Unknown reporter writes

So far, rather than running the indexer in the bg (autostart, daemon) I have been updating the index on-demand, via the recoll GUI. If, for whatever reason, I manually suspend the indexing operation… if I later attempt to again run the "update index" command later in the session, the recoll gui crashes.

Based on what I see reflected in resource usage, the "stop indexing" command isn’t having any effect. Resource usage doesn’t settle until I exit the recoll GUI. Upon restarting the recoll GUI, the "update index" command does successfully restart the indexing.

Although I can certainly avoid the habit of interrupting the indexing operation, this scenario seems like a legitimate bug.

Unknown User writes

Sorry, prior to reporting I didn’t notice this issue mentioned in issue #6 (uninterruptable xapian operation)

medoc writes

Hi and thanks for reporting this. The uninterruptible xapian operation thing is true, but normally it mostly affects querying, because indexing operations are usually of a finer granularity so that recoll takes control more often.

Could you please supply the exact recoll version, I’d like to take a closer look at this. The indexing thread has been a source of problems from the beginning, maybe it’s time I take another look.

Cheers,

jf

medoc writes

As of changeset [[https://bitbucket.org/medoc/recoll/changeset/596cbe0944d7|596cbe0944d7]], the indexing thread is gone for good. The GUI just executes a recollindex command and retrieves current status via a file. This change was comtemplated for quite a long time and has two advantages: * Resources are actually freed when indexing is done, because the indexing process exits (if you are not using real-time of course) * No more multithreading issues inside the GUI, a simpler code.

The indexing thread was just a bad idea.

So closing this, if the original problem was with 1.16.2, I can try to fix it in a minor update (sooner than 1.17), but I’d need a confirmation of the version which exhibits the problem, because stopping and restarting seems to work for me with 1.16.2