org-super-agenda-keep-order breaks for auto groups

Question

org-super-agenda-keep-order breaks for auto groups

chasecaleb opened this issue 3 years ago · comments

Using (setq org-super-agenda-keep-order t), introduced recently by !167, causes a backtrace with :auto-category or :auto-parent. I didn't test other :auto-* groups, but I'm guessing it's likely to be the case for all of them.

Example agenda view:

(setq org-agenda-custom-commands
      '(("nc" "Next: category" todo ""
         ((org-super-agenda-groups
           `((:auto-category t)))))))

Resulting backtrace when attempting to view it via C-c a n c (some sensitive data replaced with <redacted>):

Debugger entered--Lisp error: (wrong-type-argument buffer-or-string-p (:name "Category: work" :items ...))
  org-entries-lessp((:name "Category: work" :items ...) (:name "Category: prof..." :items ...))
  org-super-agenda--group-items((... ... ... ... ... ... ... ... ... ... ... ... ... ... ...))
  org-super-agenda--filter-finalize-entries(#(<redacted>))
  org-agenda-finalize-entries((... ... ... ... ... ... ... ... ... ... ... ... ... ... ...) todo)
  org-todo-list("")
  #f(compiled-function () #<bytecode 0xf90fd689e6a624d>)()
  funcall(#f(compiled-function () #<bytecode 0xf90fd689e6a624d>))
  (let ((org-super-agenda-groups '((:auto-category t)))) (funcall '#f(compiled-function () #<bytecode 0xf90fd689e6a624d>)))
  eval((let ((org-super-agenda-groups '((:auto-category t)))) (funcall '#f(compiled-function () #<bytecode 0xf90fd689e6a624d>))))
  org-agenda(nil)
  funcall-interactively(org-agenda nil)
  command-execute(org-agenda)

I don't think that this has anything to do with my org files or other config, but let me know if you need more info and I can provide it.

Context:

No problems when org-super-agenda-keep-order is not enabled
org-super-agenda version: current master HEAD - commit fb5e2ef
org version: current bugfix HEAD - commit http://git.savannah.gnu.org/cgit/emacs/org-mode.git/commit/?id=73875939a
Emacs version: 28.0.50, master HEAD as of about a day ago - commit emacs-mirror/emacs@9a5a35c5a6

Alexander-Miller commented 3 years ago

Will do.

Adam Porter · Answer 1 · Thu Sep 30 2021 04:39:25 GMT+0800 (China Standard Time)

Thanks for reporting.

@Alexander-Miller Would you be able to take a look at this?

Alexander-Miller · Answer 2 · Sat Nov 13 2021 23:48:47 GMT+0800 (China Standard Time)

Sorry for the late response, I had the grand idea to first work through whatever came up for my own packages before coming back to this bug ...

Anyway, I have a diagnosis, so now it is time to discuss the treatment. The problem is that due to the :auto-something selector you don't have a list of agenda entries that you can simply re-sort, but a list of lists - entries grouped by their parents, category etc.

There's 2 fixes I see:

Check the super-agenda-groups for the presence of an :auto-* selector and if you find one re-sort every sublist individually. That's the technically correct, but more complex option. I also assume that the nesting won't go deeper than 1 level, otherwise it gets even more complex.
Just don't do it. Only sort if you have a list of strings and document that keeping order won't work for :auto-* selectors. That's the path of least resistance and complexity, though someone might ask for option 1 later on (not me, I'm ok with this option)

I'm fine building either fix - what do you think @alphapapa ?

Caleb Chase · Answer 3 · Wed Feb 16 2022 02:02:57 GMT+0800 (China Standard Time)

Check the super-agenda-groups for the presence of an :auto-* selector and if you find one re-sort every sublist individually. That's the technically correct, but more complex option. I also assume that the nesting won't go deeper than 1 level, otherwise it gets even more complex.

What about doing this recursively? That way it wouldn't be limited to just one level. I think that would be ideal if so.

Caleb Chase · Answer 4 · Wed Jun 22 2022 22:14:36 GMT+0800 (China Standard Time)

Hi @Alexander-Miller, are you still interested in fixing this or should someone else work on it?

Alexander-Miller · Answer 5 · Thu Jun 23 2022 03:34:11 GMT+0800 (China Standard Time)

It's still on the agenda somewhere. I've just haven't been able to dedicate enough time to it since I'm struggling to fight off all the other issues that keep piling on day by day. But I do plan to have a go at this eventually.

Caleb Chase · Answer 6 · Thu Jun 23 2022 04:30:48 GMT+0800 (China Standard Time)

Understandable, thanks for the reply! If I get some spare time to prioritize it I'll try to work on this, but I'm in the same boat as you.

Andrea Orru · Answer 7 · Thu Jan 26 2023 22:23:59 GMT+0800 (China Standard Time)

Is there any plan to fix this? Right now I'm just defining explicity groups instead of auto.

Adam Porter · Answer 8 · Sun Jan 29 2023 00:49:44 GMT+0800 (China Standard Time)

Yes, it is intended to be fixed when time permits.

@Alexander-Miller I'm guessing that recursively sorting the auto-grouped lists should work well enough, if you're still interested in implementing that.

Alexander-Miller · Answer 9 · Wed Feb 15 2023 15:56:06 GMT+0800 (China Standard Time)

@alphapapa
I am. I'll give it another go in the coming days.

Alexander-Miller · Answer 10 · Thu Mar 02 2023 05:06:26 GMT+0800 (China Standard Time)

PR is up, recursive sorting seems to do the trick. Let me know if anything's amiss.

Adam Porter · Answer 11 · Sun Sep 24 2023 05:34:25 GMT+0800 (China Standard Time)

It's past time to tag a stable release of v1.3, which has several changes already, so I'm targeting this for v1.4.