2015-02-23 0:24 GMT+01:00 Stefan Monnier <monnier@iro.umontreal.ca>:

> > Now I am curious, I'll probably dive in the mailing list history to find
> > out more about this motivation.
>
> I don't think there's much history about it in the mailing-list.
> Using post-self-insert-hook has the advantage of being more modular than
> a key-binding: you can make it do heredoc expansion in sh-script without
> affecting whether it does automatic reindent (via electric-indent-mode)
> or automatic pairing (via electric-pair-mode), or ...
>

Oh I see, too many modes abusing key rebinding have lead me to believe it
was the idiom for this kind of behavior.


>
> >> > Another use-case, even if you never use here-strings, is that you
> >> > want to enter a single <, but enter << instead.  You would expect
> >> > this mistake to be corrected with a single backspace, but it's not.
> >> We could special case this, potentially, but at the same time `undo' is
> >> your friend.
> > I know, but for some reason I can't get the hang of using undo in the
> flow
> > of writing text.
>
> It's definitely more intrusive, admittedly.  I'd welcome a patch which
> makes backspace actually undo the expansion.
>

With your solution of requiring an additional (printed) character after the
<, I think accidental expansion would get much rarer already.


>
> > This is a good idea, but in this case you also need to include < < -
> > as a trigger.
>
> Indeed, and probably < < ' as well.
>

By the way, the bug report is sent now, sorry for the delay, I was without
an internet connection for the week-end.

Thibaut