path: root/vcl
AgeCommit message (Collapse)AuthorFilesLines
2014-09-25Call getLineBreak with null XHyphenator unless TEXT_DRAW_WORDBREAK_HYPHENATIONStephan Bergmann1-2/+4
So no longer instantiate XLinguServiceManager (which might e.g. bootstrap Python) unless explicitly requested, which e.g. removes a noticable delay when switching to the Options dialog's "Security" page (which causes size calculations of FixedText controls that use TEXT_DRAW_WORDBREAK but not TEXT_DRAW_WORDBREAK_HYPHENATION). Change-Id: Ib0a28d3d7d7718faf6df7a138f5a86b8b2fd7713
2014-09-23set mpFirstBand back to 0 after deleting itCaolán McNamara1-0/+1
bff + valgrind ==30470== Invalid read of size 8 ==30470== at 0xA02103F: RegionBand::isSingleRectangle() const (regionband.cxx:294) ==30470== by 0xA01F562: WriteRegion(SvStream&, Region const&) (region.cxx:1640) ==30470== by 0x9F01FD4: MetaClipRegionAction::Write(SvStream&, ImplMetaWriteData*) (metaact.cxx:2487) ... ==30470== Address 0x1d175980 is 0 bytes inside a block of size 56 free'd ==30470== at 0x4A07991: operator delete(void*) (in /usr/lib64/valgrind/ ==30470== by 0xA020BB1: RegionBand::implReset() (regionband.cxx:87) ==30470== by 0xA020EF2: RegionBand::load(SvStream&) (regionband.cxx:249) (cherry picked from commit 1ca892488a7a6a9c35fc99c41cd4af788fe0df6e) Conflicts: vcl/source/gdi/regionband.cxx Change-Id: I2468d43c7e73b3390f355358990f8af30b5f026e
2014-09-15clip PolyPolygon::Insert on .svm loadCaolán McNamara1-0/+6
valgrind + bff (cherry picked from commit fd64d444b730f6cb7216dac8f6e3f94b97d7ab60) Conflicts: tools/source/generic/poly2.cxx Change-Id: Ib3ed8a6e518c0686f8cbeaf021b9ca3a07005032 Reviewed-on: Reviewed-by: Michael Stahl <> Tested-by: Michael Stahl <>
2014-09-15bnc#719994: Fix the duotone filter to actually produce duotoneTor Lillqvist1-3/+4
Not sure what it tried to do earlier, but it was obviously wrong. Change-Id: I0b377c4baa88205a9a7ad7c134ec5bc2223e6205 (cherry picked from commit c03f7f2b427f7fe2b94dfb55ea64670426d1a957) Reviewed-on: Reviewed-by: Caolán McNamara <> Tested-by: Caolán McNamara <>
2014-09-15fdo#66701 - don't double destroy the HASH when PDF signing.Michael Meeks1-4/+14
Accidentally introduced in d73c039fa it seems. Change-Id: I846314ebca3a24520492887d4d9027f812cfb32f
2014-09-15Support version 2 ttf fonts and mac font family name encodingsCaolán McNamara2-5/+78
i.e. get loads more fonts working with LibreOffice Related: fdo#82259 support version 2 ttc fonts on Linux a lot of Mac OSX fonts are in this format, and we're not seeing them under Linux Change-Id: I8a6974ccb43afadec320331eb26bc4a2df9b13c9 (cherry picked from commit 02f8f895f2abc4ee090b4bd42963590aa018fb49) Related: fdo#82259 implement mac ttf fontname encoding support which is why these font's are not appearing under Linux Change-Id: Ib72ddb465b2df729953ece416cfa0477cd565bcd (cherry picked from commit 9436ae99438ccc24a86aff61aaa9174a9686814f) Reviewed-on: Reviewed-by: David Tardon <> Tested-by: David Tardon <>
2014-08-21only subtract height of hscroll if visibleCaolán McNamara1-2/+5
otherwise infinite loop in extensions update check Change-Id: I0dbf4f34208f1bf0854e365006a79470b26571f3 (cherry picked from commit 749f7548b4ee746d69c20a828ea6d3b853eba8d3) Reviewed-on: Reviewed-by: David Tardon <> Tested-by: David Tardon <>
2014-08-21facepalm: nAvailHeight is not derived from allocation WidthCaolán McNamara1-1/+1
Change-Id: Ic99c6519b02c546e0d3574cb756e3fb172d1ff52 (cherry picked from commit 36490fff3f4d1b4a7e4b7902a866abb177a0bf54) Reviewed-on: Reviewed-by: Andras Timar <> Tested-by: Andras Timar <>
2014-08-21check that AddToTable can be done validlyCaolán McNamara2-9/+19
Change-Id: I0ee69279e3bb1d9871feba17b908b8307f0ec5b4 (cherry picked from commit 43b39d3e47ffd179071732c0a8fc201c31b2bb46) Reviewed-on: Reviewed-by: David Tardon <> Tested-by: David Tardon <>
2014-08-21fdo#39477: Mac Keyboard shortcut Option key ko in Customize dialogJulien Nabet1-1/+1
(+ fdo#49280) I couldn't cherry-pick directly since the arbo is different between 4.2 and 4.3+ Indeed vcl/aqua is now vcl/osx Change-Id: I45a834c5ec1e4c4c5c4fd6960b48963df5656a7e Reviewed-on: Reviewed-by: Caolán McNamara <> Tested-by: Caolán McNamara <>
2014-08-21valgrind + bff: don't add 0xffff oldcodeCaolán McNamara2-4/+7
Invalid read of size 1 at 0x9DC6247: GIFLZWDecompressor::AddToTable(unsigned short, unsigned short) (decode.cxx:117) Change-Id: I9cea8dcd4a5eeaa1e793b9be96f82854b9d088af Reviewed-on: Reviewed-by: David Tardon <> Tested-by: David Tardon <>
2014-08-21fdo#81516: vcl: limit number of CFFs read from fontMichael Stahl1-7/+13
(cherry picked from commit 45b0b47d114437198c9e0872d427576e6e7e6cc6) Conflicts: vcl/source/fontsubset/cff.cxx Change-Id: I9928b9805169a2dbb41be669dc37617b30bc672b Reviewed-on: Reviewed-by: Miklos Vajna <> Tested-by: Miklos Vajna <>
2014-08-21fdo#81516 Support fonts like Source Han Sans with > 16 FDArray elementsAudrey Tang1-1/+1
Change-Id: I3f82574f434060d62dc3faeb730341c5b169ce9f (cherry picked from commit 3b38a2342e48a2aec3c2e4f5aebf883db4b84101) Reviewed-on: Reviewed-by: David Tardon <> Tested-by: David Tardon <>
2014-07-22fdo#78598 avoid use of invalidated pointersDavid Tardon1-6/+53
Change-Id: Ib81f79da696b5e8002f5a2ddcf160903231dc3f1 (cherry picked from commit 6b127d40c7d57745bc602d9ff7914392f9d3b92b) Reviewed-on: Reviewed-by: Björn Michaelsen <> Tested-by: Björn Michaelsen <>
2014-07-22fdo#80498: vcl: Revert "fdo#74124: Scale the pictures before calling ...Michael Stahl1-26/+1
... ImplDrawAlpha()." This reverts commit 3cf3700b7a903e88f5296076c40ae854bce91cdc. It may look nicer but is unusably slow when scrolling in Writer. (same as master commit ee36fc7add892690c95a969530ecdcfc1bc9decc) Conflicts: vcl/source/gdi/outdev2.cxx Change-Id: Ieb3d0b9dedeb3f1688eb30668e59750f3b569d9b Reviewed-on: Reviewed-by: Michael Meeks <> Tested-by: Michael Meeks <>
2014-07-22Related: fdo#52226 ensure graphics are swapped in on DrawingML::WriteImageCaolán McNamara1-0/+2
I imagine it would be best that the Graphics were delivered pre-swapped in by higher levels in case there are second level caches or more complex caching systemed wrapped around it, so warn about it in debug mode but give it a last-ditch shot anyway. i.e. while the .docx problem should be fixed there is a report of a very similar .xlsx problem Change-Id: Ie40ee10fe5cba8ff9c321f47b83e33ee2c1425fd (cherry picked from commit 6e580f3f53ae2de086a08c8ba1958b67874eb9c5) Reviewed-on: Reviewed-by: David Tardon <> Tested-by: David Tardon <>
2014-07-21bnc#881024 Don't world transform font size in WMF/EMF importTomaž Vajngerl2-6/+14
Conflicts: vcl/source/filter/wmf/winmtf.cxx Change-Id: Ia865b84ee2b159ff7251ab5a769a2b635dd2a1ea
2014-07-20fdo#72590 scale or map only when EMR_EXTSELECTCLIPRGN actionTomaž Vajngerl2-16/+14
Conflicts: vcl/source/filter/wmf/enhwmf.cxx vcl/source/filter/wmf/winmtf.cxx Change-Id: Ie42c855e696922b38760876d4090d572deaa689a
2014-07-19wmf: assume WinOrg/ViewportOrg = (0,0) if they are not providedTomaž Vajngerl1-6/+6
Conflicts: vcl/source/filter/wmf/winwmf.cxx Change-Id: Ia8a91a565118d102ff769da8bca65f16b5dfece2
2014-07-19fdo#78603 mixup (Left->Top) in Rectangle constructionTomaž Vajngerl1-0/+15
Conflicts: vcl/source/filter/wmf/winwmf.cxx Change-Id: I7fa566e036508d43b153c7983dc89fc6660718e7
2014-07-19vcl WMF: in non-placable images use SETWINDOW{ORG,EXT} for BoundsTomaž Vajngerl1-24/+51
If the image is non-placable the image bounds are bounds of all elements. Special cases are when both SETWINDOWEXT and SETWINDOWORG or SETVIEWPORTORG and SETVIEWPORTEXT are available - in those two cases use those bounds bounds instead. Conflicts: vcl/source/filter/wmf/winwmf.cxx Change-Id: I60e53cf6c47ccfc0f1139a11a866392f3e8e3ed1
2014-07-19fdo#70498 work-around scaling in SetClipPathTomaž Vajngerl1-8/+8
Scaling needed to be worked-around again so that the image in the bug report could be drawn correctly. The issue needs some more investigation how to detect what are the exact conditions that scaling is needed. Change-Id: Icd3f57a8ded92c49848feea92dc5a14362718874
2014-07-10Fix some OUString construction error with VS2013Tor Lillqvist1-1/+1
A bit unclear to me whether the code really did what was intended with earlier VS versions either. VS2013 said: error C2664: 'rtl::OUString::OUString(const sal_uInt32 *,sal_Int32)' : cannot convert argument 1 from 'const char *' to 'const rtl::OUString &' Reason: cannot convert from 'const char *' to 'const rtl::OUString' No constructor could take the source type, or constructor overload resolution was ambiguous Change-Id: Ie23d8382515064062a3ed46418d9d4a977b35d28 (cherry picked from commit eee44c951c104da342d55fd4de634efd51ae59ae) Signed-off-by: Michael Stahl <>
2014-07-04fdo#39944: Fix mapping for some a11y roles to ATKJacobo Aragunde Pérez1-5/+5
In particular, EDIT_BAR, EMBEDDED_OBJECT and HYPER_LINK were being mapped to custom roles when specific roles exist; fixed passing the proper name to registerRole. Besides, moved two roles that were in the section 'don't exist in ATK yet' because that's not the case now. Change-Id: I6818a0ac623f45053812b5c6dd5fc25c9abf4f65 (cherry picked from commit 7c37cd7bb26b91bda5e57d7bd0cecdbc9d1ebb22) Reviewed-on: Tested-by: Caolán McNamara <> Reviewed-by: Caolán McNamara <>
2014-07-04fdo#80014: vcl: fix drawing of linear gradientsMichael Stahl1-1/+1
There is an off-by-one error in OutputDevice::ImplDrawLinearGradient() that causes the bottom line to remain black. (regression from 04d937c1ec36c2d9fa8c90604c81a37d30e97da6) (cherry picked from commit d51f53fedb8cd68b1b3122c7199c4ea2d3a6980f) Conflicts: vcl/source/outdev/gradient.cxx Change-Id: I33948ec8065e446a9ec7efeab5e506c93fa2e9e8 Reviewed-on: Reviewed-by: Caolán McNamara <> Tested-by: Caolán McNamara <>
2014-06-29Resolves: fdo#80160 PNG with 1-bit colormap only show black and whiteCaolán McNamara1-21/+29
Change-Id: If0c9950c18e6091fafea47954a4654db436a3a44 (cherry picked from commit 986c0451d05e7bb6ff7edd9f27d0e45532bcd22c) (cherry picked from commit c1df0c41b602ac144fdde4efcf546549443afa86) Reviewed-on: Reviewed-by: Miklos Vajna <> Tested-by: Miklos Vajna <>
2014-06-29Related: fdo#33455 retain color on scaling of 1 bit depth pngsCaolán McNamara1-10/+18
load test doc and zoom in and out and at certain zoom levels its correctly black and at others its red (cherry picked from commit be9d65bb5f5a83c19eec02a037b8f808de70ecb8) Change-Id: I445ce672742ddb6d6592ef419bf5e14c5f09a5b5 Reviewed-on: Reviewed-by: Miklos Vajna <> Tested-by: Miklos Vajna <>
2014-06-25bnc#593614: Avoid rendering artefacts in presentation mode.Matúš Kukan1-23/+4
By not using XOR rendering, see also fdo#38844. The first variant seems to work better. Basically, this is just cherry-pick of bb5c7d6a79309236d4f19bb2498f2e850f735a2f Change-Id: I0fdccf60601a47c91f467f9d1321d1c25f1ef68c Reviewed-on: Reviewed-by: Andras Timar <> Tested-by: Andras Timar <>
2014-06-20coverity#736301 the 2nd arg is a gpointercp-4.2-6Caolán McNamara1-1/+1
Change-Id: I66e57d570c5ae70e69eea2d44812a72a3b67509e
2014-06-20fdo#35105: notify role change through a new UNO a11y eventJacobo Aragunde Pérez3-0/+17
A new UNO accessibility event called ROLE_CHANGED has been created. It should be triggered when an accessible object changes its role and every accessibility toolkit should use its own methods to make that change effective. Code to support the event in ATK has been added. Change-Id: I132e303bdb148967231334458c3cfa369c36ec8f Reviewed-on: Tested-by: Caolán McNamara <> Reviewed-by: Caolán McNamara <>
2014-06-20fdo#39944: fallback for DOCUMENT_* roles in ATK < 2.1.0Jacobo Aragunde Pérez1-0/+5
Set fallback to DOCUMENT_FRAME role. Change-Id: Ib83cd2f9881bcdec7b69a7660cf4166a612aaf60
2014-06-20have a stab at fixing RHEL-5 build that hasn't got new atk rolesCaolán McNamara1-4/+7
Change-Id: Idbfc9bae9593416564783974ef7237e20e6400b0
2014-06-16fdo#78799 - ignore overlayed monitors with co-incident origin.Michael Meeks1-4/+3
This happens with certain fglrx drivers etc. where 'cloned' is not set but instead both monitors are placed over the top of each other (by XFCE) -> work around that. Change-Id: I9d1846bfae2692681606717f7f5e8408df532d95
2014-06-11WaE: 'convertBaseToScreen:' is deprecated: first deprecated in OS X 10.7Tor Lillqvist1-0/+4
Change-Id: I04cc155e69d25021d294bccc588450b1f20e1fde
2014-06-11WaE: 'scrollerWidth' is deprecated: first deprecated in OS X 10.7Tor Lillqvist1-1/+4
Change-Id: I2451687c10f91e7d5242b04974cb7bdd745de51c
2014-06-11WaE: 'NSPrintSavePath' is deprecated: first deprecated in OS X 10.6Tor Lillqvist1-1/+1
Change-Id: I810b42da1d62c66c2ab6fb7bb6f71d1d00115dfa
2014-06-11WaE: comparison of two values with different enumeration typesTor Lillqvist1-1/+2
Change-Id: I611eefd20043386d00984f76a28d111e1c0495f1
2014-06-11WaE: 'setFlipped' is deprecated: first deprecated in OS X 10.6Tor Lillqvist1-2/+1
Not sure if using lockFocusFlipped: has the same effect, but will see... Change-Id: I957874bcf59c2c521bd5b2b2ca6f44fb0400e64f
2014-06-11WaE: case value not in enumerated type 'NSEventSubtype'Tor Lillqvist1-1/+2
Change-Id: Iaf5de07b7f0da7294681b9cdc152575a756a73c7
2014-06-11Initial attempt at implementing NSTextInputClient instead of NSTextInputTor Lillqvist2-11/+20
NSTextInput was deprecated in 10.6. Change-Id: I6bce9e91a68014a6ca28bff6a820c27817f9baaf
2014-06-11WaE: 'NSCountWindows' is deprecated: first deprecated in OS X 10.6Tor Lillqvist1-7/+3
Change-Id: I7163d3821ac0dfc0d471cf31e98f72953c78d878
2014-06-10Resolves: #i125000# check last status of newly created...Oliver-Rainer Wittmann1-15/+23
<Gdiplus::Bitmap> instance before using it. (cherry picked from commit b127235917610b9c68e19df29bb39af496906569) Change-Id: I97364cf963424b0e8d0b52b3c995bd4defdca067 (cherry picked from commit adb1bb21f804c62004f31ad5473d4cf447436b9b) (cherry picked from commit e0c9b690074cd476b03169397cbf1969f2239c1e) Reviewed-on: Tested-by: David Tardon <> Reviewed-by: David Tardon <>
2014-06-10Resolves: fdo#78477 ensure offset + sizeof(value) is in boundsCaolán McNamara1-6/+17
a) ptr is just added to offset, so move addition into 2nd arg Change-Id: Ia3e8145c69324f19aeec8b0dd97284ec382d20d7 (cherry picked from commit 57ef375bed4480bbedc799aca274a6bd26745008) b) sort in order of increasing offset Change-Id: I0d2c880438f47f4527037d7ffaf77cf142d24751 (cherry picked from commit 95e92889d3fb0e8a85cefdeb07a02e57130a9799) c) fdo#78477 ensure offset + sizeof(value) is in bounds check that largest offset + value to read is inside available space Change-Id: I4feac37bdfbae5061b3b75ddf44bb20fc5904656 (cherry picked from commit c888c211072f23cfb4cc488c641d8d822f930a33) (cherry picked from commit 891e0f76350890a4dd4331820bde8c118ac06ab0) Reviewed-on: Tested-by: Michael Stahl <> Reviewed-by: Michael Stahl <>
2014-06-10fdo79368 Don't process overflow menu items twice.Andrzej Hunt1-1/+7
We reuse the toolbox overflow menu for toolbarmanager's context menu -- toolbarmanger previously added its menu listener to the toolboxes menu permanently, meaning that it would try to handle overflow menu items (in addition to the context menu items which it should handle), instead we should only add the listener when we are actually using the menu as a context menu. Perhaps it would be better in the long run to actually use fully separate menus instead, and ask toolbox to specifically add its items to that rather than trying to hack the context menu on top of the overflow menu? Change-Id: Iecface2c6eae9ab79dbcdb25ffdbaf446e2885ea (cherry picked from commit 1ae89d189200c1e351a396a3f02612b84a69985f) Reviewed-on: Reviewed-by: Caolán McNamara <> Tested-by: Caolán McNamara <>
2014-06-10avoid problems with poppler's and vcl's FontInfo during runtimeMarkus Mohrhard2-6/+8
Change-Id: I1dbd256812cccae1e6ed0ad8bb34ce427e5e5be8 Conflicts: include/vcl/metric.hxx svtools/source/control/ctrlbox.cxx svtools/source/control/ctrltool.cxx svtools/source/control/stdmenu.cxx svx/source/sidebar/nbdtmg.cxx Conflicts: include/svtools/ctrltool.hxx include/vcl/metric.hxx include/vcl/outdev.hxx svtools/source/control/ctrlbox.cxx svtools/source/control/ctrltool.cxx svtools/source/control/stdmenu.cxx svx/source/sidebar/nbdtmg.cxx sw/source/core/uibase/docvw/srcedtw.cxx vcl/source/gdi/metric.cxx vcl/source/outdev/font.cxx
2014-06-10Avoid accessing freed memory and freeing already freed resourcesTor Lillqvist2-0/+9
Seen in smoketest a 4.2-based branch, but might perhaps happen in master too. Should not hurt in any case to be more careful. ImplFreeSalGDI() is called both from SalAbort() and from DestroySalInstance(). Change-Id: Ifac37e8ecfb432848bd3451dbda8634c59a06fba
2014-06-10I think SAL_FORCE_IACCESSIBLE2 is what is intended hereTor Lillqvist1-1/+1
Change-Id: If03696cfcc9a3d5053af881e00a4c6a484cbd445
2014-06-10fdo#39944: Add DOCUMENT_* accessibility UNO rolesJacobo Aragunde Pérez2-1/+12
Added a set of UNO accessibility roles for specific kinds of documents: * DOCUMENT_PRESENTATION for Impress * DOCUMENT_SPREADSHEET for Calc * DOCUMENT_TEXT for Writer The other applications still use the existing DOCUMENT role. These roles translates directly to ATK but in the other toolkits we keep using the same association that DOCUMENT role had. Change-Id: Ibac47527e5effdecb28d2314cde8558cf4fb010a Reviewed-on: Reviewed-by: Michael Meeks <> Tested-by: Michael Meeks <>
2014-06-10Improve Accessibility mapping of roles for MacNiklas Johansson1-3/+3
Since role Heading now is exposed to A11y we need to give it a better role then unknown, otherwise it gets inaccessible to VoiceOver. Until I get a better grip on how to best expose AXHeading, let's map it to textAreaRole so we at least get back to the behaviour before IA2 integration. Header and Footer is probably best mapped to AXGroup. Change-Id: I6353f4d25c18e6e550df289ca2e999ad1c01f2a8 Reviewed-on: Reviewed-by: Caolán McNamara <> Tested-by: Caolán McNamara <>
2014-06-10The accessible action "click" changed name to "press", adjust wrapper ↵Niklas Johansson1-1/+1
accordingly The commit efb23f29983f87104a684e7fab00b84fc59d131d Integrate branch of IAccessible2" changed the name of the accessible action click to press we need to adjust the action wrapper class on mac accordingly. Change-Id: I406a44e5b4fb31f4f12859ee5380b62cc5cccdf4 Reviewed-on: Reviewed-by: Caolán McNamara <> Tested-by: Caolán McNamara <>