bug#63311: 30.0.50; [PATCH] smtpmail-send-it split

[Manuel Giraud]

Eli Zaretskii <eliz@gnu.org> writes:

>> From: Manuel Giraud <manuel@ledu-giraud.fr>
>> Cc: 63311@debbugs.gnu.org
>> Date: Mon, 06 Nov 2023 16:56:33 +0100
>> 
>> Eli Zaretskii <eliz@gnu.org> writes:
>> 
>> >> Yes but how would I do that.  My idea was to left much code unmodified
>> >> (with all 'error' calls in case something goes wrong).  But how the
>> >> 'main-thread' could fetch these errors.
>> >
>> > Using thread-last-error, or at least that's what I had in mind.
>> 
>> It seems that 'thread-last-error' is global for all threads of Emacs so
>> I don't it would work.  Doesn't a thread record its error status and
>> message?
>
> Sorry, I don't understand what exactly you are asking and what you are
> saying.
>
> Yes, thread-last-error accesses a global storage, but I fail to see
> how this is a problem: if the main thread fetches the error in a
> timely fashion, it should still work.  And if, for some reason, this
> is not good enough, we could come up with a more fine-grained
> mechanism.  But first I'd like to see a description of a real problem
> with the existing mechanism: after all, it isn't expected that every
> single thread you launch will die due to an error, right?  So the
> situation where a thread dies and we need to access its error should
> be relatively rare, and a global resource could still be adequate.

So just for recap, the idea is to have a command
('message-send-asynchronously' as you suggested) that will just be
wrapper around 'message-send' and creates a thread to send the current
buffer.  Such a thread could be stored in a list and, by mean of a
timer, the main-thread would sometimes look if this thread is dead.

If it is dead and (thread-last-error) returns nil maybe we can conclude
that everything went well.  But if (thread-last-error) returns a
message, it could be message from any other Emacs thread: what to do in
such case?  Or worse: the thread is dead, (thread-last-error) returns
nil... but it was cleared from another part of Emacs.
-- 
Manuel Giraud