icomplete-mode as replacement for iswitchb

icomplete-mode as replacement for iswitchb

[Ole Laursen]

Hi!

I ending up using iswitchb many years ago after trying all other
alternatives at the time (iswitchb had the most polished buffer
switching implementation IMHO). Now I'm on 24.4 and have been trying
to switch to icomplete:

1) Turning on icomplete makes C-x C-f show a list of matches (despite
icomplete-show-matches-on-no-input being nil - this is a little
jarring, but I think I can get used to it)

2) However C-x b doesn't show any matches on no input - this breaks my
quick-switch-between-a-few-buffers-in-working-set habit (C-x b, C-s a
couple of times, RET)

3) RET in C-x b on a partial match means "open new buffer with partial
name" - I am trying to get used to C-j instead, but after a while I
think RET is more ergonomic in many cases because it frees you from
having to time a Ctrl press (e.g. f, o, o, RET is easier than f, o, o,
C-j)

I don't see why you'd want to be able to start a new buffer while
switching buffers. Switching buffers is a high-speed operation done
many times a day.

I realize some of this comes from old Emacs traditions, but from my
point of view, making sure RET has the same result as C-j when
switching buffers and either turning on
icomplete-show-matches-on-no-input as default, or even better only for
some completion scenarios where it makes sense (working set ones),
would be really helpful.


Ole

Re: icomplete-mode as replacement for iswitchb

[Stefan Monnier]

> 1) Turning on icomplete makes C-x C-f show a list of matches (despite
> icomplete-show-matches-on-no-input being nil - this is a little
> jarring, but I think I can get used to it)

I think that's a bug.  Could you M-x report-emacs bug so we don't forget it?

> 2) However C-x b doesn't show any matches on no input - this breaks my
> quick-switch-between-a-few-buffers-in-working-set habit (C-x b, C-s a
> couple of times, RET)

I guess we could change icomplete-show-matches-on-no-input so it can be
set differently for different completions.

> 3) RET in C-x b on a partial match means "open new buffer with partial
> name" - I am trying to get used to C-j instead, but after a while I
> think RET is more ergonomic in many cases because it frees you from
> having to time a Ctrl press (e.g. f, o, o, RET is easier than f, o, o,
> C-j)

I think this will have to stay a user-config (you can change that for
yourself in icomplete-mode-map).


        Stefan

Re: icomplete-mode as replacement for iswitchb

[Ole Laursen]

Thanks for the kind answer.

2014-11-11 17:31 GMT+01:00 Stefan Monnier <monnier@iro.umontreal.ca>:
>> 1) Turning on icomplete makes C-x C-f show a list of matches (despite
>> icomplete-show-matches-on-no-input being nil - this is a little
>> jarring, but I think I can get used to it)
>
> I think that's a bug.  Could you M-x report-emacs bug so we don't forget it?

I think it's perhaps triggered by the path. But yes, I'll report it.

I played around with things a bit more and found another more serious
bug so I'm back on iswitchb. I'll report that too.

>> 3) RET in C-x b on a partial match means "open new buffer with partial
>> name" - I am trying to get used to C-j instead, but after a while I
>> think RET is more ergonomic in many cases because it frees you from
>> having to time a Ctrl press (e.g. f, o, o, RET is easier than f, o, o,
>> C-j)
>
> I think this will have to stay a user-config (you can change that for
> yourself in icomplete-mode-map).

Yeah, I see that it's a useful difference for C-x C-f. It works the
way I expect for M-x completions, so perhaps my request is really to
remove the possibility of opening a new named buffer through C-x b - I
think then RET and C-j would be the same. I don't know if that's out
of the question.


Ole

Re: icomplete-mode as replacement for iswitchb

[Ole Laursen]

After getting back to iswitchb I think that while the completion stuff
in icomplete feels superior to what iswitchb is doing, the UI is at
the moment worse (in addition to the other things I've mentioned, if
you turn on matches on no input C-x b with icomplete-mode will still
show a redundant "(default <buffer>)" thing).

So while it looks promising, I don't think it's really a replacement
for iswitchb yet.


Ole

Re: icomplete-mode as replacement for iswitchb

[Stephen Leake]

Ole Laursen <olau@iola.dk> writes:

> After getting back to iswitchb I think that while the completion stuff
> in icomplete feels superior to what iswitchb is doing, the UI is at
> the moment worse (in addition to the other things I've mentioned, if
> you turn on matches on no input C-x b with icomplete-mode will still
> show a redundant "(default <buffer>)" thing).

I filed http://debbugs.gnu.org/cgi/bugreport.cgi?bug=18733 for this.

> So while it looks promising, I don't think it's really a replacement
> for iswitchb yet.

+1

-- 
-- Stephe