There is a deadlock here when storing a ODatabaseDocument on a
non-main-thread while the main thread dispatches some event that calls
into ODatabaseDocument, while holding SolarMutex.
The storing of the document also stores BASIC libraries, and since
commit fca62934f492125ea6728fd6d09f0c66c9e4fa69 the SfxLibraryContainer
uses SolarMutex for locking.
Now we could re-investigate that problem, but it seems unrealistic to
expect ODatabaseDocument's implementation will never call anything
that acquires SolarMutex.
Resistance is futile. Your locking scheme will be assimilated.
Change-Id: I337d286f3e96c6b2e0dde8682b31faab3f508d20
Reviewed-on: https://gerrit.libreoffice.org/27590
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Michael Stahl <mstahl@redhat.com>
comphelper::AsyncEventNotifier is an amazing class that dispatches
events in separate threads, no doubt implemented during times of
exuberant optimism about the tractability of shared-state
multi-threading.
Unfortunately the authors forgot to think about how all those awesome
threads will be joined, so if they are somehow blocked, then it may well
happen that the events are dispatched when the main thread is already in
DeInitVCL, and the objects required for the dispatching already smell
somewhat funny.
This happens quite reproducibly when changing dbaccess' ModelMethodGuard
to lock the SolarMutex too, then CppunitTest_dbaccess_RowSetClones
crashes in DeInitVCL() because one AsyncEventNotifier thread was blocked
until then by SolarMutexGuard, and this test never Yields once its
document is loaded.
Try to fix this by joining the "DocumentEventNotifier" threads from
DeInitVCL() itself.
Since there's no rtl::WeakReference to go with rtl::Reference, refactor
the AsyncEventNotifier and create a new AsyncEventNotifierAutoJoin
that has to be used with std::shared_ptr and std::weak_ptr.
Change-Id: I50a0749795acb04b0776e543f7125767b697ea35
Reviewed-on: https://gerrit.libreoffice.org/27581
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Michael Stahl <mstahl@redhat.com>
Based on suggestion from Aron Budea.
And do something similar to most other places keeping vectors
of weak references where the code looks like it will hold more than
a few entries.
Measurements:
the 26 page file file takes
51s without my path
15s with this patch
the 69 page file file takes
5m28 without my path
51s with this patch
the 84 page file file takes
8m28 without my path
58s with this patch
Change-Id: I8da94c525fc73ebd969e0343c6f074be4f0063b1
Reviewed-on: https://gerrit.libreoffice.org/27093
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Noel Grandin <noelgrandin@gmail.com>
uno::Reference is only allowed to used with classes that have a
::static_type member.
So convert all those places to rtl::Reference.
Maybe we need some LIBO_INTERNAL_ONLY constructors on rtl::Reference and
uno::Reference to make this a little smoother?
Change-Id: Icdcb35d71ca40a87b1dc474096776412adbfc7e3
Reviewed-on: https://gerrit.libreoffice.org/25516
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Noel Grandin <noelgrandin@gmail.com>
...which (in LIBO_INTERNAL_ONLY) for Clang expands to [[clang::fallthrough]] in
preparation of enabling -Wimplicit-fallthrough. (This is only relevant for
C++11, as neither C nor old C++ has a way to annotate intended fallthroughs.)
Could use BOOST_FALLTHROUGH instead of introducing our own SAL_FALLTHROUGH, but
that would require adding back in dependencies on boost_headers to many
libraries where we carefully removed any remaining Boost dependencies only
recently. (At least make SAL_FALLTHROUGH strictly LIBO_INTERNAL_ONLY, so its
future evolution will not have any impact on the stable URE interface.) C++17
will have a proper [[fallthroug]], eventually removing the need for a macro
altogether.
Change-Id: I342a7610a107db7d7a344ea9cbddfd9714d7e9ca
(at least for LIBO_INTERNAL_ONLY), to help further reduce the occurrences of
sal_Bool across the code base
Change-Id: I70654a0cb56655984c717b7b894f26c9ab47536e
...which only happens ot work in environments where sizeof (bool) == 1. The
simpler alternative is to use the operator <<= template without passing explicit
UNO type information, anyway.
The std::nullptr_t overloads are needed to disambiguate calls with a nullptr
argument. (Which can at least be meaningful for VOID, but for other types what
it happens to do is store a default value of the given type.) As std::nullptr_t
is only C++11, this all needs to be LIBO_INTERNAL_ONLY.
Change-Id: Iff06a6ba94250bd4ae4afc937c2a2bfa75f0888f
probably not much performance benefit, but it sure is good at
identifying leftover intermediate variables from previous
refactorings.
Change-Id: I3ce16fe496ac2733c1cb0a35f74c0fc9193cc657
Reviewed-on: https://gerrit.libreoffice.org/24026
Reviewed-by: Noel Grandin <noelgrandin@gmail.com>
Tested-by: Noel Grandin <noelgrandin@gmail.com>
Searched source for using declarations.
Checked if those symbols reappear in the source file,
even in comments or dead code but not in #include statements.
If they don't reappear, remove the declaration.
Remove includes whose symbol got removed.
Change-Id: Ibb77163f63c1120070e9518e3dc0a78c6c59fab0
Reviewed-on: https://gerrit.libreoffice.org/24148
Reviewed-by: Jochen Nitschke <j.nitschke+logerrit@ok.de>
Reviewed-by: Noel Grandin <noelgrandin@gmail.com>
Tested-by: Noel Grandin <noelgrandin@gmail.com>
Sequence.h(xx), Any.h(xx) and Type.h(xx)
and remove unused using-declarations from these files.
Add a few missing includes provided by them.
Change-Id: I6b91b6d1fdf9d0496dd546c0aab9bdcc6831a5d4
Reviewed-on: https://gerrit.libreoffice.org/23805
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Noel Grandin <noelgrandin@gmail.com>
Mark some class methods as pure.
warnings look like this:
> dbaccess/source/core/api/RowSetBase.cxx
> 593 assertWithSideEffect 398 warning Assert statement calls
> a function which may have desired side effects: 'isAfterLast'.
Change-Id: I1b69340c3714be4678b599fa9a8ca933122aa857
Reviewed-on: https://gerrit.libreoffice.org/23981
Reviewed-by: Noel Grandin <noelgrandin@gmail.com>
Tested-by: Noel Grandin <noelgrandin@gmail.com>
dbaccess, desktop and drawinglayer.
Replace with C++11 delete copy-constructur
and copy-assignment.
Nothing special, only one unused include in
dbaccess/source/filter/xml/xmlfilter.cxx.
Change-Id: Iebabbc658215162450d4caf08d4fb2f116c456d9
Reviewed-on: https://gerrit.libreoffice.org/23918
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: David Tardon <dtardon@redhat.com>
insertRow notifies listeners that it is going to insert contents,
if a script listens to that and eventually triggers insertRow again then
one inside the other causes corruption and pestilence
Change-Id: I6b568d0a67f6108536d58c407b79d02bf29f297a
bibliography: click autofilter toolbar icon to get drop down, select an
entry and assert on use of invalid iterators.
In this specific case we want to reset them all, so add a new simpler
method to just do that and not do the (always supposed to be true) comparison
note: in ORowSet::checkUpdateIterator
at m_aCurrentRow = m_pCache->m_aInsertRow; these iterators are then invalid
the whole iterator caching and fiddling here is concerning.
Change-Id: I74952896f47ac482b6bacb26afe5b1cfcf6bc49c
