I just ran into this problem: [KDE] baloo_file_extractor constantly uses > 90% of CPU · Issue #6018 · chimera-linux/cports · GitHub
It seems that on Debian baloo is switched off. On Fedora it appears to be on just after installation. (totally forgot about it when I observed a very sluggish response from a fresh install). Is this sometthing that Fedora switsches by mistake/default/desgin?
I would guess that most people would assume the indexing is enabled by default and wanted.
and rely on indexing to find man pages and other docs, but may need help configuring indexing tools when they have massive numbers of ephemeral files and system become sluggish.
Right, but once the indexing completes then it only needs momentary spurts of use with file system changes. The initial indexing on a new system may be sluggish but after that it should settle down and operate normally.
I’m not part of the KDE Sig but I would imagine this is by design. Fedora Workstation (Gnome) also has this enabled by default.
The problem: indexing never completes. It gets stuck on this file OP-1 Manual.pdf - Nextcloud
Could be a broken or malicious PDF or a bug in whatever library is being used. Neither is a reason to not to continue with the default.
explain
There’s definitely something funky about that file - I’ve just downloaded it, put it into a directory and indexed that specific directory and I too see a 100% CPU utilisation when scanning it.
baloo_file_extractor doesn’t like it at all.
Displays fine though.
Editing it in Xournal++ crashes at Page 29, immediately after 10.5 Beat Match - something knackered in there. This isn’t a baloo specific issue - it’s the file content.
Probably worth a KDE bug report asking for Baloo to at least fail more gracefully in a situation like this, and skip the document rather than chugging away in vain.
A library bug or defective PDF should be reported “upstream” and fixed. I’ve had years of experience with bad PDF’s and authors who said “it worked for me”. There are sometimes ways to repair defective PDF’s if the author isn’t in a position to make changes.
Also true, but Baloo should be defensive enough not to allow itself to be foiled by a defective PDF.
This indeed is a bug. My knee-jerk solution for buggy software is uninstall.
Haven’t checked Baloo, but generally: a) apps offload PDF related tasks to some library, and b) there are lots of PDF files that don’t honour standards, use bad fonts, etc. You rely on the library to exit in a way that apps can recognize a failure, and apps to pass the information to the user rather than crashing.
This is absolutely a bug that should be filed upstream against baloo, with the kde developers, as it should handle this edge case better.
The KDE developers are nice folks, they mostly don’t bite.