electric-pair-mode

electric-pair-mode

[Stefan Monnier]

AFAIK, most of the issues pointed out in the previous electric-pair-mode
have been resolved:
E.g. C-j and (newline) now insert a plain newline.  Some indentation
functions are marked as "unable to reindent".

I'm thinking of enabling electric-pair-mode in Emacs-24.4, and I think
the best way to do that is to actually enable it and then deal with the
complaints.  But that only works if the complaints are
sufficiently constructive.

So, please try enabling it and report any annoyance you notice.
If you just plain don't like the feature, then just disable it.


        Stefan

Re: electric-pair-mode

[Xue Fuqiao]

On Sun, Nov 17, 2013 at 7:27 AM, Stefan Monnier
<monnier@iro.umontreal.ca> wrote:
>
> AFAIK, most of the issues pointed out in the previous electric-pair-mode
> have been resolved:
> E.g. C-j and (newline) now insert a plain newline.  Some indentation
> functions are marked as "unable to reindent".
>
> I'm thinking of enabling electric-pair-mode in Emacs-24.4, and I think
> the best way to do that is to actually enable it and then deal with the
> complaints.  But that only works if the complaints are
> sufficiently constructive.
>
> So, please try enabling it and report any annoyance you notice.
> If you just plain don't like the feature, then just disable it.

You mean electric-indent-mode, I assume?  But anyway, both of them works
well for me.

-- 
http://www.gnu.org/software/emacs/

Re: electric-pair-mode

[Jorgen Schaefer]

On Sat, 16 Nov 2013 18:27:30 -0500
Stefan Monnier <monnier@iro.umontreal.ca> wrote:

> I'm thinking of enabling electric-pair-mode in Emacs-24.4, and I think
> the best way to do that is to actually enable it and then deal with
> the complaints.  But that only works if the complaints are
> sufficiently constructive.
> 
> So, please try enabling it and report any annoyance you notice.
> If you just plain don't like the feature, then just disable it.

If this is about electric-PAIR-mode:

Works nicely, except the default `electric-pair-inhibit-predicate'
makes it so that {{ results in "{{_|_}", even though I'd expect
"{{_|_}}".


If this is about electric-INDENT-mode:

The global mode does not work well with ido-mode. When opening a new
buffer or file, you have to use C-j to confirm that you really want to
open a non-existing buffer/file. electric-indent-mode inserts the ^J
verbatim instead of this.

Reproduction:

Enable ido-mode for buffers

C-x b t C-j should create a new buffer named "t". With
electric-indent-mode, it does not confirm, but adds ^J to the buffer
name. RET then opens a buffer named "t^J". (This might also be a
misfeature in ido, not sure).

Regards,
Jorgen

electric-indent-mode (was: electric-pair-mode)

[Stefan Monnier]

> AFAIK, most of the issues pointed out in the previous electric-pair-mode
> have been resolved:
> E.g. C-j and (newline) now insert a plain newline.  Some indentation
> functions are marked as "unable to reindent".

In case some of you aren't confused enough yet: I really meant
"electric-indent-mode", not electric-pair-mode!

Sorry,


        Stefan

Re: electric-pair-mode

[Stefan Monnier]

> The global mode does not work well with ido-mode. When opening a new
> buffer or file, you have to use C-j to confirm that you really want to
> open a non-existing buffer/file. electric-indent-mode inserts the ^J
> verbatim instead of this.

Indeed, thanks.  Can you make it a bug-report, so I don't forget?


        Stefan

Re: electric-pair-mode

[Xue Fuqiao]

On Mon, Nov 18, 2013 at 2:30 AM, Jorgen Schaefer <forcer@forcix.cx> wrote:
>> I'm thinking of enabling electric-pair-mode in Emacs-24.4, and I think
>> the best way to do that is to actually enable it and then deal with
>> the complaints.  But that only works if the complaints are
>> sufficiently constructive.
>>
>> So, please try enabling it and report any annoyance you notice.
>> If you just plain don't like the feature, then just disable it.
>
> If this is about electric-PAIR-mode:
>
> Works nicely, except the default `electric-pair-inhibit-predicate'
> makes it so that {{ results in "{{_|_}", even though I'd expect
> "{{_|_}}".

Another problem:

  In CPerl Mode (and maybe in other modes as well), for example, I'd
  like to type:

    print 'Hello World!';

  When I press the first apostrophe character, Emacs only inserts one
  apostrophe; when I press the second apostrophe, Emacs inserts two
  apostrophes, which is very annoying.  And I also found two other bugs
  about electric-pair-mode in Debbugs.

-- 
http://www.gnu.org/software/emacs/

electric-indent-mode (was Re: electric-pair-mode)

[Stephen Berman]

On Sat, 16 Nov 2013 18:27:30 -0500 Stefan Monnier <monnier@iro.umontreal.ca> wrote:

> AFAIK, most of the issues pointed out in the previous electric-pair-mode
> have been resolved:
> E.g. C-j and (newline) now insert a plain newline.  Some indentation
> functions are marked as "unable to reindent".
>
> I'm thinking of enabling electric-pair-mode in Emacs-24.4, and I think
> the best way to do that is to actually enable it and then deal with the
> complaints.  But that only works if the complaints are
> sufficiently constructive.
>
> So, please try enabling it and report any annoyance you notice.

Will there be a new key binding for eval-print-last-sexp in *scratch*?

Steve Berman

Re: electric-indent-mode (was Re: electric-pair-mode)

[Stefan Monnier]

> Will there be a new key binding for eval-print-last-sexp in *scratch*?

There's C-j (if you use a recent enough version from trunk).


        Stefan

Re: electric-indent-mode (was Re: electric-pair-mode)

[Stephen Berman]

On Tue, 19 Nov 2013 08:41:17 -0500 Stefan Monnier <monnier@iro.umontreal.ca> wrote:

>> Will there be a new key binding for eval-print-last-sexp in *scratch*?
>
> There's C-j (if you use a recent enough version from trunk).

Indeed, my build was 20 revisions too old; sorry for the noise.

Steve Berman

Re: electric-pair-mode

[Tassilo Horn]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

> So, please try enabling it and report any annoyance you notice.  If
> you just plain don't like the feature, then just disable it.

No annoyance so far but I'm wondering why both `electric-pair-mode' and
`electric-indent-mode' are *global* minor modes.

For example, in all kinds of lisp modes I'm using paredit which provides
both features, too, in a slightly different form.  So there I wouldn't
want to activate these modes (although it seems having them activated
doesn't do any harm).

Bye,
Tassilo

Re: electric-pair-mode

[Stefan Monnier]

> No annoyance so far but I'm wondering why both `electric-pair-mode' and
> `electric-indent-mode' are *global* minor modes.

Because in my experience, most people who like those features, like
them everywhere.

> For example, in all kinds of lisp modes I'm using paredit which
> provides both features, too, in a slightly different form.  So there
> I wouldn't want to activate these modes (although it seems having them
> activated doesn't do any harm).

Looks like there's no problem, then.


        Stefan

Re: electric-pair-mode

[Alexander Baier]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

>> No annoyance so far but I'm wondering why both `electric-pair-mode' and
>> `electric-indent-mode' are *global* minor modes.
>
> Because in my experience, most people who like those features, like
> them everywhere.
>
>> For example, in all kinds of lisp modes I'm using paredit which
>> provides both features, too, in a slightly different form.  So there
>> I wouldn't want to activate these modes (although it seems having them
>> activated doesn't do any harm).
>
> Looks like there's no problem, then.
>

In my experience electric-indent-mode screws up indentation in
python-mode buffers:

----------------------------------------------------------------
def foo(x):
    print (x)

def bar ():|
----------------------------------------------------------------

Put this content into a buffer and turn on python-mode and
electric-indent-mode.  Put the cursor where the "|" is and press RET.
On my machine the "def bar():" gets reindented so it lines up with the
preceding "print(x):" line.  This is usually not what you want.

>
>         Stefan

Regards,
  Alex

Re: electric-pair-mode

[Tassilo Horn]

Alexander Baier <lexi.baier@gmail.com> writes:

>>> For example, in all kinds of lisp modes I'm using paredit which
>>> provides both features, too, in a slightly different form.  So there
>>> I wouldn't want to activate these modes (although it seems having them
>>> activated doesn't do any harm).
>>
>> Looks like there's no problem, then.
>
> In my experience electric-indent-mode screws up indentation in
> python-mode buffers:

That's the purpose of:

,----[ C-h v electric-indent-inhibit RET ]
| electric-indent-inhibit is a variable defined in `electric.el'.
| Its value is nil
| 
|   Automatically becomes buffer-local when set.
| 
| Documentation:
| If non-nil, reindentation is not appropriate for this buffer.
| This should be set by major modes such as `python-mode' since
| Python does not lend itself to fully automatic indentation.
`----

In the current trunk, `python-mode' sets that to t.

Bye,
Tassilo

Re: electric-pair-mode

[Alexander Baier]

Tassilo Horn <tsdh@gnu.org> writes:

> Alexander Baier <lexi.baier@gmail.com> writes:
>
>>>> For example, in all kinds of lisp modes I'm using paredit which
>>>> provides both features, too, in a slightly different form.  So there
>>>> I wouldn't want to activate these modes (although it seems having them
>>>> activated doesn't do any harm).
>>>
>>> Looks like there's no problem, then.
>>
>> In my experience electric-indent-mode screws up indentation in
>> python-mode buffers:
>
> That's the purpose of:
>
> ,----[ C-h v electric-indent-inhibit RET ]
> | electric-indent-inhibit is a variable defined in `electric.el'.
> | Its value is nil
> | 
> |   Automatically becomes buffer-local when set.
> | 
> | Documentation:
> | If non-nil, reindentation is not appropriate for this buffer.
> | This should be set by major modes such as `python-mode' since
> | Python does not lend itself to fully automatic indentation.
> `----
>
> In the current trunk, `python-mode' sets that to t.
>
> Bye,
> Tassilo

Thank you, I did not know about this. I am still running 24.3, which
does not offer this. I will add this to my config as soon as 24.4 is
released.

Regards,
  Alex

Re: electric-pair-mode

[Tassilo Horn]

Alexander Baier <lexi.baier@gmail.com> writes:

Hi Alex,

>>> In my experience electric-indent-mode screws up indentation in
>>> python-mode buffers:
>>
>> That's the purpose of:
>>
>> ,----[ C-h v electric-indent-inhibit RET ]
>> | electric-indent-inhibit is a variable defined in `electric.el'.
>> | Its value is nil
>> | 
>> |   Automatically becomes buffer-local when set.
>> | 
>> | Documentation:
>> | If non-nil, reindentation is not appropriate for this buffer.
>> | This should be set by major modes such as `python-mode' since
>> | Python does not lend itself to fully automatic indentation.
>> `----
>>
>> In the current trunk, `python-mode' sets that to t.
>
> Thank you, I did not know about this. I am still running 24.3, which
> does not offer this. I will add this to my config as soon as 24.4 is
> released.

You don't need to configure anything.  With 24.4, it'll work correct out
of the box.

Bye,
Tassilo

Re: electric-pair-mode

[Stefan Monnier]

> I am still running 24.3

The 24.3 version of electric-indent-mode is definitely not fit for being
enabled by default.  If you want to help with it, use the latest trunk
code: that code is meant to be "good enough to be enabled by default".


        Stefan

Re: electric-pair-mode

[Tassilo Horn]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

>> For example, in all kinds of lisp modes I'm using paredit which
>> provides both features, too, in a slightly different form.  So there
>> I wouldn't want to activate these modes (although it seems having
>> them activated doesn't do any harm).
>
> Looks like there's no problem, then.

Ok, now I've found a mode where it's a problem.  AUCTeX has its own
TeX-ified brace pairing feature which for example knows that \left( has
to be closed by \right(, \{ by \}, etc.  With `electric-pair-mode'
enabled, when typing a { I get two closing braces.

Right now, I have "disabled" `electric-pair-mode' for auctex using:

  ;; AUCTeX brace pairing feature doesn't play nice with `electric-pair-mode'
  ;; which is a global minor mode as of emacs 24.4.
  (when (and (boundp 'electric-pair-mode)
             (boundp 'electric-pair-inhibit-predicate))
    (set (make-local-variable 'electric-pair-inhibit-predicate)
         (lambda (char) t))))

Is that the way it's supposed to be done?

I somehow have the feeling that `electric-pair-mode' should only do its
magic if the opening paren/brace/bracket is inserted with
`self-insert-command'.  For modes that bind ({[ to special commands
(like paredit and auctex), chances are high that you get some
interference with whatever special logic the mode does.  On the other
hand, there doesn't seem to be a problem with paredit, so maybe we're
just doing something wrong in auctex...
         
Bye,
Tassilo

Re: electric-pair-mode

[Stefan Monnier]

> Ok, now I've found a mode where it's a problem.  AUCTeX has its own
> TeX-ified brace pairing feature which for example knows that \left( has
> to be closed by \right(, \{ by \}, etc.  With `electric-pair-mode'
> enabled, when typing a { I get two closing braces.

Yes, that is a problem.  AUCTeX's pairing should be made aware of
electric-pair-mode (including being enabled/disabled by
calling/setting electric-pair-mode as well).

> Right now, I have "disabled" `electric-pair-mode' for auctex using:

>   ;; AUCTeX brace pairing feature doesn't play nice with `electric-pair-mode'
>   ;; which is a global minor mode as of emacs 24.4.
>   (when (and (boundp 'electric-pair-mode)
>              (boundp 'electric-pair-inhibit-predicate))
>     (set (make-local-variable 'electric-pair-inhibit-predicate)
>          (lambda (char) t))))

A simpler way to disable it is to set electric-pair-mode to nil
(buffer-locally).  No need to check boundp of anything.

> I somehow have the feeling that `electric-pair-mode' should only do its
> magic if the opening paren/brace/bracket is inserted with
> `self-insert-command'.

What happens is that AUCTeX's pairing commands call self-insert-command.


        Stefan

Re: electric-pair-mode

[Tassilo Horn]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

>> Ok, now I've found a mode where it's a problem.  AUCTeX has its own
>> TeX-ified brace pairing feature which for example knows that \left(
>> has to be closed by \right(, \{ by \}, etc.  With
>> `electric-pair-mode' enabled, when typing a { I get two closing
>> braces.
>
> Yes, that is a problem.  AUCTeX's pairing should be made aware of
> electric-pair-mode (including being enabled/disabled by
> calling/setting electric-pair-mode as well).
>
>> Right now, I have "disabled" `electric-pair-mode' for auctex using:
>
>>   ;; AUCTeX brace pairing feature doesn't play nice with `electric-pair-mode'
>>   ;; which is a global minor mode as of emacs 24.4.
>>   (when (and (boundp 'electric-pair-mode)
>>              (boundp 'electric-pair-inhibit-predicate))
>>     (set (make-local-variable 'electric-pair-inhibit-predicate)
>>          (lambda (char) t))))
>
> A simpler way to disable it is to set electric-pair-mode to nil
> (buffer-locally).  No need to check boundp of anything.

Ah, that's indeed much better.

Thanks,
Tassilo