bug#60592: 30.0.50; emacsclient: exit with a different code when the request times out

bug#60592: 30.0.50; emacsclient: exit with a different code when the request times out

[Sean Whitton]

Hello,

When calling 'emacsclient --timeout N --eval' from scripts, it would be
nice if you could determine for sure when a failure was just due to a
timeout, rather than, say, a Lisp exception.

inotifywait(1) exits with code 2 to indicate a timeout.
emacsclient could do the same.

We might then consider adding a separate exit code for a Lisp error.
Then a failure of emacsclient to connect, a request that timed out, and
an error from Lisp would be readily distinguishable.

At present the best one could do is parse emacsclient's output, but
that doesn't seem robust.

-- 
Sean Whitton

bug#60592: 30.0.50; emacsclient: exit with a different code when the request times out

[Eli Zaretskii]

> From: Sean Whitton <spwhitton@spwhitton.name>
> Date: Thu, 05 Jan 2023 22:47:49 -0700
> 
> When calling 'emacsclient --timeout N --eval' from scripts, it would be
> nice if you could determine for sure when a failure was just due to a
> timeout, rather than, say, a Lisp exception.
> 
> inotifywait(1) exits with code 2 to indicate a timeout.
> emacsclient could do the same.

It would be more portable to have the exception handler send a special
message to emacsclient, so that it would know the reason without
guessing.

bug#60592: 30.0.50; emacsclient: exit with a different code when the request times out

[Sean Whitton]

Hello,

On Fri 06 Jan 2023 at 09:37AM +02, Eli Zaretskii wrote:

>> From: Sean Whitton <spwhitton@spwhitton.name>
>> Date: Thu, 05 Jan 2023 22:47:49 -0700
>>
>> When calling 'emacsclient --timeout N --eval' from scripts, it would be
>> nice if you could determine for sure when a failure was just due to a
>> timeout, rather than, say, a Lisp exception.
>>
>> inotifywait(1) exits with code 2 to indicate a timeout.
>> emacsclient could do the same.
>
> It would be more portable to have the exception handler send a special
> message to emacsclient, so that it would know the reason without
> guessing.

I think this is wholly complimentary to what I'm suggesting, right?

-- 
Sean Whitton

bug#60592: 30.0.50; emacsclient: exit with a different code when the request times out

[Eli Zaretskii]

> From: Sean Whitton <spwhitton@spwhitton.name>
> Cc: 60592@debbugs.gnu.org
> Date: Fri, 06 Jan 2023 10:57:18 -0700
> 
> Hello,
> 
> On Fri 06 Jan 2023 at 09:37AM +02, Eli Zaretskii wrote:
> 
> >> From: Sean Whitton <spwhitton@spwhitton.name>
> >> Date: Thu, 05 Jan 2023 22:47:49 -0700
> >>
> >> When calling 'emacsclient --timeout N --eval' from scripts, it would be
> >> nice if you could determine for sure when a failure was just due to a
> >> timeout, rather than, say, a Lisp exception.
> >>
> >> inotifywait(1) exits with code 2 to indicate a timeout.
> >> emacsclient could do the same.
> >
> > It would be more portable to have the exception handler send a special
> > message to emacsclient, so that it would know the reason without
> > guessing.
> 
> I think this is wholly complimentary to what I'm suggesting, right?

I didn't think so, but maybe I was missing something.

My point is that if we make the exception known to emacsclient, then
when the timeout ends without exceptions, it should be clear it was a
real timeout.

bug#60592: 30.0.50; emacsclient: exit with a different code when the request times out

[Sean Whitton]

Hello,

On Fri 06 Jan 2023 at 08:12PM +02, Eli Zaretskii wrote:

>> From: Sean Whitton <spwhitton@spwhitton.name>
>> Cc: 60592@debbugs.gnu.org
>> Date: Fri, 06 Jan 2023 10:57:18 -0700
>>
>> Hello,
>>
>> On Fri 06 Jan 2023 at 09:37AM +02, Eli Zaretskii wrote:
>>
>> >> From: Sean Whitton <spwhitton@spwhitton.name>
>> >> Date: Thu, 05 Jan 2023 22:47:49 -0700
>> >>
>> >> When calling 'emacsclient --timeout N --eval' from scripts, it would be
>> >> nice if you could determine for sure when a failure was just due to a
>> >> timeout, rather than, say, a Lisp exception.
>> >>
>> >> inotifywait(1) exits with code 2 to indicate a timeout.
>> >> emacsclient could do the same.
>> >
>> > It would be more portable to have the exception handler send a special
>> > message to emacsclient, so that it would know the reason without
>> > guessing.
>>
>> I think this is wholly complimentary to what I'm suggesting, right?
>
> I didn't think so, but maybe I was missing something.
>
> My point is that if we make the exception known to emacsclient, then
> when the timeout ends without exceptions, it should be clear it was a
> real timeout.

Right.  This is a good addition to my proposal -- no conflicts.

-- 
Sean Whitton