Re: easy-kill: drop-in replacement for kill-ring-save

[Dmitry Gutov]

On 08.10.2013 10:25, Leo Liu wrote:
> > On the other hand, extending it with new "things" will be more
> > difficult, since each one of them will require a dedicated key and some
> > headspace estate to remember it. And there's no mechanism to "expand"
> > the region from smaller to bigger things. Maybe I don't really need
> > that, though.
>
> The extensions are provided both for development and for experienced
> users.
>
> I have made +/- expand/shrink on list or sexp more lispy could you try
> it out?

Yes, thank you, looks good. Especially since you can still press `s' or 
`l' to get the previous behavior (I haven't noticed that initially).

But the point I was trying to make is that expand-region just expands to 
the next largest thing (of any kind), so the user doesn't need to 
remember all available kinds, or their keys.

> > The one problem I encountered has to do with the next command expecting
> > an active region: mc/mark-all-like-this-dwim, from the multiple-cursors
> > package, uses the active region, if it's present.
> >
> > With expand-region, the region becomes active as a result, and
> > mc/mark-all-like-this-dwim marks all occurences of the text in the
> > region. Not so with easy-kill: the text gets copied to the kill ring,
> > but the region is inactive, so mc/mark-all-like-this-dwim just uses the
> > symbol at point.
>
> Sorry I use neither of these packages. Is the intention to have region
> active after exit easy-kill? Or is it better to have, for example, C-SPC
> exit easy-kill and activate region? I'll experiment with these ideas
> later today.

I'd say neither, ideally.

Having region always active after pressing M-w would be weird.

C-SPC is a better option, but honestly, it's way too many keystrokes 
compared to expand-region when you just want to select the current symbol.

With expand-region, it's just one keystroke (it defaults to symbols), 
and with easy-kill it is `M-w s C-SPC'.

I think, ideally, easy-kill would activate the region only as far as the 
next command is concerned (so that command can use it). Not sure how to 
implement it best.
Maybe via a one-time pre/post-command-hook that would remove itself?