mark-sexp again

mark-sexp again

[Andreas Röhler]

Hi,

coming across an implemention `clojure-mark-string' - useful for all 
people not having an appropriate extension.

Given what the hell is all implemented in Emacs, why can't commands 
around sexp take the string if within?

Cheers,

Andreas

Re: mark-sexp again

[Stefan Monnier]

> Given what the hell is all implemented in Emacs, why can't commands around
> sexp take the string if within?

I think you're looking for expand-region.el


        Stefan

Re: mark-sexp again

[Andreas Röhler]

Am 03.06.2015 um 17:24 schrieb Stefan Monnier:
>> Given what the hell is all implemented in Emacs, why can't commands around
>> sexp take the string if within?
> I think you're looking for expand-region.el
>
>
>          Stefan
>

AFAIU a string  inside a list is a sexp-expression and thus M-x 
mark-sexp RET should work accordingly.

Re: mark-sexp again

[Stefan Monnier]

> AFAIU a string  inside a list is a sexp-expression and thus M-x 
> mark-sexp RET should work accordingly.

It does, AFAICT.


        Stefan

Re: mark-sexp again

[Andreas Röhler]

Am 08.06.2015 um 15:09 schrieb Stefan Monnier:
>> AFAIU a string  inside a list is a sexp-expression and thus M-x
>> mark-sexp RET should work accordingly.
> It does, AFAICT.
>
>

When inside a string, mark-sexp should mark the string. IMO doesn't make 
sense to consider the contents then.

Example, cursor at beginning of docstring at "Just"

(defun foo1 ()
   "Just foo")

M-x mark-sexp RET marks word "Just", not the string.

Respectivly forward-sexp should reach the end of string, if called inside.

Thanks,

Andreas

Re: mark-sexp again

[Artur Malabarba]

2015-06-08 15:24 GMT+01:00 Andreas Röhler <andreas.roehler@online.de>:
> When inside a string, mark-sexp should mark the string. IMO doesn't make sense to consider the contents then.
>
> Example, cursor at beginning of docstring at "Just"
>
> (defun foo1 ()
>   "Just foo")
>
> M-x mark-sexp RET marks word "Just", not the string.
>
> Respectivly forward-sexp should reach the end of string, if called inside.

That would make it redundant with `up-list'.

As for the general suggestion, I personally like the current behavior.
For instance, in the following docstring, it's nice that forward-sexp
and kill-sexp work on `foo-bar' inside the string.

(defun ok ()
  "something `foo-bar' something")

Re: mark-sexp again

[Andreas Röhler]

Am 08.06.2015 um 18:09 schrieb Artur Malabarba:
> 2015-06-08 15:24 GMT+01:00 Andreas Röhler <andreas.roehler@online.de>:
>> When inside a string, mark-sexp should mark the string. IMO doesn't make sense to consider the contents then.
>>
>> Example, cursor at beginning of docstring at "Just"
>>
>> (defun foo1 ()
>>    "Just foo")
>>
>> M-x mark-sexp RET marks word "Just", not the string.
>>
>> Respectivly forward-sexp should reach the end of string, if called inside.
> That would make it redundant with `up-list'.

Don't think so:

(list "asdfsdf adfsadf  " "asdfsd asd asdf " "asdfasf asdfsad ")

When started a "list",  forward-sexp travels string by string, which is 
ok and also means, it recognises a string as sexp already.

Accordingly from inside a string, it should go to the end first.

>
> As for the general suggestion, I personally like the current behavior.
> For instance, in the following docstring, it's nice that forward-sexp
> and kill-sexp work on `foo-bar' inside the string.
>
> (defun ok ()
>    "something `foo-bar' something")
>

Because a kill-symbol-at-point seems missing.
No reason to maintain a quirk :)

Re: mark-sexp again

[Artur Malabarba]

2015-06-08 19:55 GMT+01:00 Andreas Röhler <andreas.roehler@online.de>:
>
> Am 08.06.2015 um 18:09 schrieb Artur Malabarba:
>>
>> 2015-06-08 15:24 GMT+01:00 Andreas Röhler <andreas.roehler@online.de>:
>>>
>>> When inside a string, mark-sexp should mark the string. IMO doesn't make
>>> sense to consider the contents then.
>>>
>>> Example, cursor at beginning of docstring at "Just"
>>>
>>> (defun foo1 ()
>>>    "Just foo")
>>>
>>> M-x mark-sexp RET marks word "Just", not the string.
>>>
>>> Respectivly forward-sexp should reach the end of string, if called
>>> inside.
>>
>> That would make it redundant with `up-list'.
>
>
> Don't think so:

I didn't mean the two commands are completely identical. I meant that
the action of moving out of a string while inside a string is already
achieved by the `up-list' command (which is also generally useful but
has no binding), so changing `forward-sexp' to do that as well would
be redundant. Specially since you'd be removing `forward-sexp's
ability to move inside a string, which is useful.

> (list "asdfsdf adfsadf  " "asdfsd asd asdf " "asdfasf asdfsad ")
>
> When started a "list",  forward-sexp travels string by string, which is ok
> and also means, it recognises a string as sexp already.
>
> Accordingly from inside a string, it should go to the end first.

You seem to think this would make forward-sexp more consistent, but it
would make it less consistent. Currently, a pair of parentheses
delineates a sexp in the same way that a pair of double quotes does.
If you place point at the "a" on the following two examples,
`forward-sexp' will do the exact same thing in both cases. That's
consistent.

    (abc def)
    "abc def"

If you want to move out of a string, just use the same command you
would use on parentheses (`up-list').
If you want to mark a whole string while inside it, just use the same
procedure you would use on parentheses (`backward-up-list' with
`mark-sexp').

Re: mark-sexp again

[Andreas Röhler]

Am 08.06.2015 um 21:37 schrieb Artur Malabarba:
> 2015-06-08 19:55 GMT+01:00 Andreas Röhler <andreas.roehler@online.de>:
>> Am 08.06.2015 um 18:09 schrieb Artur Malabarba:
>>> 2015-06-08 15:24 GMT+01:00 Andreas Röhler <andreas.roehler@online.de>:
>>>> When inside a string, mark-sexp should mark the string. IMO doesn't make
>>>> sense to consider the contents then.
>>>>
>>>> Example, cursor at beginning of docstring at "Just"
>>>>
>>>> (defun foo1 ()
>>>>     "Just foo")
>>>>
>>>> M-x mark-sexp RET marks word "Just", not the string.
>>>>
>>>> Respectivly forward-sexp should reach the end of string, if called
>>>> inside.
>>> That would make it redundant with `up-list'.
>>
>> Don't think so:
> I didn't mean the two commands are completely identical. I meant that
> the action of moving out of a string while inside a string is already
> achieved by the `up-list' command (which is also generally useful but
> has no binding), so changing `forward-sexp' to do that as well would
> be redundant. Specially since you'd be removing `forward-sexp's
> ability to move inside a string, which is useful.
>
>> (list "asdfsdf adfsadf  " "asdfsd asd asdf " "asdfasf asdfsad ")
>>
>> When started a "list",  forward-sexp travels string by string, which is ok
>> and also means, it recognises a string as sexp already.
>>
>> Accordingly from inside a string, it should go to the end first.
> You seem to think this would make forward-sexp more consistent, but it
> would make it less consistent. Currently, a pair of parentheses
> delineates a sexp in the same way that a pair of double quotes does.
> If you place point at the "a" on the following two examples,
> `forward-sexp' will do the exact same thing in both cases. That's
> consistent.
>
>      (abc def)
>      "abc def"

Expect forward-sexp to behave according to syntax of the language.
AFAIU in Emacs Lisp an sexp is either a list --empty or not-- or an 
element of a list.

Thus

(abc def)

has at least three sexp.

While a string seen from Emas Lisp is just a single one.


[ ... ]

Re: mark-sexp again

[Stefan Monnier]

> When inside a string, mark-sexp should mark the string.

Makes no sense.  C-h f mark-sexp RET:

     "Set mark ARG sexps from point.
   The place mark goes is the same place \\[forward-sexp] would
   move to with the same argument.

Again, you want expand-region.el (which should ideally make it into
Emacs at some point.  If someone could help Magnar with it, I'm sure
he'd appreciate).


        Stefan