Re: debug-ignored-errors usage

[Eric Abrahamsen <eric@ericabrahamsen.net>]

Emanuel Berg <moasenwood@zoho.eu> writes:

> Eric Abrahamsen wrote:
>
>> The docstring explains it: these are errors
>> that _won't_ trigger the debugger even if
>> you've toggled debug-on-error t.
>
> This
>
>     (/ 1 0)
>
> triggers the error (arith-error) for me but the
> second time I do it it just says "Arithmetic
> error" in the minibuffer. And the third time
> etc ...
>
> So the debugger is only triggered once per
> error type? Is this why I never see the
> debugger half-window despite not doing anything
> to not see it? I just have Emacs on and before
> long there aren't any errors I haven't
> seen already?
>
> Also, putting beginning-of-buffer in
> `debug-ignored-errors' still echoes
> "Beginning of buffer" in the minibuffer when
> that happens.
>
> So what does it really achieve: you don't get
> the debugger window _the first time either_,
> but you still get the message every time
> like always?
>
> Because perhaps that message is not part of the
> actual error? Which, if so, is wierd, because
> from the name of the error, and spelling of the
> message, it sure sounds like they are a pretty
> close match!

It means that if you've run `toggle-debug-on-error', and then one of
those errors occurs, you _won't_ get the debug window popping up, which
normally all errors would do. Some errors are basic enough or common
enough that Emacs assuming you don't need to see a debug window for
them. For instance, beginning-of-buffer. I can toggle debug on error to
t, put point at the beginning of a buffer, and hit backward-char. That's
actually an error, though all you see is the message in the minibuffer.
If the beginning-of-buffer symbol were not in `debug-ignored-errors', it
would pop up a *Debug* window showing me the error backtrace. But
there's really no point in seeing a backtrace for beginning-of-buffer.

Emanuel Berg <moasenwood@zoho.eu> writes:

> Eric Abrahamsen wrote:
>
>> Your value looks very much like ispell or
>> some other spelling package has added it's
>> own regexp in there to skip debugging when it
>> couldn't find a word to check.
>
> But why does ispell produce an error at all, if
> it automatically on installation tells Emacs to
> ignore it?

It just tells Emacs that, even if `debug-on-error' is true, the user
doesn't really need to see the debug window if there's no word to check.
There's no "deeper information" to be gleaned from a backtrace.

Eric