| 5 Oct 2025 |
 Marie | does KDE upstream know about the issue that apps start up really slow when $XDG_DATA_DIRS has so many entries? | 15:57:11 |
 K900 | Very roughly, yes, but also I don't think there's going to be a solution any time soon | 16:22:42 |
| K900 | Because this is not something Qt will normally run into | 16:23:00 |
| K900 | And you'd have to do some kind of special case optimization for this stuff | 16:23:09 |
| Marie | And we can't do anything about it from our side? | 16:27:24 |
| K900 | One thing we could do is try and reduce the number of things that end up in XDG_DATA_DIRS | 16:27:54 |
| K900 | Some of that was already done in 25.11 but there's potentially more | 16:28:03 |
| K900 | Honestly I'm personally not very interested in hyperoptimizing this because on fast enough storage it kinda becomes a non-issue | 16:28:49 |
| K900 | Like yes we are still slower but it's not a relevant amount of slower | 16:29:10 |
| Marie | it's very noticeable to me tbh | 16:29:37 |
 Sandro 🐧 | Isn't that the main source that every KDE application takes like 3 seconds to open? | 16:31:50 |
| Sandro 🐧 | eg: first time starting krunner takes a good amount of time | 16:32:06 |
| K900 | I have no idea where you're getting 3 seconds tbh | 16:32:26 |
| K900 | I'm not getting anywhere near that | 16:32:31 |
| K900 | Even on a not particularly good SSD | 16:32:43 |
| Marie | clicking on the audio menu in plasma is incredibly slow for me | 16:33:15 |
| Marie | like a second | 16:33:28 |
| K900 | That's definitely not what I'm seeing | 16:33:38 |
| K900 | I wonder if this is just a filesystem thing | 16:36:01 |
| K900 | My workstations are all bcachefs or btrfs and it's fine | 16:36:06 |
| Marie | i'm on btrfs | 16:36:12 |
| K900 | I wonder if ext4 or zfs or something is actually just That Bad | 16:36:18 |
| Marie | on a samsung 990 Pro | 16:36:37 |
| Marie | how many things are in your $XDG_DATA_DIRS? | 16:37:20 |
| Marie | printenv XDG_DATA_DIRS | tr ':' '\n' | wc -l is 211 for me | 16:37:38 |
| K900 | Yeah that sounds very wrong tbh | 16:37:59 |
| K900 | Have you tried profiling it maybe | 16:38:10 |
| K900 | 125 here | 16:39:32 |
| K900 | Inside Yakuake | 16:39:32 |
| K900 | Which is itself a wrapper | 16:39:33 |
