bug#31772: 26.1; (thing-at-point 'list) regression

[Tino Calancha]

Eli Zaretskii <eliz@gnu.org> writes:

>> Due to change
>> 
>> commit 76e297c15f6312a83599aab216be0396e9aac5c5
>> Author: Tino Calancha
>> Date:   Thu Nov 3 20:33:19 2016 +0900
>> 
>> (thing-at-point 'list) no longer works in comments; always return nil.
>
> Should it?  AFACS, in Emacs 25 the above returned nonsensical values
> when invoked inside Lisp comments.

[CC the relevant thingatpt experts]

Initially I thought that; after discuss with Emacs colleages
I changed my mine.

If you are in a buffer in emacs-lisp-mode, I just see
one list of length 3 in the following line:
("foo" "(bar)" "baz")

"(bar)" is a string.

If we comment the line, then there is no list at all; just a comment.
That is tested in thingatpt-tests.el (thing-at-point-bug24627).

You could change to fundamental-mode, then you would get again
the full list: because the line is not commented anymore.

To me this behaviour has sense, and it's predictible.  Maybe others
disagree.

PD: I must admit I am a bit worry everytime we start discussing
about this lib... :-S