Re: Putting the value of page-delimiter in some outline regexps?

[Eric Abrahamsen <eric@ericabrahamsen.net>]

Nicolas Goaziou <mail@nicolasgoaziou.fr> writes:

> Hello,
>
> Eric Abrahamsen <eric@ericabrahamsen.net> writes:
>
>> I've been trying to find a way to cordon off the bottom of my Org files,
>> to create an area for file-local variables and "LocalWords" and what
>> have you that Org doesn't consider part of the file's final heading.
>>
>> The usual Emacs thing to do seems to be using the value of
>> page-delimiter to demarcate the end of content, and the beginning of
>> meta data. I stuck that into the value of `org-outline-regexp', and also
>> hard-coded it into the "true heading" branch of `org-end-of-subtree'.
>> Over the past day this has worked fine for interactive use, and some
>> automatic Org-based tools. But I'm not so naive as to think that
>> catastrophic breakage couldn't be right around the corner as a result of
>> this!
>>
>> Anyway, I wanted to see if anyone else had tried this, or had any ideas
>> about it.
>
> I think Org already puts some care into preserving file-local variables
> at the end of the buffer, e.g., when moving around headlines. Maybe the
> issues you're encountering can be fixed the same way.

Do you mean `org-preserve-local-variables'? I think it would make sense
to put `page-delimiter' in the re-search-backward regexp. We could also
consider putting in the value of some of the ispell-*-keyword constants,
but it might also be reasonable to just ask the user to hide all that
stuff under a page-delimiter.

This still doesn't stop `org-next-visible-heading' or
`org-end-of-subtree' from treating the end matter as part of the last
heading. This would only be a minor inconvenience if I weren't trying to
write some automatic tools on top of Org. But perhaps I should be making
my tools smarter, not insisting that Org get smarter.

> Adding page delimiter in `org-outline-regexp' is clearly not subtle
> enough in any case, as it would affect every headline, not only the last
> one in the buffer.

No, I guess I didn't really think that was going to be a practical
solution.

Thanks for considering this,
Eric