Re: A few questions about desktop.el

[Juri Linkov]

>> Overwriting the desktop of another Emacs instance is not a good thing to do.
>
> Well, that's exactly what happens without this patch.
>
>> Not writing the desktop would be acceptable if it displayed a message
>> about this failure.
>
> What is the exact use case?

Something like a message saying "Can't write the desktop because it is
locked by another Emacs instance".  Not a good solution either.

>> Another alternative is to write the desktop file with a different name.
>
> You can already do that, with or without the patch. The patch offers
> you the alternative of detecting the conflict when you run the second
> Emacs instance, so you chose not to load the desktop, and then
> manually set another name for it or create it in another directory.

But this requires complicate configuration on the user's part.

>> All on all, we could leave this to the user's discretion.  The user
>> should be careful about not creating a configuration where two Emacs
>> instances automatically write to the same file.
>
> I'm not sure what this means: are you proposing not installing the
> patch? Because the point (to me anyway) is not having a configuration
> where two instances write to the same file, but detecting the case
> when that happens by accident.

No, I propose installing this patch.  I had only doubts how it would
behave with my configuration.  But with the functions Davis implemented
in his patch, it is easy to write a configuration which won't fail, e.g.:

    (unless (desktop-owner)
      (run-at-time 3600 3600 'desktop-save "~"))

that starts automatic saving only when the desktop is not locked in
another Emacs session.

> The patch is very useful and works quite well (though I've added it an
> option to configure whether the second instance should automatically
> overwrite a conflicting desktop file, automatically skip loading it,
> or ask the user, instead of always asking as Davis' original patch
> does).

Such an option would be useful.

> I propose installing it and seeing what other people thinks of it.

I agree.

-- 
Juri Linkov
http://www.jurta.org/emacs/