Closed jacobgkau closed 1 month ago
It's a difference in behavior between the cosmic and pop-shell launcher plugins.
This is because the new commit match more field of the desktop entries. In this case, it will match the "internet" categorie of firefox.
Somehow, strsim::jaro_winkler("terminal", "internet");
return a score of 0.66
, which is a lot. The plugin accept match > 0.6
.
I'm not sure the best way of fixing this. This is kinda related to https://github.com/pop-os/cosmic-launcher/issues/145. Maybe is would make sense to use different matching algorithm for different field. Like using a substring algo for Categories.
Maybe is would make sense to use different matching algorithm for different field. Like using a substring algo for Categories.
Ya, I think that could work. Alternatively, requiring a higher score like 0.8 for matching the desktop entry, and a lower score like 0.6 for matching title and app id in the toplevel plugin seems to work pretty well too.
Maybe is would make sense to use different matching algorithm for different field. Like using a substring algo for Categories.
Ya, I think that could work. Alternatively, requiring a higher score like 0.8 for matching the desktop entry, and a lower score like 0.6 for matching title and app id in the toplevel plugin seems to work pretty well too.
https://github.com/pop-os/freedesktop-desktop-entry/pull/22 does smtg like this
This seems to be caused by 172532056b404923b0bc068b524b23e81a371f18. It does not happen on ab7a71a057e3e2ce5fc200b056799a07cf3a9bb5. (FYI, @wiiznokes in case you have an idea of why this might be happening.)
Previously, search results would return application names even if a lot of windows are open:
On the latest commit, searching for the same string, I'm getting a bunch of irrelevant (but presumably closest-matching) window results, and no application results:
This essentially breaks launching apps with the launcher once more than 8 windows are open.
The good news is, this is not happening in the GNOME session (which is why I didn't catch it during regression testing).