[SOLVED] Every login baloo_file crashes

[__name__]

Every time I login I get the same notification.

"We are sorry baloo_file closed unexpectedly.

You cannot report this error, because baloo_file does not provide a bug reporting address.

Details:
Executable: baloo_file, PID: 3074, Signal: Aborted(6), Time: 3/8/24 1:18 AM"

This is the developer information:
"The generated crash information is not useful."

[NichtDerHans]

In make.conf USE="-semantic-desktop".

[__name__]

I added -semantic-desktop in my make.conf, rebuilt the affected packages, and emerge --depclean removed baloo.
No error on reboot obviously since baloo is not present.

Thank you. Is there a place to reference how USE flags change package requirements?

[Goverp]

The Gentoo wiki page on baloo is a little help, but Arch linux page on baloo lists various useful commands to clean up the database. Simply deleting (or moving to a different name) the directory ~/.local/share/baloo/ should give you a clean start. That's assuming you want baloo.

Note that NichtDerHans "solution" removes baloo completely. You might or might not want it. I gave up on baloo some time ago, and use Recoll instead.
_________________
Greybeard

[__name__]

I'm not sure if I need it or want it. I don't often search for files. I typically browse through directories. Is there a beneficial use case that you use Recoll for?

If I need it I imagine I will figure that out relatively soon.

[flexibeast]

i use Recoll to index the large number of PDFs (many of them academic papers) that i have on my system, so that i can easily find them as required (i.e. not infrequently). And as i spend a lot of my time in Emacs, it's beneficial that there's a `consult`-based interface available for it.

[Goverp]

Recoll is a content indexing system that can handle PDFs, tar balls, email, documents, other stuff, and many downloads, at the cost of having to update the index either periodically or in the background. Baloo does the same indexing, in the background (though with some sort of scan at startup in case anything's been changed since it was last running - I assume this should be faster than a proper indexing run), though I couldn't find an easy way to search using baloo!

The alternatives are tools like grep and ripgrep. Depending on the amount of stuff in your filesystem, the devices used and the speed of your hardware, they are often fast enough to be useful without the cost of the indexing runs of Recoll. Their drawback is they have no understanding of the context of the search, so you may need to take this into account both when specifying the grep search and when interpreting the results. Recoll (and presumably baloo) take care of all that for you - for example, Recoll lets you search for a string in emails, and offers previews of the results correctly formatted, whereas in grep you'd have to specify where the emails are stored, and you'd get just a filename and short text extract as context. There may be tools with similar live searching abilities but better handling of the context, but I've neither looked for nor come across them.

The big question is how much you're willing to pay for this convenience. On my pretty fast system, with a 5-disk RAID10 array holding my home directory (about 100 GB of stuff, by volume mostly photos, video and audio, but also documents, email, source code and indeed the entire home directory of my previous development environment that I just dumped en-masse into anther directory, and all the cruft of modern life) a Recoll index update takes order of 15 minutes(even though those media files don't get indexed, or at least not much, as they contain almost no text), though I can happily work on other stuff while it does its thing. When I actually want to find that note or file I sent someone about my taxes, it's well worth the effort - though I tend to only update the index when I have such a search in mind.
_________________
Greybeard

[flexibeast]

i run recollindex as a daily cron job, around 0330, when i'm usually either sleeping or trying to.

[darkphader]

Rebuilidng kde-frameworks/kfilemetadata solved this for me.
_________________
WYSIWYG - What You See Is What You Grep