This helps to avoid the expensive allocation and
freeing of OUString at some places.
Change-Id: I7d49974ce13799fcf62b989ce5d3c61b01316190
Reviewed-on: https://gerrit.libreoffice.org/34011
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Michael Meeks <michael.meeks@collabora.com>
Tested-by: Michael Meeks <michael.meeks@collabora.com>
...that had been like that ever since ff09f8d6ef2c613037f039dbcafcad6b76243a48,
"MWS_SRX644: migrate branch mws_srx644 -> HEAD" at least, for no apparent
reason.
Change-Id: I6f8bd44179b232b7b39e493d443a443a60522a5c
This apparently wanted to check whether the $ starting a potential pattern
occurrence was escaped by a preceding $. However:
* The check itself was broken, erratically looking into sPattern instead of
_inout_rFileURL for the $.
* The check was bogus, as it would have misinterpreted e.g. "$$$(loggername)".
* The resulting string is documented (in
officecfg/registry/schema/org/openoffice/Office/Logging.xcs) to be passed
through css.util.PathSubstitution. However, neither the PathSubstitution
service's documentation (in offapi/com/sun/star/util/PathSubstitution.idl) nor
its implementation (SubstitutePathVariables::impl_substituteVariable in
framework/source/services/substitutepathvars.cxx) appear to support esacping
$ with a preceding $. So it looks more reasonable to remove the check
completely here.
Change-Id: I445493d444904cd54f166adcbf870a918d1f9982
...after e57ca02849c3d87142ff5ff9099a212e72b8139c "Remove dynamic exception
specifications" dropped uses of UNO exception types from those generated files.
This needed first adapting client code that implicitly depended on those
#includes in the generated files it included ("Add missing #includes":
6dce9c6757823b9e89863716ae70ff4e8ddd4e60,
a34d9150b419ee7471b96599d877ff1091d6b567,
fed3783e77e00635e50622e7229b06c9e25451f1).
Change-Id: Ia1defda2d586b69092f53feef3644875192cbea4
Reviewed-on: https://gerrit.libreoffice.org/34014
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
This reverts commit b5e3f8a5fa98a249ecd50021c33cf2a5c7a3b4fc.
The problem is this:
==24217== Conditional jump or move depends on uninitialised value(s)
==24217== at 0x29A25FCE: SfxObjectShell::SetError(unsigned int, rtl::OUString const&) (objmisc.cxx:220)
==24217== by 0x29A35E6E: SfxObjectShell::ImportFrom(SfxMedium&, com::sun:⭐:uno::Reference<com::sun:⭐:text::XTextRange> const&) (objstor.cxx:2300)
==24217== by 0x29A3705C: SfxObjectShell::DoLoad(SfxMedium*) (objstor.cxx:765)
==24217== by 0x29A6BC48:
SfxBaseModel::load(com::sun:⭐:uno::Sequence<com::sun:⭐🫘:PropertyValue> const&) (sfxbasemodel.cxx:1802)
The commit is bogus because it introduces a
DynamicErrorInfo::GetErrorCode(), which overloads
ErrorInfo::GetErrorCode(), which is used at least in
DynamicErrorInfo_Impl::RegisterEDcr() and used to return a constructor
argument of DynamicErrorInfo but now returns pImpl->lErrId,
which is what this statement is trying to initialize.
Ultimately this causes my clang+ASAN build to fail because the
uninitialized error code happens to be detected as a mere Warning:
Test name: testMathMalformedXml::Import
assertion failed
- Expression: !xComponent.is()
- loading succeeded: sw/qa/extras/ooxmlimport/data/math-malformed_xml.docx
Change-Id: I9141144e0bc356ee54279948f2fce036d1831a86
... then don't silently ignore it but just abort. Evidently this cannot
happen, because if it were a real possibility the author of this
exception handler would have at least called SetError here.
Change-Id: Iaa5f1f1d4cac388ae08214f6031a8bc279e6fceb
Unlikely to fix the issue, but may help.
Change-Id: I3c319f550e86f2ab731b072d86d258dca12d34fa
Reviewed-on: https://gerrit.libreoffice.org/33998
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Michael Meeks <michael.meeks@collabora.com>
