Hi,
Xisco Fauli schrieb am 01.12.2025 um 10:56:
Hi Neil,
I believe that issue is the same I encountered when I tried to write a
UItest for tdf#169019 in https://gerrit.libreoffice.org/c/core/+/193104.
Later I wrote a minimal test showing the problem in
https://gerrit.libreoffice.org/c/core/+/193148 but I didn't investigate
it further. My plan was to get back to it at some point but I haven't
done it yet. I'm happy if you want to look into it.
So I submit the patch without unit test for now. A unit test needs to be
added separately then.
I can look whether I notice something in the Python code, but I work on
Windows and cannot test it.
Kind regards,
Regina
On 11/29/25 14:20, Neil Roberts wrote:
Regina Henschel <[email protected]> writes:
I'm looking for a co-author for a Python test for the patch
https://gerrit.libreoffice.org/c/core/+/194789. I work on Windows and
therefore cannot run Python tests.
Hi Regina,
I had a go at implementing the test with the patch attached below. It
works with the “gen” backend but if I use the “svp” backend (the default
for UI tests) then it gets stuck in a deadlock. At the end of the “with”
block for .uno:DataProvider, it tries to close the dialog with
executeAction and this waits for idle with
Scheduler::ProcessEventsToIdle. Somehow on the svp backend this blocks
with the solar mutex held in SvpSalInstance::DoYield on line 537. The
main thread never gets to idle because it is waiting for the “HTML Fetch
Thread” to join. That last thread is deadlocked by trying to get the
Solar Mutex in ScTable::CreateColumnIfNotExistsImpl.
At the point where the main thread is blocked, there is a comment saying
“TODO: use a SolarMutexReleaser here and drop the m_bNoYieldLock usage”.
I wonder if that means there is a bug in the SVP backend to fix before
we can make the test. I can try to look into it a bit more unless
someone already has some helpful insights?
Regards,
– Neil
