|
Previously cell-margin was unconditionnaly used to compute table position.
Office only does so on top-level tables and if tblInd is missing.
Change-Id: I183647bcf090d5822b3f2e04e097c8bbd277168d
|
|
This option didn't work if the paragraph to collapse had an alignment or
spacing defined. This fix was needed to properly import docx/doc files
Change-Id: I6b3522905714fe3def7d87cd82cca84746310793
|
|
Change-Id: I9e8dee39f63a08517eb654e33747bd8c95d84b59
|
|
Change-Id: If03f36d30a7861f5d14ed3142e339d52da0bdce0
|
|
Change-Id: Ibad37ac05c608c0211063bfa73931427d48c11f9
|
|
Given that we talk about mm100, changing 0 to 1 is not that harmful, and
it seems later core would crash without at least a small margin.
Change-Id: I913e4a5eeaba2e36b55f2d7d7fb5bdfbad8eae94
|
|
For one, --tui has negative impact on terminal behavior in certain ssh scenarios
for me. Whether and how to use it should be a user's personal choice anyway.
For another, you sometimes need to interact with gdb early, so do not
automatically start to run soffice.bin.
Change-Id: I1a1a02f5359aae2a11effa6620aebf01202357f9
|
|
Despite its name, it is a cleanup function for libxml2 that must only be called
when the process as a whole no longer needs libxml2. Calling it from a library
like liblangtag is not appropriate (just like liblangtag does not call
xmlInitParser, either).
In LibreOffice, this caused confusion about pthread keys for thread-local
storage (xmlCleanupParser calls pthread_key_delete, but apparently if libxml2 is
later used by unrelated code in LibreOffice again, it still uses the stale key,
and if that key is now reused via an unrelated pthread_key_create, arbitrary
crashes happen during thread termination in pthread's __nptl_deallocate_tsd).
Change-Id: I82cef8bc400d48f19eaa94f26d50d4b8589da106
|
|
...so that it consistenly uses the LO versions of libxml2/glibc libs at runtime.
Change-Id: I89b7fd13a8b5b7e697ec21db74fdf8ef088fdb92
|
|
Change-Id: Ie5e7aecd4f1c0e5b4dda9250ae755bf9210e048f
|
|
Change-Id: Iecd0ed802f6cc715ac77256bb4b49082b5976b4f
|
|
Change-Id: Ifc2bfb278947344d14b855ebf5527b603e333f15
|
|
Change-Id: I7de335925748b2cfe28a01f6975a630081b3172e
|
|
Change-Id: I9c5203400904902238c1f9a8f01338fdcf2904d8
|
|
Change-Id: I85016d1e70dbb5e499aefad568179800b5d00ab7
|
|
Change-Id: I2d62eddb561baa28e320f7d89a8655d2806d7b10
|
|
Change-Id: Idf8b079857638558075fa6b1fb5f7ea6b3c9f632
|
|
This reverts commit 1f77f1ac177521698ee7c978f218127cb5840046
and b590d87b380389371eca94ffb0e6f47f86d1c917.
Do this again when we will know how to get rid of published service
in offapi/type_reference/types.rdb.
|
|
Change-Id: I2cae970f532c3441a7ec7c8318987ea343aaf1cd
|
|
re-work LRU cache to really be an 8 entry LRU cache.
only force dirty pages to stay in memory, hopefully huge reads
should now stream through memory.
Change-Id: I0bedc762086f5f02202795568743750aefaaa50b
|
|
Change-Id: I7c5bc8b9375a325d4ea78f033f577eff44cfd6e9
|
|
Change-Id: I76e737f59aa36bf7982255fa9bb41ef9d6abfe8d
|
|
Change-Id: Idd33388f61461a55021b93498e2c992d7ae560d0
|
|
Change-Id: I5e966a5734308381ad305b891ecfc830dc1419f6
|
|
Implementation is gone since b2d7ede98716806cc4b3ea8547247175d5f04d18
Change-Id: I3986ca99748f0010d80d2c788723a42a1725e83c
|
|
Change-Id: I288638c4a70396ccb7354b20e901aa0c46ffe442
Reviewed-on: https://gerrit.libreoffice.org/653
Reviewed-by: Michael Meeks <michael.meeks@suse.com>
Tested-by: Michael Meeks <michael.meeks@suse.com>
|
|
Change-Id: I07d7b78be200f5f4a0674361237ef45b2bd1e3e1
Reviewed-on: https://gerrit.libreoffice.org/651
Reviewed-by: Miklos Vajna <vmiklos@suse.cz>
Tested-by: Miklos Vajna <vmiklos@suse.cz>
|
|
Change-Id: I7c1338ef2dae7831cddda8378a031a58011d7b17
|
|
Change-Id: Id2b7d3ca0881bf5b446692e1351517f1e8971fd7
|
|
Change-Id: If79fffec7388e73fd85f5279e773af363c5f97e3
Reviewed-on: https://gerrit.libreoffice.org/645
Reviewed-by: Miklos Vajna <vmiklos@suse.cz>
Tested-by: Miklos Vajna <vmiklos@suse.cz>
|
|
That is a much better place for it. The gb_C(XX)FLAGS don't get passed
on to "external" modules, so they should not contain flags that are
essential for the compilation to work at all. I think.
At least for me, in my MacOSX SDK 10.4 -based build tree, using Xcode
3 installed in /Xcode3, on OS X 10.8, the -isysroot is essential for
liborcus to compile.
Change-Id: I1f11c690585e4ba512eb87d2fddf11872b91dd57
|
|
Causes way too much problems with the tinderboxes, thanks to -Werror
getting passed to the liborcus configury, and liborcus apparently is
quite -Werror-unclean.
Thus also revert the initial fixes for those WaEs. (Many more would be
needed, it seems. I rather leave it to Kohei to make upstream liborcus
warning-free.
This reverts commit 083fa89dd8f894f9cd0ef217a900fab777db9f2a.
This reverts commit 7e871bbaf7b208dd884528259211385b8074da39.
This reverts commit 21b5706f83edecad8cd3d0ab3ce5110b82e75a46.
This reverts commit bed5557e8bdfe5c698a45a4cc82619a7c22de001.
|
|
Change-Id: I1cc06ac0608ef159eb3027bea614b2dade8b857c
|
|
Change-Id: Ibbbe331421434e15c4659b5c566dadecf441dff8
|