- 18 Eyl, 2015 40 kayıt (commit)
-
-
Stephan Bergmann yazdı
...for one, this avoids sending seek requests down the GIO/GVFS stack, which can silently fail with corrupt data read from certain broken servers for current versions of the GIO/GVFS stack; for another, it should considerably speed up loading documents via the GIO UCP, as LO's document type detection is notorious for issuing lots of seek and (small-chunk and/or re-) read operations on a file. (This issue has become more relevant after 51e0d789 "rhbz#1134285: Access dav, davs URLs via GVFS," where the old route via the WebDAV UCP was apparently not affected by those broken servers, so happened to start affecting opening certain remote files in LO via Nautilus.) Change-Id: I91f91128b2d1a16f976eafeacf216a91747f4df1 (cherry picked from commit 93a0696e) Reviewed-on: https://gerrit.libreoffice.org/18612Reviewed-by:
Caolán McNamara <caolanm@redhat.com> Tested-by:
-
Eike Rathke yazdı
EndListeningContext holds BroadcasterStoreType iterators in its ColumnBlockPositionSet and collects broadcasters to purge them at the end. Removing broadcasters from ScColumn::maBroadcasters in between invalidates the iterators. Hence calling the "normal" EndListening() that removes a broadcaster when all listeners are gone while an EndListeningContext is in use is bad. Change-Id: Ibdd88469e91e6173ceff1f391c23ef7cb7c6f596 (cherry picked from commit e4a8ae0b) Reviewed-on: https://gerrit.libreoffice.org/18615Tested-by:
Jenkins <ci@libreoffice.org> Reviewed-by:
-
Caolán McNamara yazdı
Change-Id: Ib066ff1fad686d611a86510e768b1109ef7a5e75 (cherry picked from commit 7aa2f8c7) Reviewed-on: https://gerrit.libreoffice.org/18640Reviewed-by:
David Tardon <dtardon@redhat.com> Tested-by:
-
Eike Rathke yazdı
It is not sufficient to collect the cells, one has to broadcast on the collection. Change-Id: I11f889936aff43f958c56789e539809289819752 (cherry picked from commit 97ad6393) Reviewed-on: https://gerrit.libreoffice.org/18573Tested-by:
Tor Lillqvist <tml@collabora.com>
-
Tomaž Vajngerl yazdı
Change-Id: Ibe8d9140f7a54016f2cd684198856ac27d880aa3 (cherry picked from commit b85d24e4) Reviewed-on: https://gerrit.libreoffice.org/18622Reviewed-by:
Michael Meeks <michael.meeks@collabora.com> Tested-by:
Jan Holesovsky <kendy@collabora.com>
-
Tomaž Vajngerl yazdı
Not needed anymore.. This reverts commit 812e85a1. Change-Id: I2b25180a3ddaf5b838197a82eeec2de1c202a7d7 Reviewed-on: https://gerrit.libreoffice.org/18648Reviewed-by:
-
Michael Meeks yazdı
Each has rather different assumptions about how life should be. Change-Id: I85fffc77d14f3a5335a077fcb541a2b31c372043 Reviewed-on: https://gerrit.libreoffice.org/18644Reviewed-by:
-
Jan Holesovsky yazdı
Change-Id: I81ee00114d6e1e942e41e61a50940691a2a41a40 Reviewed-on: https://gerrit.libreoffice.org/18639Reviewed-by:
-
Tomaž Vajngerl yazdı
Change-Id: Idd80b7390694038ab0913edab0e496593beb0e15 (cherry picked from commit 7993663c) Reviewed-on: https://gerrit.libreoffice.org/18625Reviewed-by:
-
Tomaž Vajngerl yazdı
Change-Id: Iacd75beecc14023173a9aa52a30298bbfe787d61 (cherry picked from commit 0eb9f13d) Reviewed-on: https://gerrit.libreoffice.org/18623Reviewed-by:
-
Marco Cecchetti yazdı
Reviewed-on: https://gerrit.libreoffice.org/18555Tested-by:
Tomaž Vajngerl <quikee@gmail.com> Signed-off-by:
-
Tomaž Vajngerl yazdı
Change-Id: Ic853457871b914f9c1beb2f648bf7d9d18dce957 (cherry picked from commit 4823b6d4) Reviewed-on: https://gerrit.libreoffice.org/18624Reviewed-by:
-
Tomaž Vajngerl yazdı
Fixes shrinking shapes with gradients when the VirtualDev is reused to create a alpha mask (and the texture is reused and reading back from just one part of the texture which uses glReadPixels code-path). Binding texture is not necessary when we use and bind it to the framebuffer. Change-Id: Ie3994f749e1a2c17d4d3df44710b7453d7a4f45f (cherry picked from commit 0165da09) Reviewed-on: https://gerrit.libreoffice.org/18621Reviewed-by:
-
Michael Meeks yazdı
Unfortunately, since we can have 2x SalGraphics' on a OutputDevice, and one of these can be a printer - things got very confused around which context to glFlush. This de-tangles the various reference-counts. Conflicts: vcl/inc/generic/genpspgraphics.h vcl/win/source/gdi/salgdi.cxx Change-Id: I1062be0b02a91fc9009deaa3ec29c5dbb227df20 Reviewed-on: https://gerrit.libreoffice.org/18611Reviewed-by:
-
Michael Meeks yazdı
Change-Id: Id4ac387174fa03c5886a990d89fd53a4f18db11c Reviewed-on: https://gerrit.libreoffice.org/18610Tested-by:
-
Michael Meeks yazdı
Fixes crasher exiting slideshow. Change-Id: I14b6a2c1abc946f590bf91cab19099753172fd95 Reviewed-on: https://gerrit.libreoffice.org/18609Reviewed-by:
-
Michael Meeks yazdı
Using vdev's associated with old contexts, is un-necessary and a recipe for context switching & hence flicker. Change-Id: I37ed967a7816e5ca0240908ab4793ce1e68570ee Reviewed-on: https://gerrit.libreoffice.org/18602Tested-by:
-
Michael Meeks yazdı
Avoids a rather horrible flickering problem in GL mode. Squashing: tdf#94213 - release offscreen texture properly on re-size. We need to ensure that we use an initialized context, and that (when we re-parent) we DeInit and so reset the previous OpenGLContext. Make UseContext more paranoid as well for good measure. Squashing: tdf#94213 - cleanup associated GL contexts properly when DCs released. Change-Id: I6b9fb899777d8e460999ac3ff038a1302e434bb5 Reviewed-on: https://gerrit.libreoffice.org/18607Tested-by:
-
Tor Lillqvist yazdı
Will make it less demanding on low-end hardware, where the device driver is unresponsive for too long when a OpenCL kernel handling lots of data is executing. This makes Windows restart it which is problematic. I tried several approaches of splitting, both at higher levels in sc and at the lowest level just before creating and executing the OpenCL kernel(s). This seems to be the most minimal and local approach. Doing it at the lower level would have required too much poking into our obscure OpenCL code, like passing an offset parameter to every kernel. Use a simple heuristic to find out whether to split. On the problematic low-end devices, CL_DEVICE_PREFERRED_VECTOR_WIDTH_FLOAT is 4, while for more performant devices it is 1 or 8. Change-Id: If16d152710057b34d09ef0203960e1fbb9ac067f Reviewed-on: https://gerrit.libreoffice.org/18613Reviewed-by:
-
Matthias Freund yazdı
I redo the sc_instergraphic to 1px outline instead 2px. New crop icon. sc_fillcolor now behind the color line sc_insertbookmark now without the plus. Change-Id: I9dd62bbe378735dea322b5ead03951d7db9a3162 Reviewed-on: https://gerrit.libreoffice.org/18596Reviewed-by:
Yousuf Philips <philipz85@hotmail.com> Tested-by:
Adolfo Jayme Barrientos <fitojb@ubuntu.com> Tested-by:
-
Caolán McNamara yazdı
Change-Id: Ic2eec30cfb5f61c53777eefeeb8bad6f552da2fc Reviewed-on: https://gerrit.libreoffice.org/17355Tested-by:
Michael Stahl <mstahl@redhat.com>
-
Winfried Donkers yazdı
in Calc functions NETWORKDAYS and WORKDAY.INTL Change-Id: I2d42ab956e4ab9f2187a0c6bc3c64c9306ca892a Reviewed-on: https://gerrit.libreoffice.org/18559Tested-by:
Eike Rathke <erack@redhat.com> Tested-by:
-
Vasily Melenchuk yazdı
InetURLObject class does not work correctly with internal document links, containing not full url, but just a fragment. This case added as an exception for HTML link export. Change-Id: I44496a1cf186836d4194fc65ce7bf9aeb2f79b65 Reviewed-on: https://gerrit.libreoffice.org/17993Reviewed-by:
Thorsten Behrens <Thorsten.Behrens@CIB.de> Tested-by:
-
Miklos Vajna yazdı
The bug document has: <style:page-layout-properties ... style:repeat="no-repeat"> <style:background-image ... style:repeat="stretch"/> </style:page-layout-properties> When a stretched background image is set on a page style using the Writer UI, then these two style:repeat attributes always match, but not in the bugdoc. The later used to have priority, till commit 7d9bb549 (Related: #i124638# Second step of DrawingLayer FillAttributes..., 2014-06-02). Fix the problem by extending XMLBackgroundImageContext::EndElement(): if we know that the <style:page-layout-properties> sets the FillBitmapMode property, then don't blindly set BackGraphicLocation, but try to overwrite the exiting FillBitmapMode one. Change-Id: I64ab4363b20cc95003d35acd63ea421472b1c071 (cherry picked from commit 6621da38) Reviewed-on: https://gerrit.libreoffice.org/18582Tested-by: -
Michael Stahl yazdı
5 ms timers cause SwTextFrm::CollectAutoCmplWrds() to return early, and unlike the auto-spelling stuff there is nothing to store the already-checked range of the paragraph, so on the next iteration it will start from the beginning and time-out again. Prevent that by excluding TIMER events here, as is already done for the ONLINE_SPELLING case. Change-Id: Iac781f10ce0aef902fa921030e61b4cff65d0cb3 (cherry picked from commit b4f35a74) Reviewed-on: https://gerrit.libreoffice.org/18590Reviewed-by:
-
Michael Stahl yazdı
So the auto-spell-checking is hyper-optimized to add the words to the auto-complete list as well, and call SetAutoCompleteWordDirty(). But if you disable the auto-spell-checking, then a separate function SwTextFrm::CollectAutoCmplWrds() will be called, which is buggy because it only resets the dirty flag if at least one word of sufficient length was found in the paragraph, which is never the case for an empty paragraph. Change-Id: Idec64fc3c379301426a44e06a1114c474de36014 (cherry picked from commit 97c6dac6) Reviewed-on: https://gerrit.libreoffice.org/18589Reviewed-by:
-
Samuel Mehrbrodt yazdı
Need to update the tooltip also on state change. This partially reverts the fix for tdf#83558, 1fb8724f Verified that the fix for tdf#83558 still works. Change-Id: I023a5e4b101dc91522f19b0d3ed2ed0c4a47e64b Reviewed-on: https://gerrit.libreoffice.org/18586Tested-by:
-
Tomáš Chvátal yazdı
Otherwise this would needlessly fail on some compilers. Change-Id: I4e46e9e452de0703b6556cd55e4606b02be5c41c Reviewed-on: https://gerrit.libreoffice.org/18584Tested-by:
-
Caolán McNamara yazdı
Change-Id: Ie066c7f83ad15bec198f2091a3b084468c502766 (cherry picked from commit 064fd834) Reviewed-on: https://gerrit.libreoffice.org/18499Reviewed-by:
Miklos Vajna <vmiklos@collabora.co.uk> Tested-by:
-
Caolán McNamara yazdı
Change-Id: Id0be728602b3c58b2853ff464336d68303531efe (cherry picked from commit f3f1919a) Reviewed-on: https://gerrit.libreoffice.org/18434Reviewed-by:
-
Szymon Kłos yazdı
While opening folders in SvtFileView using doubleclick, sometimes GtkSalFrame::gestureLongPress method is executed with a null frame pointer and LO crashes. I noticed this only with remote dirs, probably this bug occurs only when the doubleclick handler routine takes a lot of time. (cherry picked from commit db33c785) Change-Id: I432046994b3e1662bd7e499681bd20e9696b2d52 Reviewed-on: https://gerrit.libreoffice.org/18571Reviewed-by:
-
Julien Nabet yazdı
Change-Id: Ic57611be96f160037fbff2e9452f9206083c80e4 Reviewed-on: https://gerrit.libreoffice.org/18324Reviewed-by:
-
Katarina Behrens yazdı
Regression from commit 8b779997, which replaced named GtkFrame with GtkBox, but didn't rename the box accordingly. Change-Id: I321dbee05faded81dbf575bca3c578ecb09f1a3e Reviewed-on: https://gerrit.libreoffice.org/18562Reviewed-by:
-
Caolán McNamara yazdı
Change-Id: I233c2fff9c06a81117f8114ccee83b53ea4026db (cherry picked from commit b43e0335) Reviewed-on: https://gerrit.libreoffice.org/18505Tested-by:
-
Caolán McNamara yazdı
Change-Id: I3e8aa5c48ba802cd363688502b44e27bfdf67f01 (cherry picked from commit b02f1c58) Reviewed-on: https://gerrit.libreoffice.org/18464Tested-by:
-
Julien Nabet yazdı
+ Typo: excecptionTypeSizeArray->exceptionTypeSizeArray Cherry-picked from 81d62c5f Change-Id: I6fac3bea1eba094e87717d20a08ec7cf6151e2df Reviewed-on: https://gerrit.libreoffice.org/18527Tested-by:
Noel Grandin <noelgrandin@gmail.com>
-
László Németh yazdı
with disabled rulers. (This fix was suggested by Tomaž Vajngerl.) (Cherry-picked from the commit ed031895) Conflicts: svtools/source/control/ruler.cxx Change-Id: I161237cdb4941c0eaf934223b078acd94d72e21d Reviewed-on: https://gerrit.libreoffice.org/18507Tested-by:
-
Michael Stahl yazdı
There is a sort of intentional infinite loop in the idle spell checking handler: while the user is typing a word, it should not be marked as invalid yet, in order not to annoy them with red underlines. So the word where the cursor is positioned always remained dirty, unless you happen to have a grammar checker enabled, which clears the paragraph's dirty flag from a separate thread. To avoid the infinite loop, add another spell checking state "PENDING" which is the same as dirty except that it should cancel the idle spell checking. The idle spell checking will run again when the user does the next editing operation. Notably this means if the user just moves the cursor out of the wrongly spelled word, it won't be underlined yet, but that appears a minor issue, and checking when the cursor leaves the word appears too hard to implement. (cherry picked from commit 4c91e94e) Change-Id: Ifb3d6d17f94f9f1cfad82e70dfa79f1594c38647 Reviewed-on: https://gerrit.libreoffice.org/18511Tested-by:
-
Michael Stahl yazdı
This reverts commit 9099e21b. It causes the problem that frames anchored to the same node where redlines start or end get duplicated during Hide. Jan-Marek said that the original mail-merge related problem this change was fixing is now most likely fixed differently, and the test in testMultiPageAnchoredDraws() still passes. (cherry picked from commit e84f0a9b) sw: add unit test for the redline frame duplication regression (cherry picked from commit d5ffcba0) Change-Id: Ie84fed3f64be7696782bc557004eb18fccc5b64b Reviewed-on: https://gerrit.libreoffice.org/18509Reviewed-by:
-
Lubosz Sarnecki yazdı
Change-Id: Ib139e81f770531f7d808764dd3a77c7ac8d6fa3c Signed-off-by:
-
