Errors in timers

Errors in timers

[Lennart Borgman (gmail)]

How is errors in timers supposed to be handled? It looks like those 
errors are only seen if you set debug-on-signal.

If there are not already any better way to handle this in Emacs I think 
it would be good if something to do it was added.

Here is some test code:

   (run-with-idle-timer 1 nil
                      (lambda()
                        (lwarn 't :warning "before")
                        (let ((x (/ 1 0))))
                        (lwarn 't :warning "after")
                        ))

Re: Errors in timers

[Andreas Schwab]

"Lennart Borgman (gmail)" <lennart.borgman@gmail.com> writes:

> How is errors in timers supposed to be handled? It looks like those errors
> are only seen if you set debug-on-signal.

Timer functions are run inside condition-case, so error are caught and
ignored.  See timer-event-handler.

Andreas.

-- 
Andreas Schwab, SuSE Labs, schwab@suse.de
SuSE Linux Products GmbH, Maxfeldstraße 5, 90409 Nürnberg, Germany
PGP key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: Errors in timers

[Richard Stallman]

    How is errors in timers supposed to be handled? It looks like those 
    errors are only seen if you set debug-on-signal.

Could you make the question more specific?  I am not sure
what "handled" means here.

Re: Errors in timers

[Lennart Borgman (gmail)]

Andreas Schwab wrote:
> "Lennart Borgman (gmail)" <lennart.borgman@gmail.com> writes:
> 
>> How is errors in timers supposed to be handled? It looks like those errors
>> are only seen if you set debug-on-signal.
> 
> Timer functions are run inside condition-case, so error are caught and
> ignored.  See timer-event-handler.

Yes, I do not remember exactly why it was added now, but I think it 
would be very good to be able to see the errors to. Currently the only 
way is to set debug-on-signal to t and that is a real sledge hammer.

As I see it it would be good to take care of this individually for 
different timers.

I have no proposal for exactly how. The need might be a bit different 
from when running a function directly. Timers can for example be run 
repeatedly and maybe the fail sometimes and otherwise perform well. 
Stopping them with a trace back might not be that good then. Perhaps 
some kind of logging is better in such cases.

I myself currently put a highlighted traceback in the message buffer and 
give a single message line to the user about that.