https://bugs.kde.org/show_bug.cgi?id=477068
--- Comment #12 from Stefan Brüns <stefan.bru...@rwth-aachen.de> --- (In reply to tagwerk19 from comment #11) > (In reply to tagwerk19 from comment #10) > > Maybe when baloo_file does its initial scan through the filestructure to set > > up iNotify watches, it could do a sanity check of the folder name against > > the DocId. > OK, I think that would catch this particular issue. I mentioned already in the FSID change that there is no guarantee a document in the DB is currently reachable by its path, see network mounts, see thumbdrives, etc. But unfortunately everyone ignored my reservations, as everyone else apparently knows better how Baloo works than me ... > It seems that the FSID change triggers the issue of (comment #3): > > Confusing "empty" results when searching from a particular directory (as > > opposed to "Your Files"). That's the Bug 474973 > but there's no guarantee that this is the only cause. Doing a sanity check > of ensuring the folder is indexed before creating an iNotify watch would > catch any similar issues. Whatever the root cause. You fall into the same trap as others have before, these are not synchronous. First, inotify works with paths, and second, it works with a sequence of events. You will have intermediate states which may no longer be valid, e.g. when doing renames. Trying to "synchronize" inotify and DB state have been the source of inconsistent DB states in the past, this will only make things worse. The DB must be self-consistent, but may temporarily be in a different state than the FS. But actually, these comments are off-topic here. This BR is about DB inconsistencies caused by code changes, nothing else. -- You are receiving this mail because: You are watching all bug changes.
