cl-functions do not honor common-lisp-indent-function

cl-functions do not honor common-lisp-indent-function

[Thierry Volpiatto]

Hi, some cl-functions are not indented correctly when
`lisp-indent-function' is set to `common-lisp-indent-function'

--8<---------------cut here---------------start------------->8---
(setq lisp-indent-function 'common-lisp-indent-function)

(defun foo ()
  (cl-flet ((test ()
                  (ignore)))
    (test)))

(defun foo ()
  (flet ((test ()
           (ignore)))
    (test)))
--8<---------------cut here---------------end--------------->8---

is this wanted?

-- 
  Thierry
Get my Gnupg key:
gpg --keyserver pgp.mit.edu --recv-keys 59F29997 

Re: cl-functions do not honor common-lisp-indent-function

[Stefan Monnier]

> Hi, some cl-functions are not indented correctly when
> `lisp-indent-function' is set to `common-lisp-indent-function'

> --8<---------------cut here---------------start------------->8---
> (setq lisp-indent-function 'common-lisp-indent-function)

> (defun foo ()
>   (cl-flet ((test ()
>                   (ignore)))
>     (test)))

> (defun foo ()
>   (flet ((test ()
>            (ignore)))
>     (test)))
> --8<---------------cut here---------------end--------------->8---

> is this wanted?

No, but it's not a surprise either, since `cl-flet' is not defined in
common-lisp ;-).


        Stefan

Re: cl-functions do not honor common-lisp-indent-function

[Thierry Volpiatto]

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

>> Hi, some cl-functions are not indented correctly when
>> `lisp-indent-function' is set to `common-lisp-indent-function'
>
>> --8<---------------cut here---------------start------------->8---
>> (setq lisp-indent-function 'common-lisp-indent-function)
>
>> (defun foo ()
>>   (cl-flet ((test ()
>>                   (ignore)))
>>     (test)))
>
>> (defun foo ()
>>   (flet ((test ()
>>            (ignore)))
>>     (test)))
>> --8<---------------cut here---------------end--------------->8---
>
>> is this wanted?
>
> No, but it's not a surprise either, since `cl-flet' is not defined in
> common-lisp ;-).
The prefix 'cl-' mean common lisp no ? ;-) 
so the cl-* family should be added to cl-indent.el, isn't it ?

(let ((l '((cl-flet . flet)
           (cl-labels . flet)
           (cl-macrolet . flet)
           etc...etc...)))
  (dolist (el l)
       [...]

-- 
  Thierry
Get my Gnupg key:
gpg --keyserver pgp.mit.edu --recv-keys 59F29997 

Re: cl-functions do not honor common-lisp-indent-function

[Helmut Eller]

On Wed, Dec 26 2012, Thierry Volpiatto wrote:

> Stefan Monnier <monnier@iro.umontreal.ca> writes:
>
>> No, but it's not a surprise either, since `cl-flet' is not defined in
>> common-lisp ;-).
> The prefix 'cl-' mean common lisp no ? ;-) 
> so the cl-* family should be added to cl-indent.el, isn't it ?

No, because common-lisp-indent-function is for programs written in
actual Common Lisp.  For Emacs Lisp -- and those cl-* symbols are just
Emacs Lisp -- use the normal lisp-indent-function and if you whish to
add rules for cl-lib add it to lisp-indent-function.

Those cl-* symbols have no (predefined) meaning for Common Lisp so
shouldn't be in cl-indent.el; for the same reason there shouldn't be
rules for save-excursion or condition-case.

Helmut

Re: cl-functions do not honor common-lisp-indent-function

[Stephen J. Turnbull]

Helmut Eller writes:

 > Those cl-* symbols have no (predefined) meaning for Common Lisp so
 > shouldn't be in cl-indent.el; for the same reason there shouldn't be
 > rules for save-excursion or condition-case.

That is irrelevant.  If the user thinks of these symbols as equivalent
to "real" Common Lisp standard symbols, then the user has every right
to ask Emacs to treat them as Common Lisp symbols.  That user should
not need to know precisely how that is done, or implement it herself.

In fact, I would argue that the implementer of the "cl-foo is just
like Common Lisp foo" feature probably shouldn't need to know, either.
Instead these should be a minor mode which, when enabled,
automagically looks up every cl-* symbol without the prefix for the
purposes of source code formatting (including syntax highlighting).

I have no idea how to implement that, of course.<wink/>  And I do
agree that purists shouldn't have to put up with that, so it needs to
be modal.

Re: cl-functions do not honor common-lisp-indent-function

[Helmut Eller]

On Wed, Dec 26 2012, Stephen J. Turnbull wrote:

> Helmut Eller writes:
>
>  > Those cl-* symbols have no (predefined) meaning for Common Lisp so
>  > shouldn't be in cl-indent.el; for the same reason there shouldn't be
>  > rules for save-excursion or condition-case.
>
> That is irrelevant.

Why?  Why should common-lisp-indent-function, by default, implement
something other than the Common Lisp conventions?  Should
scheme-indent-function also support Emacs Lisp macros like cl-flet?

> If the user thinks of these symbols as equivalent
> to "real" Common Lisp standard symbols, then the user has every right
> to ask Emacs to treat them as Common Lisp symbols.

An Emacs Lisp user or for that matter a Scheme user has not "every
right" to dictate the indentation rules for Common Lisp users.

Helmut

Re: cl-functions do not honor common-lisp-indent-function

[Stephen J. Turnbull]

Helmut Eller writes:
 > On Wed, Dec 26 2012, Stephen J. Turnbull wrote:
 > 
 > > Helmut Eller writes:
 > >
 > >  > Those cl-* symbols have no (predefined) meaning for Common Lisp so
 > >  > shouldn't be in cl-indent.el; for the same reason there shouldn't be
 > >  > rules for save-excursion or condition-case.
 > >
 > > That is irrelevant.
 > 
 > Why?  Why should common-lisp-indent-function, by default, implement
 > something other than the Common Lisp conventions?

It shouldn't.  And certainly not by default (perhaps I should have
used the word "optional" instead of "modal").  But you've got the
wrong question.  The right one is the inverse:

    Why should anything other than `common-lisp-indent-function'
    implement the Common Lisp conventions?

 > > If the user thinks of these symbols as equivalent
 > > to "real" Common Lisp standard symbols, then the user has every right
 > > to ask Emacs to treat them as Common Lisp symbols.
 > 
 > An Emacs Lisp user or for that matter a Scheme user has not "every
 > right" to dictate the indentation rules for Common Lisp users.

Of course users of other Lisp dialects should not dictate to Common
Lisp users.  But I was talking about *Emacs* users.  I don't have any
trouble imagining that some Emacs users who program in both Common
Lisp and cl.el would like to see cl-* expressions follow the
corresponding Common Lisp conventions.  After all, the OP is evidently
an example.

Steve

Re: cl-functions do not honor common-lisp-indent-function

[Helmut Eller]

On Wed, Dec 26 2012, Stephen J. Turnbull wrote:

>     Why should anything other than `common-lisp-indent-function'
>     implement the Common Lisp conventions?

E.g. to avoid conflicts: some other language might have similar but not
quite the same conventions and to avoid the risk of conflicts you
re-implement some of the conventions.

>  > > If the user thinks of these symbols as equivalent
>  > > to "real" Common Lisp standard symbols, then the user has every right
>  > > to ask Emacs to treat them as Common Lisp symbols.
>  > 
>  > An Emacs Lisp user or for that matter a Scheme user has not "every
>  > right" to dictate the indentation rules for Common Lisp users.
>
> Of course users of other Lisp dialects should not dictate to Common
> Lisp users.  But I was talking about *Emacs* users.  I don't have any
> trouble imagining that some Emacs users who program in both Common
> Lisp and cl.el would like to see cl-* expressions follow the
> corresponding Common Lisp conventions.  After all, the OP is evidently
> an example.

When I'm programming Emacs Lisp I use emacs-lisp-mode and I assume that
the default indentation function, i.e. lisp-indentation-function, will
handle all relevant macros.  I fail to see why using
common-lisp-indent-function in emacs-lisp-mode is such a bright idea.

Helmut

Re: cl-functions do not honor common-lisp-indent-function

[Thierry Volpiatto]

Helmut Eller <eller.helmut@gmail.com> writes:

> On Wed, Dec 26 2012, Stephen J. Turnbull wrote:
>
>>     Why should anything other than `common-lisp-indent-function'
>>     implement the Common Lisp conventions?
>
> E.g. to avoid conflicts: some other language might have similar but not
> quite the same conventions and to avoid the risk of conflicts you
> re-implement some of the conventions.
>
>>  > > If the user thinks of these symbols as equivalent
>>  > > to "real" Common Lisp standard symbols, then the user has every right
>>  > > to ask Emacs to treat them as Common Lisp symbols.
>>  > 
>>  > An Emacs Lisp user or for that matter a Scheme user has not "every
>>  > right" to dictate the indentation rules for Common Lisp users.
>>
>> Of course users of other Lisp dialects should not dictate to Common
>> Lisp users.  But I was talking about *Emacs* users.  I don't have any
>> trouble imagining that some Emacs users who program in both Common
>> Lisp and cl.el would like to see cl-* expressions follow the
>> corresponding Common Lisp conventions.  After all, the OP is evidently
>> an example.
>
> When I'm programming Emacs Lisp I use emacs-lisp-mode and I assume that
> the default indentation function, i.e. lisp-indentation-function, will
> handle all relevant macros.
It is not handling all these macros/functions, so what do we do ?

> I fail to see why using common-lisp-indent-function in emacs-lisp-mode
> is such a bright idea.
Because emacs-lisp-mode we may use common-lisp style functions (flet,
labels etc...).
So when you indent e.g a flet clause with emacs-lisp style it is indented
badly.

> Helmut
>
>
>

-- 
  Thierry
Get my Gnupg key:
gpg --keyserver pgp.mit.edu --recv-keys 59F29997 

Re: cl-functions do not honor common-lisp-indent-function

[Helmut Eller]

On Wed, Dec 26 2012, Thierry Volpiatto wrote:

>> When I'm programming Emacs Lisp I use emacs-lisp-mode and I assume that
>> the default indentation function, i.e. lisp-indentation-function, will
>> handle all relevant macros.
> It is not handling all these macros/functions, so what do we do ?

Improve lisp-indentation-function.

>> I fail to see why using common-lisp-indent-function in emacs-lisp-mode
>> is such a bright idea.
>
> Because emacs-lisp-mode we may use common-lisp style functions (flet,
> labels etc...).
> So when you indent e.g a flet clause with emacs-lisp style it is indented
> badly.

But common-lisp-indent-function will indent "if" like:

(if x
    a
    b)

while the Emacs Lisp convention is:

(if x
    a
  b)

Helmut

Re: cl-functions do not honor common-lisp-indent-function

[Thierry Volpiatto]

Helmut Eller <eller.helmut@gmail.com> writes:

> On Wed, Dec 26 2012, Thierry Volpiatto wrote:
>
>>> When I'm programming Emacs Lisp I use emacs-lisp-mode and I assume that
>>> the default indentation function, i.e. lisp-indentation-function, will
>>> handle all relevant macros.
>> It is not handling all these macros/functions, so what do we do ?
>
> Improve lisp-indentation-function.
>
>>> I fail to see why using common-lisp-indent-function in emacs-lisp-mode
>>> is such a bright idea.
>>
>> Because emacs-lisp-mode we may use common-lisp style functions (flet,
>> labels etc...).
>> So when you indent e.g a flet clause with emacs-lisp style it is indented
>> badly.
>
> But common-lisp-indent-function will indent "if" like:
>
> (if x
>     a
>     b)
>
> while the Emacs Lisp convention is:
>
> (if x
>     a
>   b)

Not a problem for me, one or the other is not bad, just a different
style. (I prefer the common-lisp style for `if')

I understand your point of view, the problem is to indent and fontlock
correctly all the common-lisp style functions used in emacs-lisp-mode,
in one way or the other.
I agree that the best would be to handle this for emacs-lisp-mode
without using an indentation destinated for another mode (lisp).
Unfortunately it is missing, not implemented, so we use the "less bad"
solution, using common-lisp indentation style for all with some tweaking
for the cl-* ones that are not supported in both style.

-- 
  Thierry
Get my Gnupg key:
gpg --keyserver pgp.mit.edu --recv-keys 59F29997 

Re: cl-functions do not honor common-lisp-indent-function

[Stephen J. Turnbull]

Helmut Eller writes:

 > I fail to see why using common-lisp-indent-function in
 > emacs-lisp-mode is such a bright idea.

"You fail" is precisely correct.  Nothing wrong with that, it's not
your job to think like anybody but you.  That's why it should be an
option ("minor mode", eg, cl-shadows-common-lisp-mode).

Others who habitually use cl want *their* Elisp to "look and feel" as
much like Common Lisp as possible, *including any local mods they may
have made to "standard" Common Lisp formatting conventions*.  By the
same token, it's nobody else's job to think like you.  That's why it
should be an option.  I tend to agree with you that it should default
"off", but that's up to the maintainers, and it doesn't matter much.

It is Emacs's job to adapt as well as possible to *both* opinions.
That's certainly possible here.  Add DRY, and the bright idea follows.

Re: cl-functions do not honor common-lisp-indent-function

[Helmut Eller]

On Thu, Dec 27 2012, Stephen J. Turnbull wrote:

> Helmut Eller writes:
>
>  > I fail to see why using common-lisp-indent-function in
>  > emacs-lisp-mode is such a bright idea.
>
> "You fail" is precisely correct.  Nothing wrong with that, it's not
> your job to think like anybody but you.  That's why it should be an
> option ("minor mode", eg, cl-shadows-common-lisp-mode).
>
> Others who habitually use cl want *their* Elisp to "look and feel" as
> much like Common Lisp as possible, *including any local mods they may
> have made to "standard" Common Lisp formatting conventions*.  By the
> same token, it's nobody else's job to think like you.  That's why it
> should be an option.  I tend to agree with you that it should default
> "off", but that's up to the maintainers, and it doesn't matter much.
>
> It is Emacs's job to adapt as well as possible to *both* opinions.
> That's certainly possible here.  Add DRY, and the bright idea follows.

Apparently we are in wild agreement: common-lisp-indent-function should
not handle cl-* symbols by default.

Helmut

Re: cl-functions do not honor common-lisp-indent-function

[Thierry Volpiatto]

Helmut Eller <eller.helmut@gmail.com> writes:

> On Wed, Dec 26 2012, Stephen J. Turnbull wrote:
>
>> Helmut Eller writes:
>>
>>  > Those cl-* symbols have no (predefined) meaning for Common Lisp so
>>  > shouldn't be in cl-indent.el; for the same reason there shouldn't be
>>  > rules for save-excursion or condition-case.
>>
>> That is irrelevant.
>
> Why?  Why should common-lisp-indent-function, by default, implement
> something other than the Common Lisp conventions?  Should
> scheme-indent-function also support Emacs Lisp macros like cl-flet?
Are you going to use cl-* symbols in your scheme or common-lisp files ?
I don't think so, so what will change in the indentation of your
common-lisp files ? Nothing.

>> If the user thinks of these symbols as equivalent
>> to "real" Common Lisp standard symbols, then the user has every right
>> to ask Emacs to treat them as Common Lisp symbols.
>
> An Emacs Lisp user or for that matter a Scheme user has not "every
> right" to dictate the indentation rules for Common Lisp users.
>
> Helmut
>
>

-- 
  Thierry
Get my Gnupg key:
gpg --keyserver pgp.mit.edu --recv-keys 59F29997 

Re: cl-functions do not honor common-lisp-indent-function

[Helmut Eller]

On Wed, Dec 26 2012, Thierry Volpiatto wrote:

> Helmut Eller <eller.helmut@gmail.com> writes:
>
>> On Wed, Dec 26 2012, Stephen J. Turnbull wrote:
>>
>>> Helmut Eller writes:
>>>
>>>  > Those cl-* symbols have no (predefined) meaning for Common Lisp so
>>>  > shouldn't be in cl-indent.el; for the same reason there shouldn't be
>>>  > rules for save-excursion or condition-case.
>>>
>>> That is irrelevant.
>>
>> Why?  Why should common-lisp-indent-function, by default, implement
>> something other than the Common Lisp conventions?  Should
>> scheme-indent-function also support Emacs Lisp macros like cl-flet?
> Are you going to use cl-* symbols in your scheme or common-lisp files ?

No probably not.

> I don't think so, so what will change in the indentation of your
> common-lisp files ? Nothing.

See, since those cl-* symbols are not often used in Common Lisp there is
no justification for your rules.

Helmut

Re: cl-functions do not honor common-lisp-indent-function

[Didier Verna]

Stephen J. Turnbull wrote:

> Helmut Eller writes:
>
>  > Those cl-* symbols have no (predefined) meaning for Common Lisp so
>  > shouldn't be in cl-indent.el; for the same reason there shouldn't be
>  > rules for save-excursion or condition-case.
>
> That is irrelevant.  If the user thinks of these symbols as equivalent
> to "real" Common Lisp standard symbols, then the user has every right
> to ask Emacs to treat them as Common Lisp symbols.  That user should
> not need to know precisely how that is done, or implement it herself.

  cl-indent is meant for providing indentation in *real* Common Lisp
  source files. The cl-* functions will not appear in Common Lisp code
  so cl-indent shouldn't know about them.

  The cl-* functions may however appear in Emacs Lisp code, so indenting
  them properly (that is, in a CL style because these functions provide
  CL functionality to elisp) is a job for (lisp-indent-function). It is
  currently missing.

  Note that I happen to *not* use (lisp-indent-function) at all, but
  always (common-lisp-indent-function), even in Emacs Lisp files. In
  such a situation, I still consider that I'm doing something
  non-standard, and so I wouldn't complain because the cl-* functions
  are not indented properly by cl-indent. I think it's my job, as an
  end-user who's doing something weird, to support them by hacking my
  configuration.

  Of course, if there were also a shortcut such as
  (cl-indent-add-cl-lib-support), that would be nice.

-- 
ELS 2013, June 3/4, Madrid, Spain:  http://els2013.european-lisp-symposium.org

Scientific site:   http://www.lrde.epita.fr/~didier
Music (Jazz) site: http://www.didierverna.com

Re: cl-functions do not honor common-lisp-indent-function

[Stefan Monnier]

> Instead these should be a minor mode which, when enabled,
> automagically looks up every cl-* symbol without the prefix for the
> purposes of source code formatting (including syntax highlighting).

No need for a minor mode.  emacs-lisp-mode's indentation code should do that.


        Stefan