bug#21435: 25.0.50; file-notify has problems after renames

[Michael Albinus]

Eli Zaretskii <eliz@gnu.org> writes:

>> > However, if all we want is to make sure the destination directory gets
>> > a notification (so it could auto-revert), then this already happens on
>> > MS-Windows (see the 'created' event above), and therefore nothing
>> > should be done on Windows to support the user request above.
>> 
>> It's not only the destination directory, it's also the source directory
>> which matters. Remember the initial use case, two dired buffers under
>> `auto-revert-mode' control. The moved file must appear in the
>> destination dired buffer, and it must disappear in the source dired buffer.
>
> Yes, and both events happen in this scenario, on Windows as well as on
> GNU/Linux.  So the fact that the move is not reported as a move will
> not cause any problems in this use case.

For that dired case, yes. I was answering to your statement "if all we
want is to make sure the destination directory ..."

>> Not necessary I believe. Due to inotify cookie mechanism, it will work
>> robustly. And don't forget gfilenotify, which sends a `rename' event
>> already on low-level.
>
> I'm guessing that gfilenotify only does that when its back-end is
> inotify.  E.g., I doubt that it could do this when it uses its
> fallback polling method.

Maybe, I haven't checked. But the point is that we send a `renamed'
event only we can trust there is a file move. Otherwise, we send
`deleted' and `created'.

>> >  . I'm not sure this kind of non-trivial logic is something that
>> >    belongs to filenotify.el; it could well have a better place in
>> >    auto-revert.el instead, as that is the level where the logic is
>> >    needed and understood, or even in the Dired-specific function that
>> >    auto-reverts a directory
>> 
>> If we only deliver `removed' and `created' events, none of those
>> libraries would have a chance to pair them to a rename action.
>
> They shouldn't rely on that in the first place, since this is
> unreliable, as we just saw.

Nope. When filenotify.el sends a `renamed' event, it must be
reliable. Because we got it from gfilenotify, because we could pair the
events via inotify cookies, whatever. When it is not reliable, we send
`deleted' and `created', which has less semantics than `renamed'.

> And in the case in point, it's unnecessary anyway, since all you need
> is to have events in both source and destination.  These events do not
> have to be 'rename' events.

In the use cases we know today, you are right. But there might be other
use cases where it matters. And again, `renamed' events provide more
information than single `deleted' and `created' events.

>> Essential information, like inotify cookies, will be lost.
>
> On filenotify.el level, yes.  I thought filenotify.el exists to try to
> present a more or less unified interface independent of the back-end.
> If such differences in back-end behavior are seen by clients of
> filenotify.el, then how is it different from invoking the back-end
> directly?

There is already a difference: native gfilenotify gives us a `rename'
event. Shall we convert it to `deleted' and `created'? This would reduce
the information.

>> And yes, this information will be needed. Recently, I saw a discussion
>> on sx, whether Emacs' `auto.revert-mode' could also support file
>> renaming. That is, when a buffer is associated by a file, and that file
>> is renamed outside Emacs, Emacs shall rename `buffer-name' and
>> `buffer-file-name', and then revert. Nice idea ...
>
> The problem we are discussing does not exist in this scenario, AFAIU.

That scenario would work only if there is a `renamed' event. How else
autorevert could decide, that a file has been moved? What is its new name?

Best regards, Michael.