What am I doing wrong in this org-ql-block query using parent or ancestors?

[al3xandru]

The following 2 queries are giving me surprising results making me realize that I don't fully understand how they work.

(org-ql-block '(parent (heading "[INBOX]")))

(org-ql-block '(and
                 (todo "TODO" "WIP")
                 (tags "work")
                 (ancestors (heading "[INBOX]"))))

The structure of the org file is:

* [INBOX]
** TODO Task a
** TODO Work task A :work:
* [NOTES]
* 2024
** 2024-W08

The first returns every item in the file.
The second returns every TODO/WIP tagged with work across the file.

It looks like my usage of parent and ancestors is not correct. I've read the docs and searched for similar examples. I'm probably missing something or skipped an important element as I have 2 more usages of ancestors that are working well (the only exception being that they are deeper in the outline structure).

Could anyone point me into the right direction? thank you

[alphapapa]

Hi Alex,

Do you get the same incorrect results when using org-ql-search instead of org-ql-block?

[al3xandru]

1. Yes.
2. It seems there is an issue that starts with (heading "[INBOX]") which instead of returning a single result it returns 4 items.
3. The other 3 results are children of "[INBOX]" and thus would not explain the query going outside that parent.
4. What bothers me is that the exact type of queries return the results I expect when matching deeper level items

[al3xandru]

Taking a hint from point 4 above, I have moved the INBOX item and all its children to a lower level. The parent & ancestors based queries returned the correct results (well, the results I expected at least)

My understanding of heading predicate is that it matches if the heading text matching the string(s). Is that correct? I can't find a good explanation why there are 4 results for the (heading "[INBOX]") query (the parent heading is not prepended to the child, right? otherwise all children would match)

[alphapapa]

Well, I don't understand what's going on here, because I wasn't able to reproduce the problem. Here's what I did:

1. Using https://github.com/alphapapa/with-emacs.sh, run with-emacs.sh -i org-ql.
2. Make file /tmp/test.org, paste your contents into it, and save it.
3. Run org-ql-search on that file with each of the queries you gave. They both returned the correct, expected results, without any headings that didn't belong.

I'd suggest trying to reproduce the problem with those steps. Then be sure to mention what versions of Emacs, Org, and org-ql you're using.

[al3xandru]

1. I have tried to use with-emacs.sh and it complains about unsupported o option.
2. In the with-emacs version, I have installed org-ql using package-install, and then required it
3. Copied over my org file. Then chopped it down to a short version that still returns what I consider invalid results:

#+TITLE: my life organized
* [INBOX]
:PROPERTIES:
:ID:       967A3D3F-D8B9-487B-9803-60BAC1F9853A
:END:
** TODO Subtask 1
** TODO Subtask 2
* 2024
:PROPERTIES:
:ID:       0FCB3663-B044-4BF1-8A3C-CDAA3F2C61F0
:END:
** 2024-W08
:PROPERTIES:
:ID:       04B8D416-24EA-4710-976C-ABF60F7BA317
:END:
*** [#B] Weekly Program [52%]
**** TODO Power over
*** 2024-02-20 Tuesday
***** DONE Start work 
SCHEDULED: <2024-02-20 Tue 17:00>
***** DONE Do Good Work
CLOSED: [2024-02-20 Tue 17:20]
****** DONE Estimate available work time (1h30ʹ)
CLOSED: [2024-02-20 Tue 17:12]
21:45-17:15=4h30ʹ-1h30ʹ=3h
- 1.5h Meetings

4. executed the query (ancestors (heading "[INBOX]")) which returns 4 results instead of 2

[al3xandru]

As regards the with-emacs.sh does this command look ok to you:

Running: emacs --quick --title Emacs (config: /var/folders/f4/5nbbgd055px6mv0hftysx5cw0000gn/T/tmp.KYt0p3Cj8J) --eval (setq user-emacs-directory (file-truename "/var/folders/f4/5nbbgd055px6mv0hftysx5cw0000gn/T/tmp.KYt0p3Cj8J")) --eval (setq user-init-file (expand-file-name "init.el" user-emacs-directory)) -l package --eval (when (boundp 'native-comp-eln-load-path) (push (expand-file-name "eln-cache" user-emacs-directory) native-comp-eln-load-path)) --eval (add-to-list 'package-archives '("gnu" . "https://elpa.gnu.org/packages/") t) --eval (add-to-list 'package-archives '("melpa" . "https://melpa.org/packages/") t) --eval (add-to-list 'package-archives '("org" . "https://orgmode.org/elpa/") t) --eval (setq package-user-dir (expand-file-name "elpa" user-emacs-directory)) --eval (package-initialize) --eval (package-refresh-contents) --eval (when (file-exists-p user-init-file) (load-file user-init-file)) ./with-emacs.sh -o d:e:hi:OPR -l dir:,debug,emacs:,help,install:,no-package,no-org-repo,no-refresh-packages -- -i org-ql

[alphapapa]

This appears to be #365

[al3xandru]

Ha... I did not pick up 365 when doing a search prior to posting.

Any suggestions related to with-emacs? It's an interesting utility that could prove useful. I'm on a macOS environment.

[alphapapa]

Ha... I did not pick up 365 when doing a search prior to posting.

And a quick check does not reveal the cause of the problem. It appears that org-ql--value-at is returning an incorrect non-nil value for the parent node of the nodes that should not be in the result, but I don't know why.

I guessed you are on a Mac. You will need to install GNU userland (coreutils, or whatever it's called in Homebrew) and make sure your PATH is configured so the GNU commands come first in your environment. Please consult other Mac users.

[alphapapa]

This appears to be #365

Correction, this appears to have been a bug in the heading predicate, which just happened to be noticed in conjunction with the parent predicate here. I'll be releasing a fix in a v0.8.5 release.

[alphapapa]

See also new release which fixes ancestors and parent: https://github.com/alphapapa/org-ql/releases/tag/v0.8.5

[al3xandru]

If anyone reads this discussion again and wonders about with-emacs on macOS:

1. if running brew, then brew install gnu-getopt
2. macOS has getopt in /usr/bin/getopt thus, put on the gnu-opt bin dir first in the PATH (e.g. /opt/homebrew/opt/gnu-getopt/bin) before running with-emacs.sh

(@alphapapa if you think this is useful for other with-emacs potential users, please do use it as you see fit)

[alphapapa]

Thanks. It's also discussed at alphapapa/with-emacs.sh#3, which links to a similar discussion.