bug#2914: a signal to just slap emacs' face, not kill it

bug#2914: a signal to just slap emacs' face, not kill it

[jidanni]

Recently I made emacs so busy that I could get a C-g in edgewise, and I
had to wait a few minutes for it to finish what it was doing.

I could just from the shell do killall -s HUP emacs but that would just
kill it, without giving me a chance to save my files, etc.

So maybe there should be a way to send it a signal to just slap its face
a little, to get its attention, but not actually kill it.

Perhaps just to do what a C-g would have done, if it had got thru.

(It's besides the point, but I was using
http://test.wikipedia.org/wiki/Special:Allmessages?uselang=zh-tw and C-g
nor C-c C-k could break thru.)

bug#2914: a signal to just slap emacs' face, not kill it

[Stefan Monnier]

> Recently I made emacs so busy that I could get a C-g in edgewise, and I
> had to wait a few minutes for it to finish what it was doing.

That's clearly undesirable, tho I do know that it happens in
various cases.  In many cases it can be fixed pretty easily, tho.

> (It's besides the point, but I was using
> http://test.wikipedia.org/wiki/Special:Allmessages?uselang=zh-tw and
> C-g nor C-c C-k could break thru.)

So if you provide a testcase, or at least explain what it is you were
doing with that URL, maybe we can actually fix the problem.


        Stefan


PS: One common situation where C-g is unresponsive is when the redisplay
is too slow (e.g. navigating in a vvveeerrrryyy long line).  In such
a case, iconifying the frame displaying the "offending" buffer may get
you back in business (provided you have another frame or the
emacs-server open).

bug#2914: a signal to just slap emacs' face, not kill it

[jidanni]

Stefan Monnier <monnier@iro.umontreal.ca> writes:
> What were you doing?
I'm sorry. I was using the w3m add on package's
(w3m "http://test.wikipedia.org/wiki/Special:Allmessages?uselang=zh-tw")
and apparently had given it too many other keystokes so that C-g and C-c
C-k wouldn't get thru, or it was too busy rendering or something to
"listen to any phone calls". Anyway, that's all besides the point.

I'm thinking that maybe emacs could catch USR1 or USR2 signals, and
interpret them as C-g's, when the real C-g's just aren't getting thru.
But that is probably a bug in the first place.

Anyway, to test, just do something that makes emacs too busy to hear the
C-g's or maybe you typed a lot of other things that it hasn't read yet
before it can read your C-g's...

bug#2914: a signal to just slap emacs' face, not kill it

[Glenn Morris]

Version: 24.1

SIGUSR2 should now cause Emacs to stop and enter the debugger.
C-h v debug-on-event