bug#48584: 28.0.50; Incorrect hook ordering between local and global hooks with depth

[Philipp]

> Am 06.06.2021 um 11:12 schrieb Lars Ingebrigtsen <larsi@gnus.org>:
> 
> Philipp <p.stephani2@gmail.com> writes:
> 
>>> If I'm reading run_hook_with_args correctly, it'll loop over the local
>>> hook first (in order), and when it happens upon a t in that value, it'll
>>> then loop over the global value (in order), and then finish up the rest
>>> of the local ones.
>> 
>> Yes - and that's a reasonable behavior to expect, and we should document it.
> 
> I think it's an implementation detail that users should not depend on.
> Where that `t' ends up being depends on many subtle things like the
> order of your add-hook/remove-hook calls in your .emacs file.

What we think about this doesn't matter much.  As I explained, people depend on 
observable/observed behavior.

Besides, hook ordering for abnormal hooks is crucial, so we can't just say 
"don't depend on it."

> 
>> Correct hook ordering is crucial for abnormal hooks.  We already rely
>> on the very specific ordering behavior several times in the Emacs
>> codebase alone.  I've searched a bit through the Emacs codebase, and
>> found the following places where Emacs runs an abnormal hook with
>> `run-hook-with-args-until-success/failure' that also gains a local
>> part somewhere the codebase:
> 
> [...]
> 
>> That is, we rely on this "undefined" behavior already in very basic
>> operations such as saving buffers to file.  In Eldoc, we even
>> explicitly tell users to add functions to the local part of this
>> abnormal hook (eldoc-documentation-functions), thereby telling them to
>> rely on "undefined" behavior!  This hopefully shows that the behavior
>> is far from being undefined.
> 
> I don't think it shows any such thing.

Why not?  Clearly there are hooks that are both abnormal and partially-local in 
the Emacs codebase, and for those hooks this issue arises.

> 
> I think it'd be a good idea to implement (and document) something in
> this area that actually allows users to control the hook running order
> properly, which currently just isn't possible.
> 

Yes, and arguably the current implementation with the "shadow" ordering stored 
in a private symbol property is already kind of a hack.  However, the 
second-best option is still to document the current behavior.  As I've tried to 
explain, just stating "it's undefined" won't fly.