Orderless matching for `org-set-tags-command`

[oantolin]

Orderless matching for `org-set-tags-command`

[oantolin]

Hi Omar, you weren't very active on github any more so I hope you are doing well

Hi Clemens! I'm fine, just busy with work. I hope you are doing well too.

This is an interesting sounding bug. I'll see if I can figure it out. I use org mode a lot, but don't really use tags much, so I wouldn't have noticed this myself. Thanks for reporting.

[clemera]

Good to hear, I'm fine, too. Just spend to much time in Emacs while I actually wanted to start learning more about Browser APIs 😆

I have looked a bit into it and the problem seems to be that orderless-filter let binds completion-regexp-list which affects the completion table org-tags-completion-function which uses all-completions and try-completion internally (which make use of completion-regexp-list internally).

I tried to simply rebind completion-regexp-list around the table and this will fix the described problem but when you partially input a tag it won't narrow down the list or insert a completion but continues to show all completions. I think it's related to how try-completion action is handled by orderless (the org completion table needs to be passed to try-completion while orderless-try-completion delegates to orderless-filter which uses all-completions which means the table does not get passed the right flag.)

[oantolin]

I wonder why the org-set-tags-command doesn't use completing-read-multiple for this. If you change the completing-read call to completing-read-multiple:

(org-trim (completing-read-multiple
           "Tags: "
           org-last-tags-completion-table
           nil nil (org-make-tag-string current-tags)
           'org-tags-history))

things seem to work OK.

[clemera]

Yeah, there are also problems with helm and selectrum which would be fixed by using completing-read-multiple, see here which contains some discussion about this, I wanted to fix it upstream as mentioned there but wasn't motivated enough to proceed.

[minad]

I believe this can be closed since it is simply not the fault of Orderless. Substring also does not work. See https://github.com/minad/vertico#org-set-tags-command for some workaround. And one can always use completing-read-multiple as a replacement.