bug#17955: 24.3.92; octave.el: indentation following ... or \

bug#17955: 24.3.92; octave.el: indentation following ... or \

[Leo Liu]

As reported on 2013-07-03 in
http://article.gmane.org/gmane.comp.gnu.octave.maintainers/33796

In octave mode:

a = \
b

b is aligned to a. Previously there was an offset per
octave-continuation-offset.

The octave smie lexer may or may not return the continuation token (...
or \), any idea why it is made to behave like this?
octave-continuation-offset is also unused. Comments?

Leo

bug#17955: 24.3.92; octave.el: indentation following ... or \

[Stefan Monnier]

> As reported on 2013-07-03 in
> http://article.gmane.org/gmane.comp.gnu.octave.maintainers/33796

> In octave mode:

> a = \
> b

> b is aligned to a. Previously there was an offset per
> octave-continuation-offset.

> The octave smie lexer may or may not return the continuation token (...
> or \), any idea why it is made to behave like this?
> octave-continuation-offset is also unused. Comments?

Same as for sh-script's SMIE support, the new indentation code performs
indentation of continued lines as if the "backslash newline" was
a normal newline except it doesn't contain an implicit semi-colon.
So you should get indentation like:

   a = b + a * \
           c

So you should be able to control indentation of "b" above by tweaking
the (:after . "=") or (:before . "=") rule.

That indentation style is incompatible with octave-continuation-offset
(which basically assumes that continued lines are just a single line
wrapped).  But, just like we did in sh-script.el, we could/should
probably add an indentation rule to SMIE to make sure that continued
lines are indented at least as much as octave-continuation-offset.


        Stefan

bug#17955: 24.3.92; octave.el: indentation following ... or \

[Lars Ingebrigtsen]

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

> So you should be able to control indentation of "b" above by tweaking
> the (:after . "=") or (:before . "=") rule.

I've never played around with the SMIE indentation before, but looking
at other examples, I came up with this:

diff --git a/lisp/progmodes/octave.el b/lisp/progmodes/octave.el
index a1a5192ee1..5d877fc6ba 100644
--- a/lisp/progmodes/octave.el
+++ b/lisp/progmodes/octave.el
@@ -460,7 +460,8 @@ octave-smie-rules
          (smie-rule-parent octave-block-offset)
        ;; For (invalid) code between switch and case.
        ;; (if (smie-rule-parent-p "switch") 4)
-       nil))))
+       nil))
+    ('(:after . "=") octave-block-offset)))
 
 (defun octave-indent-comment ()
   "A function for `smie-indent-functions' (which see)."

And the results look OK to me:

a = \
    b
foo = \
      dasd

So I've pushed this to Emacs 28; feel free to tweak further.

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

bug#17955: 24.3.92; octave.el: indentation following ... or \

[Stefan Monnier via Bug reports for GNU Emacs, the Swiss army knife of text editors]

Lars Ingebrigtsen [2021-05-29 07:06:41] wrote:
> Stefan Monnier <monnier@iro.umontreal.ca> writes:
>> So you should be able to control indentation of "b" above by tweaking
>> the (:after . "=") or (:before . "=") rule.
>
> I've never played around with the SMIE indentation before, but looking
> at other examples, I came up with this:
>
> diff --git a/lisp/progmodes/octave.el b/lisp/progmodes/octave.el
> index a1a5192ee1..5d877fc6ba 100644
> --- a/lisp/progmodes/octave.el
> +++ b/lisp/progmodes/octave.el
> @@ -460,7 +460,8 @@ octave-smie-rules
>           (smie-rule-parent octave-block-offset)
>         ;; For (invalid) code between switch and case.
>         ;; (if (smie-rule-parent-p "switch") 4)
> -       nil))))
> +       nil))
> +    ('(:after . "=") octave-block-offset)))
>  
>  (defun octave-indent-comment ()
>    "A function for `smie-indent-functions' (which see)."
>
> And the results look OK to me:
>
> a = \
>     b
> foo = \
>       dasd
>
> So I've pushed this to Emacs 28; feel free to tweak further.

I suspect this behavior, is still not quite what users would want:
why break the line if the continuation starts where you broke?
Maybe

    ('(:after . "=") (smie-rule-parent octave-block-offset))))

would work better?

And we should add tests for these changes, since regressions are all too
easy to introduce when it comes to indentation rules.


        Stefan

bug#17955: 24.3.92; octave.el: indentation following ... or \

[Lars Ingebrigtsen]

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

> I suspect this behavior, is still not quite what users would want:
> why break the line if the continuation starts where you broke?
> Maybe
>
>     ('(:after . "=") (smie-rule-parent octave-block-offset))))
>
> would work better?
>
> And we should add tests for these changes, since regressions are all too
> easy to introduce when it comes to indentation rules.

OK; now done.

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no