Re: [elpa] master 6c3842e 2/6: Add scala-mode to excluded-modes (#88)

Re: [elpa] master 6c3842e 2/6: Add scala-mode to excluded-modes (#88)

[Stefan Monnier]

> --- a/aggressive-indent.el
> +++ b/aggressive-indent.el
> @@ -136,6 +136,7 @@ Please include this in your report!"
>      netcmd-mode
>      python-mode
>      sass-mode
> +    scala-mode
>      slim-mode
>      special-mode
>      shell-mode

Isn't aggressive-indent-excluded-modes a bit blunt/arbitrary?

I mean, shouldn't we argue for each mode *why* it shouldn't use
aggressive-indent, and then maybe come up with a way to make this
variable unnecessary (because we can determine it automatically)?

E.g. maybe some of those modes shouldn't use aggressive-indent simply
because auto-reindentation can't be done reliably (in which case,
aggressive-indent could check something like electric-indent-inhibit
which was introduced for the same kind of needs)...


        Stefan

Re: [elpa] master 6c3842e 2/6: Add scala-mode to excluded-modes (#88)

[Artur Malabarba]

On 12 October 2016 at 16:48, Stefan Monnier <monnier@iro.umontreal.ca> wrote:
>> --- a/aggressive-indent.el
>> +++ b/aggressive-indent.el
>> @@ -136,6 +136,7 @@ Please include this in your report!"
>>      netcmd-mode
>>      python-mode
>>      sass-mode
>> +    scala-mode
>>      slim-mode
>>      special-mode
>>      shell-mode
>
> Isn't aggressive-indent-excluded-modes a bit blunt/arbitrary?

Indeed. :-(

> I mean, shouldn't we argue for each mode *why* it shouldn't use
> aggressive-indent, and then maybe come up with a way to make this
> variable unnecessary (because we can determine it automatically)?

Scala-mode was added because it takes the liberty of doing some
whitespace cleanup during indent-region, so it was preventing users
from typing properly. See:
https://github.com/Malabarba/aggressive-indent-mode/issues/87

I didn't have time to look for a config option to disable that, (and
even then, I don't think I want aggressive-indent to go around
changing semi-unrelated) so I disabled it in scala-mode.

> E.g. maybe some of those modes shouldn't use aggressive-indent simply
> because auto-reindentation can't be done reliably (in which case,
> aggressive-indent could check something like electric-indent-inhibit
> which was introduced for the same kind of needs)...

Yes. A lot of these are there for this reason, and we could use
`electric-indent-inhibit` for that (I didn't know about it).

Other modes are there because indentation makes no sense for them at
all. Such as shell-mode, which I see doesn't set
electric-indent-inhibit. We can't avoid these modes simply by
restricting ourselves to modes derived from prog-mode, because some
modes (like html or xml) do make sense to indent but are not derived
from prog-mode. Now that I have the advantage of hindsight, I see it
would have been easier to restrict to prog-mode and implement a
white-list, instead of taking everything and implementing a
black-list.

Re: [elpa] master 6c3842e 2/6: Add scala-mode to excluded-modes (#88)

[Stefan Monnier]

> Scala-mode was added because it takes the liberty of doing some
> whitespace cleanup during indent-region, so it was preventing users
> from typing properly. See:
> https://github.com/Malabarba/aggressive-indent-mode/issues/87

BTW, another way to avoid this problem is to skip reindentation if
(and (eolp) (memq (char-before) '(?\s ?\t))).

>> E.g. maybe some of those modes shouldn't use aggressive-indent simply
>> because auto-reindentation can't be done reliably (in which case,
>> aggressive-indent could check something like electric-indent-inhibit
>> which was introduced for the same kind of needs)...
> Yes.  A lot of these are there for this reason, and we could use
> `electric-indent-inhibit` for that (I didn't know about it).

Maybe we should have another variable that's not bound to
electric-indent, something like `indent-is-not-reliable'.

> Other modes are there because indentation makes no sense for them at
> all.  Such as shell-mode, which I see doesn't set
> electric-indent-inhibit.

AFAIK shell-mode's indentation works just fine and reliably.  Maybe you
don't need it when you write 2-line scripts, but for larger scripts
(such as https://gitlab.com/monnier/bugit/blob/master/bugit), it makes
just as much sense as for any other programming language.

> from prog-mode.  Now that I have the advantage of hindsight, I see it
> would have been easier to restrict to prog-mode and implement a
> white-list, instead of taking everything and implementing a
> black-list.

electric-indent does OK with a black list, so maybe it was not such
a bad decision.


        Stefan

Re: [elpa] master 6c3842e 2/6: Add scala-mode to excluded-modes (#88)

[Artur Malabarba]

> AFAIK shell-mode's indentation works just fine and reliably.  Maybe you
> don't need it when you write 2-line scripts, but for larger scripts
> (such as https://gitlab.com/monnier/bugit/blob/master/bugit), it makes
> just as much sense as for any other programming language.

You're thinking of sh-mode. `shell-mode' is the one you get with `M-x shell'.

Re: [elpa] master 6c3842e 2/6: Add scala-mode to excluded-modes (#88)

[Stefan Monnier]

>> AFAIK shell-mode's indentation works just fine and reliably.  Maybe you
>> don't need it when you write 2-line scripts, but for larger scripts
>> (such as https://gitlab.com/monnier/bugit/blob/master/bugit), it makes
>> just as much sense as for any other programming language.
> You're thinking of sh-mode. `shell-mode' is the one you get with `M-x shell'.

Duh!  comint-derived modes don't make much sense there, indeed.
But you don't need to list it: just check the value of
indent-line-function and weed out the `indent-relative' case.


        Stefan