How did that happen?

How did that happen?

[Harald Hanche-Olsen]

I just had emacs getting into this strange state: Everything seemed
normal, except for one little thing: The cursor would jump to the
beginning of the line after I inserted any character. This happened
everywhere, including in the minibuffer.

Needless to say, this made emacs difficult to use to say the least.
The only thing I could think of checking was if something had gotten
into after-change-functions, but its value was as expected:
(mew-draft-dynamic-highlight t), global value nil. (What fun to type
stuff like C-h v -retfa C-e -ahc C-e nuf TAB in order to investigate!)

Unfortunately, I haven't the faintest idea how to trigger this
behaviour, or I would submit a bug report right away.

Does anyone here have any notion what could possibly trigger this
behaviour? Or suggestions for what I might try, should I encounter it
again? I had to terminate that instance – it is just too painful to
run two instances of emacs simultaneously, or I might have kept it
around so I could investigate it some more.

Version info:
GNU Emacs 24.4.50.1 (x86_64-apple-darwin13.4.0, NS appkit-1265.21
Version 10.9.5 (Build 13F34)) of 2014-09-27 on airy

Built from this commit:
commit 13559b3ff594d19244a691ddaf966e0965654537
Author: Stefan Monnier <monnier@iro.umontreal.ca>
Date:   Sat Sep 27 12:26:54 2014 -0400

    * lisp/epg-config.el (epg-gpg-program): Use the plain program names rather
    than their absolute file name.

– Harald

PS. Sorry if posting to this list is inappropriate, but I suspect this
is too esoteric to fit on help-gnu-emacs. Correct me if I'm wrong.

Re: How did that happen?

[Andreas Schwab]

Harald Hanche-Olsen <hanche@math.ntnu.no> writes:

> Needless to say, this made emacs difficult to use to say the least.
> The only thing I could think of checking was if something had gotten
> into after-change-functions, but its value was as expected:
> (mew-draft-dynamic-highlight t), global value nil. (What fun to type
> stuff like C-h v -retfa C-e -ahc C-e nuf TAB in order to investigate!)

If you have the server running you can try to use emacsclient --eval to
inspect the Emacs instance.

> Unfortunately, I haven't the faintest idea how to trigger this
> behaviour, or I would submit a bug report right away.

Also check post-command-hook, for example.

Andreas.

-- 
Andreas Schwab, SUSE Labs, schwab@suse.de
GPG Key fingerprint = 0196 BAD8 1CE9 1970 F4BE  1748 E4D4 88E3 0EEA B9D7
"And now for something completely different."

RE: How did that happen?

[Drew Adams]

> I just had emacs getting into this strange state: Everything seemed
> normal, except for one little thing: The cursor would jump to the
> beginning of the line after I inserted any character. This happened
> everywhere, including in the minibuffer.
> 
> Needless to say, this made emacs difficult to use to say the least.
> The only thing I could think of checking was if something had gotten
> into after-change-functions, but its value was as expected:
> (mew-draft-dynamic-highlight t), global value nil. (What fun to type
> stuff like C-h v -retfa C-e -ahc C-e nuf TAB in order to
> investigate!)
> 
> Unfortunately, I haven't the faintest idea how to trigger this
> behaviour, or I would submit a bug report right away.
> 
> Does anyone here have any notion what could possibly trigger this
> behaviour? Or suggestions for what I might try, should I encounter
> it again? I had to terminate that instance – it is just too painful to
> run two instances of emacs simultaneously, or I might have kept it
> around so I could investigate it some more.

My wild guess is that an error causes the current operation to be
interrupted, and point is returned to bol (where it was sent by
part of the operation).  If it happens after each insertion, perhaps
it occurs during a function on `post-command-hook'.

You might not be able to use `debug-on-entry' for those functions,
but if it happens again you could at least do a binary search to
determine which of them is problematic, and then, say, use `message'
to track its progress.