Hey, > The crash is not specific to jack_capture_90.mp3; except for a period of > time a couple of weeks long leading up to the original bug. But do you see any pattern? I cannot reproduce it, too. > Has anyone on the team tried testing baloo with a large $HOME? Yes I think myself. Baloo indexed 1.387.086 files (mixed content) ; 650GB source file =>10GB baloo database. According your description of the bug, that is bigger than your dataset and be enough to reproduce. Why do you think we do not use the software they maintain? Baloo is working great for me, I haven't seen issues for a long long time. Okay what can you do to tackle down the bug: * debug symbols for libKF5BalooEngine.so.5 and liblmdb.so.0, so that we know what line triggers the segfault. because the interesting part is without debug symbols. * can you reproducible trigger the crash with jack_capture_90.mp3, or any other file? * did you checked upstream bug tracker - to find any matching bug? As this is an upstream issue, it needs to be fixed by upstream anyways, so we need to clarify if the issue is still at a current version. Otherwise create a bugreport at bugs.kde.org, they should also know better what infos they need to fix the issue. * start a Debian unstable VM/ KDE neon VM and try to reproduce the issue > It could take up to a month to reproduce this bug, and by then the reply will be "ignore for bullseye". This is a contributing factor to why baloo is in perpetual beta (I'd argue alpha, but subjectivity ;-)) Yes I know the timeframe small between the freeze and release to tackle down complicated issues. But please don't forget the other side. What should maintainers do if, they cannot reproduce the issue and upstream does not have stable LTS branches and doing monthly releases? The first thing is checking the current version: If it is fixed, than one can search for the bugfix to backport it or file the bug upstream otherwise. But without checking against the current version upstream reply most of the time: please update first. So I cannot reproduce it nor Norbert. Mark the bug as "unreproducible" is a valid approach, as we cannot tell if it is fixed or not. With more information it may be possible to decide. That's why the question is so important: Can you reproduce it somehow, so we are able to see the same issue? hefee
Attachment:
signature.asc
Description: This is a digitally signed message part.