Sorting doesn't respect roman numerals
To reproduce:
Create folders labeled 'I - Discovery', 'II - Threshold', 'III - Freedom'
Expected sort order is 'I - Discovery', 'II - Threshold', 'III - Freedom'
Actual sort order is 'I - Discovery', 'III - Freedom', 'II - Threshold'
Meanwhile, bash has no problem sorting it correctly
Activity
Thank you for reporting !
It's the same for all gtk based file manager (Tested caja, nemo and pcmanfm)
The origin of the problem is g_utf8_collate_key_for_filename.
gtk based file managers don't compare file names directly, but use some smartness (like needed for this issue) in order to generate a key for comparison, the so called "collate key".
So this issue should not be fixed in thunar, but upstream, in glib. It would be great if you could report an issue there: https://gitlab.gnome.org/GNOME/glib/-/issues
Actually, maybe your issue is the same one as this one: https://gitlab.gnome.org/GNOME/glib/-/issues/1150 .. than it should be sufficient to add a comment.
- closing this issue, since it should not be fixed in thunar
added 3. Not Xfce label
Thank you for looking into this @alexxcons . Should've known that it would be GTK as the underlying issue. Unfortunately, I can't report the issue at gnome gitlab, I'm permanently banned there.
-
Actually, I cannot reproduce the 'good' result via console
schwinn@schwinn-gsi:~/misc/programming/xfce/thunar_test/issue #1015_glib_sorting_bug/misc$ ls -al insgesamt 40 drwxrwsrwx+ 5 schwinn users 4096 11. Aug 22:23 . drwxrwsrwx+ 4 schwinn users 4096 10. Aug 17:37 .. -rw-rw-rw-+ 1 schwinn users 0 10. Aug 17:35 'FWave I - Discovery' -rw-rw-rw-+ 1 schwinn users 0 10. Aug 17:35 'FWave III - Freedom' -rw-rw-rw-+ 1 schwinn users 0 10. Aug 17:35 'FWave II - Threashold' drwxrwsrwx+ 2 schwinn users 4096 11. Aug 22:18 'Wave I - Discovery' drwxrwsrwx+ 2 schwinn users 4096 11. Aug 22:19 'Wave III - Freedom' drwxrwsrwx+ 2 schwinn users 4096 11. Aug 22:19 'Wave II - Threashold'I am not sure if such a roman number 'smartness' should be added at all. E.g. how you can know if a single V is meant to be a romand 5, or if it is just supposed to be a single V ?
Let's better not make assumptions, which can be wrong in many use-cases.
Instead, in case it is important for you, you could try using the roman number unicode table: https://symbl.cc/en/collections/roman-numerals/ and see if sorting works correctly for that (i doubt so)
I don't know why your console output is different to mine. I did find a file manager that is able to sort this correctly:
As for not being sure about roman number 'smartness', does your logic and reasoning part of your brain tell you "this is okay" when you look at the sequence I, III, II? Mine just screams "this is wrong"
