Age | Commit message (Collapse) | Author |
|
On Mac OS X, it apparently suffices to execute "new java.awt.Dimension()" to run
into "Apple AWT Java VM was loaded on first thread -- can't start AWT" (see
<http://developer.apple.com/library/mac/#technotes/tn2005/tn2147.html> "JNI
Development on Mac OS X - Thread-Safe JNI Programming - Calling AWT/Swing From
AppKit").
The solution here is simple, in that uses of java.awt.Dimension can be replaced
with com.sun.star.awt.Size without loss of functionality. However, there are
still occurrences of java.awt.Image and java.awt.Toolkit lurking
(reportbuilder/java/com/sun/star/report/SOImageService.java,
reportbuilder/java/com/sun/star/report/pentaho/output/ImageProducer.java,
reportbuilder/java/com/sun/star/report/pentaho/output/OfficeDocumentReportTarget.java)
for which it might be sheer luck that they do not seem to cause trouble yet.
Change-Id: I33e9f74c50ebddc11bd1d9a48c55dc8f8700345d
Signed-off-by: Lionel Elie Mamane <lionel@mamane.lu>
|
|
The problem was that we tried to insert a page break before reaching the
first section break, where section properties are sent.
Additionally, the continuous section break at the end of the doc caused
trouble, so ignore it explicitly.
(cherry picked from commit 1efa576ef88141c4deb5da9818537e053dc6517b)
Change-Id: I22bc355994991beeadb41d26b44ce3e2beedbdb2
Reviewed-on: https://gerrit.libreoffice.org/497
Reviewed-by: Norbert Thiebaud <nthiebaud@gmail.com>
Tested-by: Norbert Thiebaud <nthiebaud@gmail.com>
|
|
* fdo#53968 revealed that multiple soffice.bin instances can run removeTree in
parallel. Therefore, demoted failures from exceptions to SAL_WARNs. (And
keeping fingers crossed.)
* a8cdce148c76c93c5d41820610d6e6ac175e03a7 "fdo#53655: Ignore failure to remove
directories (as happens on Windows XP)" was due to a forgotten
osl::Directory::close before calling osl::Directory::remove after all.
* UserInstallations have been seen in the wild where no extensions were
installed per-user (any longer), but user/uno_packages/cache/registry/
com.sun.star.comp.deployment.component.PackageRegistryBackend/*.rdb files
contained data nevertheless. To reliably clean out any old junk,
refreshBundledExtensionsDir has been extended to cleanExtensionsCache which in
tandem with an extended Desktop::SynchronizeExtensionRepositories now cleanly
re-installs all bundled, shared, and per-user extensions after a LO upgrade.
Change-Id: Ic6b5b6c1945d76eb3a65b6cd4512a657b7a835a0
Reviewed-on: https://gerrit.libreoffice.org/477
Reviewed-by: Norbert Thiebaud <nthiebaud@gmail.com>
Tested-by: Norbert Thiebaud <nthiebaud@gmail.com>
|
|
removeRdbFiles suffered from a confusion that ImplementationInfo.uri denotes the
corresponding component (.so, .jar, etc.), but not the .rdb file. So removing
an .rdb file silently failed to remove the corresponding implementations, so re-
installing a similar enough .rdb (as typically happens during extension update)
would fail due to duplicate implementation names.
Change-Id: I25d4ff72656c99a3af509eef09e89c18cfd0aabe
Reviewed-on: https://gerrit.libreoffice.org/476
Reviewed-by: Norbert Thiebaud <nthiebaud@gmail.com>
Tested-by: Norbert Thiebaud <nthiebaud@gmail.com>
|
|
dbtools were added in 01089138cc76f5717244bd04f25a868602aeaab6 only to
create dependency and fix the build.
This creates circular dependency for libmerged, dbtools is not that
common library.
It's better to create real dependency in connectivity_inc package.
To avoid another circular dependency gb_YaccTarget_get_header_target
must not depend on external headers target.
Change-Id: I431513d0b7a895b471eeec854b3b3f1fd69503af
Reviewed-on: https://gerrit.libreoffice.org/470
Reviewed-by: Norbert Thiebaud <nthiebaud@gmail.com>
Tested-by: Norbert Thiebaud <nthiebaud@gmail.com>
|
|
An "archive" installer contains one empty share/extensions/dict-XY/
directory for every dictionary (except the 3 default ones), causing
numerous assertions about missing META-INF/manifest.xml and an extension
manager cluttered; this is due to overzealous removal of extension
"prereg" code. (regression from
6dcb3d4ef46312729bb6f16c473b433474863f68, partially revert that)
(cherry picked from commit 8098a03d025a0c8fee90322e3835079edb45ade3)
Conflicts:
solenv/bin/modules/installer/simplepackage.pm
...plus cherry-pick of follow-up 7b4394f35b3557ffe61c8e9b0f0b630b31ca0977
"...but remove obsolete prereg-specific code again."
Change-Id: I3a063ebfed1012aeb27ec3076cdd5ca545f918a2
Reviewed-on: https://gerrit.libreoffice.org/464
Reviewed-by: Norbert Thiebaud <nthiebaud@gmail.com>
Tested-by: Norbert Thiebaud <nthiebaud@gmail.com>
|
|
Change-Id: Ibb543b8da8c89c7fa3799fada03c787dabe2b3cb
(cherry picked from commit a251863970e0512a1dc2bd857913a3adf7148bb3)
Reviewed-on: https://gerrit.libreoffice.org/472
Reviewed-by: Norbert Thiebaud <nthiebaud@gmail.com>
Tested-by: Norbert Thiebaud <nthiebaud@gmail.com>
|
|
Change-Id: I682bd186b571973acbe98a00af7f6a83ce83e9d4
Reviewed-on: https://gerrit.libreoffice.org/498
Reviewed-by: Kohei Yoshida <kohei.yoshida@gmail.com>
Tested-by: Kohei Yoshida <kohei.yoshida@gmail.com>
|
|
This patch disables row height calculations ( and uses the hard heights read
from the xml instead ) The bug here partially exisits because the height
calculation doesn't work correctly because the factor from
ScDocShell::GetOutputFactor() has not been calculated yet ( and has a default
value of 1 ) This number is used in setting up the correct nPPTX and because it
isn't setup correctly the height calculated for the text is incorrect. We use
EnableAdjustHeight to inhibit auto height calculations that result from various
changes ( like style changes etc. ) from the start of the load and re-enabled
it at the end.
Change-Id: Ib6715c39b7cb5793458f15b1f0b3c9ecd16412f9
Signed-off-by: Kohei Yoshida <kohei.yoshida@gmail.com>
|
|
Change-Id: I7d94525daaeb7c8268fb10594c8a21ebcd52311c
Signed-off-by: Noel Power <noel.power@suse.com>
|
|
...it apparently leads to crashes, but is probably not good from a usability
perspective anyway (as the menu closes again when the dialog appears/is operated
on by the user).
For now, just disable the Java specific interaction handler here; might make
sense to address this more generally though (there's framework::QuietInteraction
btw).
Change-Id: I6ae303c0084549b5339d219e158cdb89e5a6b331
(cherry picked from commit 9cfc64c66ff9e0f0251a006f039bf4e26a63147a)
Reviewed-on: https://gerrit.libreoffice.org/504
Reviewed-by: Björn Michaelsen <bjoern.michaelsen@canonical.com>
Tested-by: Björn Michaelsen <bjoern.michaelsen@canonical.com>
|
|
Change-Id: I420c044c9bcc49e479956f9118fb173a0350fda2
Signed-off-by: Kohei Yoshida <kohei.yoshida@gmail.com>
|
|
This change replaces lcl_IsSkippableWhitespace with a call to ICU's u_isspace, which covers all Unicode separators. It also updates and fixes one of the SwScanner unit tests.
Bug details:
SwScanner::NextWord skips whitespace before calling into ICU's BreakIterator. The function used to identify whitespace (lcl_IsSkippableWhitespace) doesn't cover the full category of Unicode separators (code [Zs], 18 in total. See: http://www.fileformat.info/info/unicode/category/Zs/index.htm).
Since 0xA0 (no-break space) is not identified as whitespace and not skipped, we end up calling ICU starting at the position 0xA0, asking it to get us the boundary of the next word forward. ICU sees that it's called at the end of a word, and reverses the query direction to backward, and returns the word before. This causes NextWord to think we've hit the end of the string and call it a day, terminating word count for the rest of the line.
Change-Id: I29c89ddb0b26e88da822501253898856b28e3fa5
Reviewed-on: https://gerrit.libreoffice.org/453
Reviewed-by: Andras Timar <atimar@suse.com>
Tested-by: Andras Timar <atimar@suse.com>
(cherry picked from commit 3ba107606682b5e675127483a514f0e6580ecfd1)
Signed-off-by: Caolán McNamara <caolanm@redhat.com>
|
|
Change-Id: I83cbb2574849c69fba3ca315d7507358654ff2b9
|
|
In MOOXML '[1]Sheet 4'!$A$1 represents a reference to an external
document's cache index with a sheet name containing blank, and not
[1]'Sheet 4'!$A$1
Change-Id: I5caf9e1d9a80154116dc7a72b33d34229092839c
|
|
Change-Id: Ib17bb0a2fc95207da69bfef11b2fab59aefa719d
Signed-off-by: Fridrich Štrba <fridrich.strba@bluewin.ch>
|
|
Change-Id: I9bdc9997e260a75682177c8641695b60df0c81a6
Signed-off-by: Andras Timar <atimar@suse.com>
|
|
Change-Id: I27d056bdb8329dce302f2737bad5c5dc55791e74
Signed-off-by: Andras Timar <atimar@suse.com>
|
|
Change-Id: I1228b3a064623e810265b47776e33f81f4bf0e51
Signed-off-by: Kohei Yoshida <kohei.yoshida@gmail.com>
|
|
Change-Id: I586a31a5d563ea8a11282836ca1d54b0c8be1bf4
Signed-off-by: Arnaud Versini <arnaud.versini@gmail.com>
Signed-off-by: Kohei Yoshida <kohei.yoshida@gmail.com>
|
|
Because the "fix" for i#8288 only set the point of the cursor and not
the mark, it can happen that after a Replace All weird things are
selected; fix that (and i#8288 properly) by using the SwCrsrShell stack,
and don't do it when searching in the selection.
This cannot be done in SwCrsrShell itself because ExecSearch sets the
cursor to the beginning/end of the document already.
(regression from cda75d6f358d438f7ed36ed7fa296ac23e3c4a04)
Change-Id: Ie9f8d7da2fa26000a0237c24f8742e47ffa29213
(cherry picked from commit b3233579ccb2528912835deedde37c1d48417d6c)
Reviewed-on: https://gerrit.libreoffice.org/481
Reviewed-by: Miklos Vajna <vmiklos@suse.cz>
Tested-by: Miklos Vajna <vmiklos@suse.cz>
|
|
Fix typo that had intro bitmap always be default in 3.6 series.
Change-Id: I7822ab75c718040128231f5d50913c3b128d49fd
Reviewed-on: https://gerrit.libreoffice.org/475
Reviewed-by: Miklos Vajna <vmiklos@suse.cz>
Tested-by: Miklos Vajna <vmiklos@suse.cz>
|
|
regression from 399e425fc9b9a89045bb1653769f581ed5fa5028
Change-Id: I2907a03ca731f43630e5f36cf78682de8a364924
(cherry picked from commit cb0aa85747fa6ada9a7dc4b32e5c5bf3d3850f35)
Reviewed-on: https://gerrit.libreoffice.org/474
Reviewed-by: Rafael Dominguez <venccsralph@gmail.com>
Tested-by: Rafael Dominguez <venccsralph@gmail.com>
Reviewed-by: Eike Rathke <erack@redhat.com>
Tested-by: Eike Rathke <erack@redhat.com>
|
|
This reverts commit 9efd8deb867c89baf0f5df410b87e8310fff524b.
|
|
This patch disables row height calculations ( and uses the hard heights read
from the xml instead ) The bug here partially exisits because the height
calculation doesn't work correctly because the factor from
ScDocShell::GetOutputFactor() has not been calculated yet ( and has a default
value of 1 ) This number is used in setting up the correct nPPTX and because it
isn't setup correctly the height calculated for the text is incorrect. We use
EnableAdjustHeight to inhibit auto height calculations that result from various
changes ( like style changes etc. ) from the start of the load and re-enabled
it at the end.
Change-Id: Ib6715c39b7cb5793458f15b1f0b3c9ecd16412f9
|
|
Change-Id: I4ea33f03594c83568f93c036e3143f84e72e2983
Reviewed-on: https://gerrit.libreoffice.org/471
Reviewed-by: Tomáš Chvátal <tchvatal@suse.cz>
Tested-by: Tomáš Chvátal <tchvatal@suse.cz>
|
|
This reverts commit 7f18b80a3bd2b3c126ac852c7ecd5ceb376f389d.
Change-Id: Ie577edba915ad78eaa82a66280a51ddd6ba9398e
|
|
For some reason older gcc versions don't manage to dynamic_cast
to the necessary cast. I'm not quite sure why, forcing sal/osl/unx/module.cxx
to always use RTLD_GLOBAL does not seem to help. Most probably
compiler bug. Changing the cast to two simpler ones helps.
Signed-off-by: Miklos Vajna <vmiklos@suse.cz>
|
|
Introduced by b95d203bc17c83ec0fe5139f519d53ed1d842d3a .
Change-Id: I2497ee7cd55776fbd801d07470968e5aa47e971a
Signed-off-by: Cédric Bosdonnat <cedric.bosdonnat@free.fr>
|
|
Change-Id: Ie8322b93745c48970766c2033a23cca2c710a516
(cherry picked from commit 442d0766eb9575a45572e29f2018bed2831c177f)
Reviewed-on: https://gerrit.libreoffice.org/466
Reviewed-by: Bosdonnat Cedric <cedric.bosdonnat@free.fr>
Tested-by: Bosdonnat Cedric <cedric.bosdonnat@free.fr>
|
|
Change-Id: I5e84cf342ab9a7017bf61dbb455548105b867e50
Reviewed-on: https://gerrit.libreoffice.org/468
Reviewed-by: Norbert Thiebaud <nthiebaud@gmail.com>
Tested-by: Norbert Thiebaud <nthiebaud@gmail.com>
|
|
It appears that since cd0046bca0691fdca45a030903b7cde3bfee544c,
2 tinderboxes fail on including the generated sqlbison.hxx;
it turns out one of the inclusions of that in svx is unnecessary and the
other is used to call a virtual function, which can apparently be done
without linking; just link svx to dbtools anyway, that should fix the
build.
Change-Id: I7f7b1c4e9561c4dd2b820b173d05bdd5d4c6ba6d
(cherry picked from commit 01089138cc76f5717244bd04f25a868602aeaab6)
Reviewed-on: https://gerrit.libreoffice.org/456
Reviewed-by: Miklos Vajna <vmiklos@suse.cz>
Tested-by: Miklos Vajna <vmiklos@suse.cz>
|
|
Since CWS swrefactormarks2 the SwXBookmarks collection handles not only
bookmarks but at least cross-ref marks as well, which then bother users
when they show up in the Insert->Hyperlink dialog; remove non-bookmarks
again.
(regression from df6d312ca537402463e4eb0530f22b956600fc02)
Change-Id: I6a64ba8a43468dd3ce1569e944371d3ef71f8824
(cherry picked from commit 45be3ac8151d63ccb61879f876696704542a4ce7)
Reviewed-on: https://gerrit.libreoffice.org/459
Reviewed-by: Miklos Vajna <vmiklos@suse.cz>
Tested-by: Miklos Vajna <vmiklos@suse.cz>
|
|
error: attempt to copy-construct an iterator from a singular iterator.
(regression from 39b8a5f87f55abe53488c9c3c35b65fb0df84cde)
Change-Id: Iab29f4c356ea1cb5ca0f687bcfc5e54f185fbba3
(cherry picked from commit c47505a4525c342694ba4196544715467c2bdb8e)
Reviewed-on: https://gerrit.libreoffice.org/458
Reviewed-by: Miklos Vajna <vmiklos@suse.cz>
Tested-by: Miklos Vajna <vmiklos@suse.cz>
|
|
Embedding graphics in a report builder report is not implemented,
so do not propose it to the user.
It already was not proposed the dialog raised by
"Insert->Picture->From File"
This commit disables embedding the UI when selecting
an image for an already existing Graphic control
(as opposed to inserting a new graphic control).
Change-Id: I107bc59bef8bb29a75140542cc5c2779a37ed113
Signed-off-by: Lionel Elie Mamane <lionel@mamane.lu>
|
|
Change-Id: I65fa22ceeba37a15b70fe41b1dee26f1dde7d759
Signed-off-by: Michael Meeks <michael.meeks@suse.com>
|
|
This would result in the pivot cache being refreshed twice. This block
of code was a leftover from the previous, non-cached version of pivot
table, and this logic no longer applies today.
In today's pivot table code, the cache gets refreshed first, then the
table gets refreshed afterward. We should never clear the cache data
during table refresh.
Change-Id: Ida6fa1805463facf842cb9ea94f4c1afab716595
Signed-off-by: Michael Meeks <michael.meeks@suse.com>
|
|
It was due to a simple misplacement of nRow = 0. Each record was
labeled row 0. No wonder it screws up the result badly.
Change-Id: Ia94f42a62f98a9bd7c9d34df3606351bb487931b
Signed-off-by: Michael Meeks <michael.meeks@suse.com>
|
|
Added a preferred extension property to the filter which overwrites
one(s) given by the default file format type associated with it.
Normally this property is not used for the majority of the filters,
but for plain text filters, we need to use .txt when saving from
writer whereas we do .csv from calc.
(cherry picked from commit ba03f2dc817c692fa8709391ec83d8e01a0f8b28)
(cherry picked from commit 84d1ff7be178ec3572710e092e85fe548ce3042b)
(cherry picked from commit 45b03c888042052a47cb79bc5a2837d04e115c5e)
Squashed these 3 commits into 1.
Also needed rtl::OUString("*.") instead of OUString("*.") in
sfx2/source/bastyp/fltfnc.cxx
Change-Id: Ic6d769bc1180f88fd75f40909e5e054d45719178
Signed-off-by: Eike Rathke <erack@redhat.com>
|
|
These cannot be order only because if the source file is changed the
generated cxx file has to be recompiled as well.
Change-Id: I6c1d8e2d90d98a80ce747f8cd1d118e0a83ed379
(cherry picked from commit cd0046bca0691fdca45a030903b7cde3bfee544c)
Reviewed-on: https://gerrit.libreoffice.org/450
Reviewed-by: Björn Michaelsen <bjoern.michaelsen@canonical.com>
Tested-by: Björn Michaelsen <bjoern.michaelsen@canonical.com>
|
|
This was a regression introduced with 2af9040d38af7c7353855415dbea0134585058f3
"Cleaned up utl::UCBContentHelper." GetSize exhibited the same problem and has
also been fixed. IsYounger exhibits a similar problem, but has not been
addressed, as it is unclear what to return in case no dates can be compared; it
is only used in one place (handling SID_EDITDOC in
SfxViewFrame::ExecReload_Impl, sfx2/source/view/viewfrm.cxx) where, it appears,
only contents for which DateModified /does/ yield a non-void value are relevant.
Change-Id: Iff9f5e1ed6f45eede117713d64a265b98f6862b1
Reviewed-on: https://gerrit.libreoffice.org/449
Reviewed-by: Michael Stahl <mstahl@redhat.com>
Tested-by: Michael Stahl <mstahl@redhat.com>
|
|
Change-Id: I41fd1463b6432818ee36e7f90e920a5fd98c7106
(cherry picked from commit 1316cfe8a72f7c7bc2f5ad7644a2d4c54b77fc8b)
Signed-off-by: Eike Rathke <erack@redhat.com>
|
|
Change-Id: Ic2b79e78f95e298a08794fa11f227d76152244f5
Reviewed-on: https://gerrit.libreoffice.org/444
Reviewed-by: Andras Timar <atimar@suse.com>
Tested-by: Andras Timar <atimar@suse.com>
|
|
Change-Id: I50d3f1002468d41d5b5882de170699d44fe5223c
Reviewed-on: https://gerrit.libreoffice.org/443
Reviewed-by: Andras Timar <atimar@suse.com>
Tested-by: Andras Timar <atimar@suse.com>
|
|
Arguably such annoyingly thin double borders don't make much sense
anyway, because they're essentially 2 hairlines with ~no space between,
but unfortunately older LO versions are able to create them;
since the refactoring in 2d045cdb69176b280812dda0b813371cf1ac72e2,
which changed the BorderWidthImpl::Get* methods to return 0 due to
rounding, they were ignored at least in the HTML import, which is a
regression.
So add a special purpose hack that essentially rounds up the first line
to 1 but not the other lines so the visual result is a hairline single
border.
Change-Id: I20ac4675bcf67ea58a6931a40bff3605390e9c0d
(cherry picked from commit 601bfe2ce3113719b2f8edaba2ccb6b630051a9a)
Reviewed-on: https://gerrit.libreoffice.org/451
Reviewed-by: Miklos Vajna <vmiklos@suse.cz>
Tested-by: Miklos Vajna <vmiklos@suse.cz>
|
|
This applies to two scenarios: 1) cycling through cell values, and 2)
cycling through cell functions in a formula tip window. Both were broken
and this commit fixes both.
Change-Id: I7826efbada827b8ec3edc5f34a1876c84e5e6236
Reviewed-on: https://gerrit.libreoffice.org/441
Reviewed-by: Norbert Thiebaud <nthiebaud@gmail.com>
Tested-by: Norbert Thiebaud <nthiebaud@gmail.com>
|
|
Change-Id: Idb997e672e4f997ad324bdb5d49ab7dae4a9e9e9
|
|
Change-Id: I5525eb750dfbed282fea272de4a736e6c70e51a9
Reviewed-on: https://gerrit.libreoffice.org/442
Reviewed-by: Andras Timar <atimar@suse.com>
Tested-by: Andras Timar <atimar@suse.com>
|
|
...plus, add error codes to other failure's exception strings, just in case...
Change-Id: Ic21c52e0ab52ed1752745e86bde214aee7c3c208
Signed-off-by: Michael Meeks <michael.meeks@suse.com>
|
|
Turns out 608fe962cc649ad62c489811d3a8666e0e06e5e7 "Let osl::FileStatus getters
assert programming errors" was overly optimistic and misunderstood that
osl_getFileStatus(..., nMask) /can/ return a FileStatus for which isValid(nMask)
is false, esp. for stuff like file names and URLs of Windows drives and servers.
That in turn leads to existing code now calling
rtl::OUString(_aStatus.ustrFileName) etc. with null argument and crashing.
Change-Id: Icd2168e209aa1c7a6df30cd954513d01034923db
Signed-off-by: Michael Meeks <michael.meeks@suse.com>
|