- 24 Nis, 2019 30 kayıt (commit)
-
-
Jens Carl yazdı
Move TableColumn Java tests to C++ for ScTableColumnObj. Change-Id: I3f17fa1f25c2676124b56c384364e36736fac849 Reviewed-on: https://gerrit.libreoffice.org/71261 Tested-by: Jenkins Reviewed-by: Jens Carl <j.carl43@gmx.de>
-
Jens Carl yazdı
Change-Id: I08d5b060b6756a89972eeec57c1f9c85e6f65500 Reviewed-on: https://gerrit.libreoffice.org/71262 Tested-by: Jenkins Reviewed-by: Jens Carl <j.carl43@gmx.de>
-
andreas kainz yazdı
Change-Id: I90ea66c8da67c4fcdc8143f1cc94e80db207ee31 Reviewed-on: https://gerrit.libreoffice.org/71257 Tested-by: Jenkins Reviewed-by: andreas_kainz <kainz.a@gmail.com>
-
Miklos Vajna yazdı
Commit a51b7a1c (tdf#103831, tdf#100986: Force using GDI when needed, 2017-03-03) noted that the DirectWrite text renderer doesn't support stretched text, add support for this now by setting a DirectWrite transform matrix that only does horizontal scaling. With this, tdf#113076 is kept fixed, but at the same time manually stretched text keeps working. Previously the glyphs of the text had the correct size and position, but the glyphs themselves where not streched, but simply aligned to the left. Change-Id: I8fe8e74d3edc0d71ed2f16fcce66c6f5009ed264 Reviewed-on: https://gerrit.libreoffice.org/71245Reviewed-by: Miklos Vajna <vmiklos@collabora.com> Tested-by: Jenkins
-
Stephan Bergmann yazdı
For one, it is deprecated since Java 9, and for another it causes a NullPointerException when called from the JNI Invocation API (i.e., without a Java caller frame) at least with some Java 12. (Found when doing a test build with Java 12 and JAVA_SOURCE/TARGET_VER explicitly configured as 7 with d365f363 "Allow to pass JAVA_SOURCE/TARGET_VER into configure".) I have entered information about the NullPointerException at <https://bugreport.java.com/bugreport/start_form.do>, but haven't heard back yet, so duplicate that information here: Issue Type: > Bug Component: > Core Libraries Operating System: > Linux 64-bit Java Release: > 12 Synopsis: > Calling java.lang.Class.newInstance from JNI Invocation API causes NullPointerEx Description: > I can reproduce this with the OpenJDK 12 RPMs on Fedora 30, but from looking at <https://hg.openjdk.java.net/jdk/jdk12/file/06222165c35f/src/java.base/share/classes/java/lang/Class.java> it looks plausible that it is a common issue that Reflection.getCallerClass() returns null when java.lang.Class.newInstance is called from the JNI Invocation API, and that null is propagated to jdk.internal.reflect.Reflection.verifyMemberAccess as parameter currentClass, which dereferences it at <https://hg.openjdk.java.net/jdk/jdk12/file/06222165c35f/src/java.base/share/classes/jdk/internal/reflect/Reflection.java#l130>. > This is known to have been working with older OpenJDK (at least 1.8). Steps to Reproduce: > Compile the provided test.cc with `g++ -I/usr/lib/jvm/java-12-openjdk-12.0.0.33-1.ea.1.rolling.fc30.x86_64/include -I/usr/lib/jvm/java-12-openjdk-12.0.0.33-1.ea.1.rolling.fc30.x86_64/include/linux /usr/lib/jvm/java-12-openjdk-12.0.0.33-1.ea.1.rolling.fc30.x86_64/lib/server/libjvm.so test.cc` and run it with `LD_LIBRARY_PATH=/usr/lib/jvm/java-12-openjdk-12.0.0.33-1.ea.1.rolling.fc30.x86_64/lib/server ./a.out`. Expected Result: > exit 0 Actual Result: > Exception in thread "main" java.lang.NullPointerException > at java.base/jdk.internal.reflect.Reflection.verifyMemberAccess(Reflection.java:130) > at java.base/java.lang.reflect.AccessibleObject.slowVerifyAccess(AccessibleObject.java:673) > at java.base/java.lang.reflect.AccessibleObject.verifyAccess(AccessibleObject.java:666) > at java.base/java.lang.reflect.AccessibleObject.checkAccess(AccessibleObject.java:638) > at java.base/java.lang.reflect.Constructor.newInstanceWithCaller(Constructor.java:490) > at java.base/java.lang.reflect.ReflectAccess.newInstance(ReflectAccess.java:166) > at java.base/jdk.internal.reflect.ReflectionFactory.newInstance(ReflectionFactory.java:404) > at java.base/java.lang.Class.newInstance(Class.java:590) > Aborted Source code for an executable test case: > #include <cstdlib> > > #include "jni.h" > > void handleError(JNIEnv * env) { > if (env->ExceptionCheck()) { > env->ExceptionDescribe(); > std::abort(); > } > } > > int main() { > JavaVM * vm; > JNIEnv * env; > JavaVMInitArgs args{JNI_VERSION_1_2, 0, nullptr, true}; > if (JNI_CreateJavaVM(&vm, reinterpret_cast<void **>(&env), &args) != JNI_OK) { > std::abort(); > } > auto const c1 = env->FindClass("java/lang/Class"); > handleError(env); > auto const id = env->GetMethodID(c1, "newInstance", "()Ljava/lang/Object;"); > handleError(env); > auto const c2 = env->FindClass("java/lang/Object"); > handleError(env); > env->CallObjectMethod(c2, id); > handleError(env); > } Workaround: > Call via JNI the suggested replacement of clazz.getDeclaredConstructor().newInstance() instead of the deprecated java.lang.Class.newInstance(). Change-Id: I85ff7102a18b98561f188e609873753546bc050d Reviewed-on: https://gerrit.libreoffice.org/71240 Tested-by: Jenkins Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
-
Mike Kaganski yazdı
... which was obviously omitted in commit 6b678c13, which introduced the correct value of 0x0491 (1169) in addition to the pre-existing 0x043C (1084). The new value is used when user sets relevant entry in Region and Language control panel applet (intl.cpl); having the old value in MSI summary information blocks auto-detection of proper installer language. [MS-LCID] Appendix A mentions that the new code is available since Win7, so we are good with the fix. [MS-LCID] https://docs.microsoft.com/en-us/openspecs/windows_protocols/ms-lcid/a9eac961-e77d-41a6-90a5-ce1a8b0cdb9cCo-authored-by: Eike Rathke <erack@redhat.com> Change-Id: Ic03b0e8eaf145c662fde48846cd5307e362d4d63 Reviewed-on: https://gerrit.libreoffice.org/71242Reviewed-by: Mike Kaganski <mike.kaganski@collabora.com> Tested-by: Mike Kaganski <mike.kaganski@collabora.com>
-
Stephan Bergmann yazdı
Had found > sh: -c: line 0: syntax error near unexpected token `(' > sh: -c: line 0: `file /Users/stephan/Software/lo/core/instdir/LibreOffice.app/Contents/Frameworks/LibreOfficePython.framework/Versions/3.5/lib/python3.5/site-packages/setuptools/script (dev).tmpl' in workdir/CustomTarget/instsetoo_native/install/install.phony.log on macOS. Change-Id: I4b01b5f0a292424a60558cdb8b431a3d8dbdf1d4 Reviewed-on: https://gerrit.libreoffice.org/71234 Tested-by: Jenkins Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
-
Noel Grandin yazdı
re-arrange the conditions so we avoid the dynamic_cast most of the time. This takes the file-open time from 40s to 29s for me. Change-Id: If65a63502fd545003137747605332be65d5e91ca Reviewed-on: https://gerrit.libreoffice.org/71229Reviewed-by: Noel Grandin <noel.grandin@collabora.co.uk> Tested-by: Noel Grandin <noel.grandin@collabora.co.uk>
-
Noel Grandin yazdı
Inherit from tools::WeakBase non-virtually, so that we can use a static_cast in tools::WeakReference::get instead of a dynamic_cast. This takes the file-open time from 1m21 to 40s for me. Add a clang plugin to make sure we don't accidentally end up inheriting from tools::WeakBase more than once. Change-Id: I9c7c36403333f99094e1f9d8cce2ecd9200377f9 Reviewed-on: https://gerrit.libreoffice.org/71231Reviewed-by: Noel Grandin <noel.grandin@collabora.co.uk> Tested-by: Noel Grandin <noel.grandin@collabora.co.uk>
-
Tor Lillqvist yazdı
(I know Swedish, this is not guesswork, there are no orthography differences.) Change-Id: I6f652f01dc95aa3a89fec1964c2a49109eb5e30b
-
Tor Lillqvist yazdı
On macOS, the system says it has a generic "de" dictionary. But on iOS, it says it is specificaly "de_DE". Assume it is good enough for the other German variants, too, for now. Change-Id: I745cc02c1ffa774dbcbfdc1d775ce358957f04f8
-
Caolán McNamara yazdı
spec says "An unsigned integer that MUST be 0xFFFFFFFF" for version so at least insist on that Change-Id: I8c11659d86fe122b1c8f61b1bf99e7b4b53ff306 Reviewed-on: https://gerrit.libreoffice.org/71227 Tested-by: Jenkins Reviewed-by: Caolán McNamara <caolanm@redhat.com> Tested-by: Caolán McNamara <caolanm@redhat.com>
-
Mike Kaganski yazdı
Change-Id: If846bb08ec6eceb307228bc628a4b186a3058051 Reviewed-on: https://gerrit.libreoffice.org/71230Reviewed-by: Mike Kaganski <mike.kaganski@collabora.com> Tested-by: Mike Kaganski <mike.kaganski@collabora.com>
-
Tor Lillqvist yazdı
Now with "Unipoll" this actually caused the app to hang. Unclear whether this ifdef had ever been strictly needed. (But before "Unipoll" it did not cause any harm.) Change-Id: I5cc399f76fcdb267eb96bf7790474c14f2ce9955
-
Tor Lillqvist yazdı
Change-Id: If5781ae101e58c26368e57787cb301f84ffa82a5
-
Stephan Bergmann yazdı
in its javac invocations, so we apparently need to set those variables instead of ant.build.javac.source/target (which are documented at <https://ant.apache.org/manual/javacprops.html> to be supported since Ant 1.7, but are apparently not used here). (Found when doing a test build with Java 12 and JAVA_SOURCE/TARGET_VER explicitly configured as 7 with <https://gerrit.libreoffice.org/#/c/71218/> "Allow to pass JAVA_SOURCE/TARGET_VER into configure", which caused building external/languagetool to nevertheless complain about unsupported 1.6 source/target switches.) Change-Id: I624f77b65ef81bc592a78559e67062a42cfff05a Reviewed-on: https://gerrit.libreoffice.org/71219 Tested-by: Jenkins Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
-
Gabor Kelemen yazdı
Found with bin/find-unneeded-includes Only removal proposals are dealt with here. Change-Id: I07d78df2d5aacbdaff7344a09812db91de17d535 Reviewed-on: https://gerrit.libreoffice.org/71113 Tested-by: Jenkins Reviewed-by: Miklos Vajna <vmiklos@collabora.com>
-
Stephan Bergmann yazdı
...for those who pretend they know what they are doing (e.g., to experiment with building against Java 12 which no longer supports Java 1.6 source/target, even though that results in a build that is not baseline compatible). Change-Id: I6e8617bc5b90212473652bdfac40bb48e9623a08 Reviewed-on: https://gerrit.libreoffice.org/71218 Tested-by: Jenkins Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
-
Noel Grandin yazdı
use template to make return type match input type, so we don't need to cast at the call sites Change-Id: I1e65f362c67f74c9a230cdbc1db12545b28eb499 Reviewed-on: https://gerrit.libreoffice.org/71216 Tested-by: Jenkins Reviewed-by: Noel Grandin <noel.grandin@collabora.co.uk>
-
Noel Grandin yazdı
regression from commit 7282014e Date: Fri Feb 1 15:15:16 2019 +0100 tdf#50916 Makes numbers of columns dynamic. Change-Id: I3913e14cadc11efa1d9e7405cdbaf96fe094508a Reviewed-on: https://gerrit.libreoffice.org/71220 Tested-by: Jenkins Reviewed-by: Noel Grandin <noel.grandin@collabora.co.uk>
-
Rtch90 yazdı
Change-Id: I5f94c8204a4c35470e891eb89cfbe74aad852bf4 Reviewed-on: https://gerrit.libreoffice.org/71084 Tested-by: Jenkins Reviewed-by: Katarina Behrens <Katarina.Behrens@cib.de>
-
andreas kainz yazdı
Change-Id: I3114afc6856f2a44127b2959408337c464d9ff88 Reviewed-on: https://gerrit.libreoffice.org/71147 Tested-by: Jenkins Reviewed-by: andreas_kainz <kainz.a@gmail.com>
-
DarkByt31 yazdı
Change-Id: I11c2ac4855a5fec2d26255ec3ab14ffe50e0fbbe Reviewed-on: https://gerrit.libreoffice.org/71066 Tested-by: Jenkins Reviewed-by: Mike Kaganski <mike.kaganski@collabora.com>
-
Luboš Luňák yazdı
Was broken in d077b19a. Change-Id: I5f0e0ca7c60f0f09381ee7eebd7d49d5533e07af Reviewed-on: https://gerrit.libreoffice.org/71145 Tested-by: Jenkins Reviewed-by: Luboš Luňák <l.lunak@collabora.com>
-
Noel Grandin yazdı
Change-Id: Ifdee7c4bd9ed3306530c7bc4ecf3017008e90c3d Reviewed-on: https://gerrit.libreoffice.org/71215 Tested-by: Jenkins Reviewed-by: Noel Grandin <noel.grandin@collabora.co.uk>
-
Mike Kaganski yazdı
For some reason, we have never exported the default column width to XLSX, although we have such export for XLS. This led to bugs like tdf#100946. Workarounds applied to those bugs have made LO to export columns which shouldn't have been exported, abusing "customWidth" attribute in XclExpColinfo::XclExpColinfo depending on if the column has size different from an app-global default; after that, sheet-local default was determined, and then columns which have mbCustomWidth set (i.e., different from app default), as well as those different from sheet-local default, were stored. Effectively, the hack had disabled the removal of defaults from maColInfos in XclExpColinfoBuffer::Finalize. We even had unit tests explicitly testing that we export those columns that Excel skips. The effect of that is not only unnecessary data in the file; the data was actually wrong (customWidth actually means that the width was edited manually by user, even if equal to the default, thus changing Excel handling of the column); and also Calc initializes all columns to the right of last used column for such a file. Only in case when app-global default happened to match sheet-local one, columns would have properties allowing them to be removed from maColInfos in the end of XclExpColinfoBuffer::Finalize, which still resulted in problems similar to the workarounded one. This patch implements proper export of the default column width to XLSX, thus partially reverting changes made for tdf#100946 in commit 40d892a2. Actually, our export to XLSX does not depend on the 5-pixel correction (see ECMA-376-1:2016 18.3.1.81), since the exported default depends only on most-used column width. XclExpDefcolwidth implementation was edited to only take the correction into account when exporting to XLS (to keep status quo; it's clear that XLS widths export and import implementation is incorrect: some empirical formula is used in XclTools::GetXclDefColWidthCorrection, that was introduced in commit 555d7029 for i#3006, that doesn't match any documentation, and changes widths of columns in exported XLS - something to be fixed independently). Change-Id: I227aca17e56247cbb839444717066a898987c4f1 Reviewed-on: https://gerrit.libreoffice.org/71132Reviewed-by: Mike Kaganski <mike.kaganski@collabora.com> Tested-by: Mike Kaganski <mike.kaganski@collabora.com>
-
Luboš Luňák yazdı
b1118883 changed msfilter to use GraphicFilter::ImportUnloadedGraphic() to lazy-load images from the document. However, that function in some cases simply reads the entire rest of the passed SvStream, which in this case is the entire .xls file. And the document from tdf#124828 is ~50MiB and contains ~4000 images => 100+ GiB memory required. Change-Id: I74926383204ec642eabb28b62e2cf2e1ff8054a9 Reviewed-on: https://gerrit.libreoffice.org/71136 Tested-by: Jenkins Reviewed-by: Miklos Vajna <vmiklos@collabora.com>
-
Miklos Vajna yazdı
Meant to fix the build problems reported in the mail thread starting at <https://lists.freedesktop.org/archives/libreoffice/2019-March/082340.html> "Build failure with latest ICU 64.1". Change-Id: I006b92f4737f5e56e50527dd954e8c0d339e75dc Reviewed-on: https://gerrit.libreoffice.org/71143 Tested-by: Jenkins Reviewed-by: Miklos Vajna <vmiklos@collabora.com>
-
Noel Grandin yazdı
This takes the loading time from 13s to 10.5s Avoid expensive text-invalidation while we are loading Change-Id: Ia74b84575fbf635b5ab5618c74ea2efc9f6ad827 Reviewed-on: https://gerrit.libreoffice.org/71043 Tested-by: Jenkins Reviewed-by: Noel Grandin <noel.grandin@collabora.co.uk> Tested-by: Noel Grandin <noel.grandin@collabora.co.uk>
-
Andrea Gelmini yazdı
Change-Id: I9cc5e13681b047fe772704b42796e210ad070601 Reviewed-on: https://gerrit.libreoffice.org/71148 Tested-by: Jenkins Reviewed-by: Julien Nabet <serval2412@yahoo.fr>
-
- 23 Nis, 2019 10 kayıt (commit)
-
-
Regina Henschel yazdı
When reworking command U the shape 'RingOutside' was not considered. Although it is a 'mso-' shape, it has angles in degree and not in fixed-point number in its defaults. The patch removes in addition the changes introduced by Symphony, which erroneously used total width and height instead of radii for the ellipse size. Change-Id: Iede8e346681796fb3e147dbcfaaa08ed8fe87c54 Reviewed-on: https://gerrit.libreoffice.org/70941 Tested-by: Jenkins Reviewed-by: Regina Henschel <rb.henschel@t-online.de>
-
Katarina Behrens yazdı
Use-case here is .ods document with 16k tracked changes. For every tracked change, Calc generates a formula cell. For every formula cell, clipboard content is queried (via ScDocument::IsClipboardSource, ScModule::GetClipDoc respectively). This is dog-slow on Windows Therefore don't query clipboard content while the document is in the process of being loaded. Instead return from ScDocument:: IsClipboardSource early and return 'false' because at this point of time it can't be reliably determined whether this doc is a clipboard source anyway Change-Id: If5aa62cbfb62fb326a3c73b4d9be839127d3c03b Reviewed-on: https://gerrit.libreoffice.org/71121Reviewed-by: Thorsten Behrens <Thorsten.Behrens@CIB.de> Tested-by: Thorsten Behrens <Thorsten.Behrens@CIB.de>
-
Caolán McNamara yazdı
Change-Id: I4ab78f35cb5a6ea31235662b6fad8d1e6c5d33b5 Reviewed-on: https://gerrit.libreoffice.org/71140 Tested-by: Jenkins Reviewed-by: Caolán McNamara <caolanm@redhat.com> Tested-by: Caolán McNamara <caolanm@redhat.com>
-
Stephan Bergmann yazdı
Change-Id: I7a36dd235a100b0d283eec86646dc167ac224ba7 Reviewed-on: https://gerrit.libreoffice.org/71133 Tested-by: Jenkins Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
-
Jim Raykowski yazdı
This reverts commit 220ce1a5. Including a header file of another module and even a module that is above is a no-go. Change-Id: I3c8107d8c667c34259bdfee9d83afb2651f5e11c Reviewed-on: https://gerrit.libreoffice.org/71142 Tested-by: Jenkins Reviewed-by: Jim Raykowski <raykowj@gmail.com>
-
Noel Grandin yazdı
comment from Miklos: > Won't this do a comparison of two null-terminated strings, while > the strings here are utf16, i.e. abc is a\0b\0c\0? That would mean > in practice you compare the first char only. introduced in commit 003d11f4 Date: Sat Apr 20 08:32:33 2019 +0200 tdf#81765 slow loading of .ods with >1000 of conditional formats Change-Id: I60ea18772753f50f880c65d8cff594267d724e76 Reviewed-on: https://gerrit.libreoffice.org/71114 Tested-by: Jenkins Reviewed-by: Noel Grandin <noel.grandin@collabora.co.uk>
-
Noel Grandin yazdı
regression from commit 1a5b12aa Date: Thu Apr 18 15:13:19 2019 +0200 optimise find/insert pattern Change-Id: I24edd5a89beed392c86058cd60155caf94bcf9ca Reviewed-on: https://gerrit.libreoffice.org/71137 Tested-by: Jenkins Reviewed-by: Noel Grandin <noel.grandin@collabora.co.uk>
-
Caolán McNamara yazdı
FillInfo is called more often than Initialize, so we've ended up with multiple AcceptChgDat strings Change-Id: I99e9398757d63e2c6315ec9c3101910d5978b13b Reviewed-on: https://gerrit.libreoffice.org/71119 Tested-by: Jenkins Reviewed-by: Caolán McNamara <caolanm@redhat.com> Tested-by: Caolán McNamara <caolanm@redhat.com>
-
Caolán McNamara yazdı
because we know all these get a label/tooltip set at runtime Change-Id: Icdcb341db6fe26b0a451c3b7229e8c0c4b0b2f86 Reviewed-on: https://gerrit.libreoffice.org/71109 Tested-by: Jenkins Reviewed-by: Caolán McNamara <caolanm@redhat.com> Tested-by: Caolán McNamara <caolanm@redhat.com>
-
Miklos Vajna yazdı
Change-Id: I3b1929d591fb3c44504b4818330a4e59aa3d587f Reviewed-on: https://gerrit.libreoffice.org/71125Reviewed-by: Miklos Vajna <vmiklos@collabora.com> Tested-by: Jenkins
-