bug#35286: 26.2; indent-sexp broken

[Leo Liu <sdl.web@gmail.com>]

On 2019-04-15 20:54 -0400, Noam Postavsky wrote:
> The minimal example would be (with point before "foo", the "xx)"
> part gets indented too):
>
> foo (bar
>  xx)
>
> I'm not sure if anyone relies on that (I broke it in 26.1 and nobody
> reported it).  The main motivating example is
>
> #s(foo
> bar)
>
> which is due to forward-sexp not handling #s correctly for Emacs Lisp
> (it's treated as a separate sexp rather than as a whole record
> expression).

Thanks for the examples.

>> Not just comments though it can be anything. For example, put the
>> following in prolog-mode. Move point to the first `{' char and then M-x
>> indent-sexp.
>
> Hmm, I'm a bit confused here.  I thought indent-sexp is only good for
> Lisp-based languages.  Do lisp-indent-calc-next/calculate-lisp-indent
> give usable results for prolog or ruby??

Treating things between parentheses as sexp and using indent-sexp to
indent it is not lisp-specific (imagine if we couldn't use forward-sexp
on parentheses in non-lisp languages). I think I have been doing it for
over a decade now. Sometimes indirectly. For example, I use paredit to
input parentheses everywhere and paredit uses indent-sexp to indent the
inserted pair. It's been working nicely forever until the change in 26.2
breaks it.