triggered when I noticed a class doing acquire() in the
constructor and then release() in the destructor.
found mostly by
git grep -n -B5 -e '->release()'
Change-Id: I96e43a3d30ffd9ae9a34275f24cd914d8f7b026f
Reviewed-on: https://gerrit.libreoffice.org/25806
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Noel Grandin <noelgrandin@gmail.com>
This code hasn't been touched (on purpose) in years, it appears
to have been used to resize part of the console. (It's prefed off)
Change-Id: Iaaeaa8a95c71ae91299bbeeea24869bf2a237b6b
Reviewed-on: https://gerrit.libreoffice.org/24663
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: jan iversen <jani@documentfoundation.org>
Tested-by: jan iversen <jani@documentfoundation.org>
regression from...
commit 789055bc2acb4c71483fd60ea258d158bd5aec10
Date: Tue Apr 12 16:39:03 2016 +0200
clang-tidy performance-unnecessary-copy-initialization
probably not much performance benefit, but it sure is good at
identifying leftover intermediate variables from previous
refactorings.
This case has a comment explaining the problem that appeared post
change.
Change-Id: Ib0c0883c57f103656cda00e3a94399a515d7fe41
The plan for the near future is to still ship the executable but replace
the interal use if possible with using the static library.
At some point when it is not needed for debugging anymore and everything
works correctly we should only build the uploader executable in dev
configurations. The huge disadvantage of the interal solution is that it
is nearly impossible for a user to upload a crash report if LibO starts
working correctly. Also LibO overwrites the file with the information
after the upload whereas the executable does not.
Change-Id: Ib9854946be3a34e580964c18e1a9c0cce16221d1
Reviewed-on: https://gerrit.libreoffice.org/25862
Reviewed-by: Markus Mohrhard <markus.mohrhard@googlemail.com>
Tested-by: Markus Mohrhard <markus.mohrhard@googlemail.com>
this list initialisation is short enough
to remove the local helper function
Change-Id: I17fab47e51fc0b1a0aa0462d517d7861f8a278e0
Reviewed-on: https://gerrit.libreoffice.org/25884
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Julien Nabet <serval2412@yahoo.fr>
Firebird 3.0 uses new database file format (ODS, On Disk Structure) so
that we need to use either ODS11 or ODS12 version of test database for
dbaccess_firebird test, depending on whether building against Firebird
2.5 or 3.0 libraries.
Change-Id: Idecdc35b0ac87ab7f46cb79b5c044c65423a2c7e
Reviewed-on: https://gerrit.libreoffice.org/25846
Reviewed-by: jan iversen <jani@documentfoundation.org>
Tested-by: jan iversen <jani@documentfoundation.org>
...that doesn't burn CPU by always directly returning again from a
dbus_connection_read_write call with zero timeout. But still doesn't look like
it uses DBus the way it's intended to. Help appreciated.
Change-Id: I0d130adfb921409a27a847053b0b3646dc566a86
Remove ViewBar and put contents in ActionBar
Change-Id: I6551d36503f659f076ecf2b6766c6a14db771a1d
Reviewed-on: https://gerrit.libreoffice.org/25833
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Samuel Mehrbrodt <Samuel.Mehrbrodt@cib.de>
exists for the lifetime of the ScGroupTokenConverter
otherwise in tdf#98880 ScDocument::InterpretDirtyCells releases
that backing storage that the DoubleVectorRefToken relies on, and
the ScVectorRefMatrix relies on that, so...
when sc/source/core/tool/interpr4.cxx calls ::IsString on the ScVectorRefMatrix
which calls ensureFullMatrix. That makes use of rArray.mpStringArray where
rArray's mpStringArray is set to that rArray by
FormulaGroupContext::ensureStrArray and the storage of mpStringArray belongs to
the FormulaGroupContext, but that context was reset and destroyed up the stack
in ScDocument::InterpretDirtyCells so the data is now invalid
We could turn the unique_ptr into a shared_ptr and have the ScGroupTokenConverter
take a ref to the currently active FormulaGroupContext to ensure any generated
DoubleVectorRefToken/SingleVectorRefToken point to valid data during the
lifetime of the ScGroupTokenConverter
Change-Id: Id457934cdff18570961cb261cf5c46b6ef8ea083
Reviewed-on: https://gerrit.libreoffice.org/25815
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Eike Rathke <erack@redhat.com>
Tested-by: Eike Rathke <erack@redhat.com>
It is a complicated scenario, but I think something like this is going
on:
Before Windows 10, only OpenGL 1.1 was supported in RDP sessions. Our
code detects RDP and tries to disable the use of OpenGL in that
case. This code for this is in the neighbourhood of the blacklisting
of OpenGL implementations. But the use of OpenGL in GLTF and chart2
does not go through code paths that would take that into
consideration, so it tries to use OpenGL anyway, and eventually we end
up in InitGLEW(), where the recently added check for OpenGL 3.0 or
newer causes an (intentional) abrupt termination of the process.
Change-Id: Ia57bfc8d2cefe4e6d7a00282973608d17967d3fd
ReadLegacyAddons modifies its vectors of maDecks and maPanels in this case, but
a load of things have (c++) references contents of the original contents.
Its such a rats nest that the easiest thing seems to be to make them
vectors of shared_ptrs and hold DeckDescriptor and PanelDescriptor
by shared_ptr and it all works out
Change-Id: I3f628e12c7d5f4224d14d5e0769e450ce893fb54
This is a follow-up of 09981cd6383ecb99e4b6c83b98b03af5cf3ff59b.
Inspired by coverity#1362478 etc.
Change-Id: If5507c52552393cccbb255cbf4b7e9f4b80db917
Reviewed-on: https://gerrit.libreoffice.org/25840
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Michael Stahl <mstahl@redhat.com>
so rename it to that and add a OpenThenSwitchToDeck that actually
does that, using the Toggle varient as the callback from the
sidebar button which toggles the current deck on/off
which retains the features of
// tdf#67627 Clicking a second time on a Deck icon will close the Deck
// tdf#88241 Summoning an undocked sidebar a second time should close sidebar
but means that calls to OpenThenSwitchToDeck from e.g. slide layout
don't auto close it if that deck is already open
Change-Id: I3e3724626b93447a7ab6bc7032e9c6839dabcf55
Means nothing to this class and reduces spam in our unit test logs
Change-Id: I56a1d01b98372691003410b6d7b0e5326b2ee7eb
Reviewed-on: https://gerrit.libreoffice.org/25828
Tested-by: Noel Grandin <noelgrandin@gmail.com>
Reviewed-by: Noel Grandin <noelgrandin@gmail.com>
A script to search our test logs and sort the messages by how common
they are so we can start to reduce the noise a little.
Change-Id: I8a6e6167c42447f9869ac700300d1b243f055e2b
this is similar to
commit 6acd5c45c764d81aea1539e66adbfadb51df0aa3
Author: Caolán McNamara <caolanm@redhat.com>
Date: Wed Jan 21 15:25:03 2015 +0000
Resolves: fdo#87601 specific html doc hangs on load
and is a consequence of the old changeover from unsigned shorts for old string
class to signed int for new string class
Change-Id: I47022c641cbc329d8802765927005d43211b0d88
