bug#41522: 26.3; Narrowing can prevent syntax propertization

[Ikumi Keita <ikumi@ikumi.que.jp>]

Hi Dmitry, thanks for your reply.

>>>>> Dmitry Gutov <dgutov@yandex.ru> writes:
> Our current stance on the matter is that, in the simple case, code
> should call syntax-ppss from "widened" state.

I suppose that this means that the code should call syntax-ppss (or
something like it) "before" narrowing is done, right? This code works
for me
----------------------------------------------------------------------
(progn (goto-char (point-max))
       (syntax-propertize (point)) ; <-
       (narrow-to-region (point) (- (point) 50))
       (syntax-ppss)
       (widen))
----------------------------------------------------------------------
while these don't:
----------------------------------------------------------------------
(progn (goto-char (point-max))
       (narrow-to-region (point) (- (point) 50))
       (syntax-ppss)
       (widen)
       (syntax-ppss) ; <-
)
----------------------------------------------------------------------
----------------------------------------------------------------------
(progn (goto-char (point-max))
       (narrow-to-region (point) (- (point) 50))
       (syntax-ppss)
       (widen)
       (syntax-propertize (point)) ; <-
)
----------------------------------------------------------------------

> And the code which doesn't should be fixed. Or if it really need to do
> that (for some reason), it should take care to handle any undesired
> sife-effects itself.

Hmm, then elisp programmers have to check whether each function used in
the code uses narrowing+syntactic functions inside it if this kind of
problem happens. I feel it a bit troublesome but at the same time I
understand it is very difficult to solve this problem cleanly.

OK, I try to be aware of this matter afterwards. Thanks for your
advice!

Regards,
Ikumi Keita