Re: Proposed tweak to org-agenda-skip-entry-when-regexp-matches-in-subtree

[Bernt Hansen <bernt@norang.ca>]

eric johnson <johnson.eric@gmail.com> writes:

> org-mode is fantastic.  Part of what makes it so awesome is that one can keep
> tweaking the software and the process.  In looking at my own usage, I
> noticed that
> I really wanted to simplify my org-todo-keywords.  I had a separate
> list of TODOs
> for projects (PROJ->PRST->DONE) and was wondering why I had to have that.
> Why couldn't I just get by with TODO->STARTED->DONE for everything, tasks
> and projects, and mark up projects with a tag.
>
> You can see what I'm aiming for with this example.
>
> (setq org-tags-exclude-from-inheritance '("project"))
> (setq org-todo-keywords '(
>               (sequence "TODO" "STARTED" "WAITING" "|" "DONE" "CNCL"))
> (setq org-stuck-projects '("project/STARTED" ("TODO" "WAITING"
> "STARTED") nil ""))
>
> * STARTED Stuck project           :project:
> ** DONE This was done
> * STARTED Not stuck project      :project:
> ** TODO Next project
>
> C-a # won't show "Stuck Project".  That's because the project line's "STARTED"
> is found in org-agenda-skip via the
> org-agenda-skip-entry-when-regexp-matches-in-subtree.
> I really want it to be
> org-agenda-skip-entry-when-regexp-ONLY-matches-in-subtree.
>
> To do that, I hacked up the function to capture a "begin" point after
> the headline.
>
> (defun org-agenda-skip-entry-when-regexp-matches-in-subtree ()
>   "Checks if the current subtree contains match for `org-agenda-skip-regexp'.
> If yes, it returns the end position of the current entry (NOT the tree),
> causing agenda commands to skip the entry but continuing the search in
> the subtree.  This is a function that can be put into
> `org-agenda-skip-function' for the duration of a command.  An important
> use of this function is for the stuck project list."
>   (let ((begin (save-excursion (org-end-of-line) (1- (point))))
> 	(end (save-excursion (org-end-of-subtree t)))
> 	(entry-end (save-excursion (outline-next-heading) (1- (point))))
> 	skip)
>     (save-excursion
>       (goto-char begin)
>       (setq skip (re-search-forward org-agenda-skip-regexp end t)))
>     (and skip entry-end)))
>
> If this change is too radical, it might make sense to modify
> org-agenda-list-stuck-projects to let the user define the skip function via
> an element in org-stuck-projects.
>
> I'm thinking something like this...
>
>   (let* ((org-agenda-skip-function
>            (or (nth 4 org-stuck-projects)
> 'org-agenda-skip-entry-when-regexp-matches-in-subtree))
>
> That would enable everyone to control the org-agenda-skip-function.

Hi Eric,

I've already moved to this type of a system with lazy project
definitions.  I changed my STARTED keyword to NEXT and clocking in
changes TODO to NEXT only if there are no unfinished subtasks for the
headline.

Stuck project views can be configured in a custom agenda view and that
is what I use now -- I don't use the standard stuck project definition
anymore - I just override the # key selection in the agenda so the keys
are all the same.

Not changing tasks with actionable subtasks to STARTED or NEXT on clock
in keeps the standard stuck project determination working.  So if you
clock time on the top-level task it just stays as TODO since there are
subtasks available to work on.

Details of my current set up are at http://doc.norang.ca/org-mode.html

HTH,
Bernt