rainbow-delimiters-mode and syntax-begin-function

rainbow-delimiters-mode and syntax-begin-function

[fatiparty--- via Users list for the GNU Emacs text editor]

I am going through 

(define-minor-mode rainbow-delimiters-mode

There is for instance 

;; Obsolete equivalent of `syntax-begin-function'.
(when (boundp 'font-lock-beginning-of-syntax-function)
      (set (make-local-variable
            'font-lock-beginning-of-syntax-function) nil)) )

What is this about?  And how is one modifying `syntax-begin-function' ?

rainbow-delimiters-mode, syntax-begin-function and font-lock-beginning-of-syntax-function

[fatiparty--- via Users list for the GNU Emacs text editor]

Jan 23, 2022, 19:58 by help-gnu-emacs@gnu.org:

Would sections [1] and [2] still be relevant in this code? 

It includes the claim  of not using `syntax-begin-function' and
that `font-lock-beginning-of-syntax-function' could be obsolete.

Have looked for `font-lock-beginning-of-syntax-function' but
could not find information of possible obsolescence.
(define-minor-mode rainbow-delimiters-minor-mode
  "Colour Brace Marks according to their depth."
  nil
  ""   ; exclude modeline indicator
  nil 
  
  (font-lock-remove-keywords nil my-font-lock)

  (when rainbow-delimiters-minor-mode
    (font-lock-add-keywords nil rainbow-delimiters-font-lock 'append)
    (set (make-local-variable 'jit-lock-contextually) t)

    ;; flush-cache to avoid old values when modifying `syntax-begin-function'
    (when (or (bound-and-true-p syntax-begin-function)
              (bound-and-true-p font-lock-beginning-of-syntax-function))
      (syntax-ppss-flush-cache 0))

    ;; [1] Don't use `syntax-begin-function', as it may break the
    ;; assumption that `syntax-ppss' is equivalent to
    ;; `parse-partial-sexp' from `point-min'.

    (when (boundp 'syntax-begin-function)
      (set (make-local-variable 'syntax-begin-function) nil))

    ;; [2] Obsolete equivalent of `syntax-begin-function'.
    (when (boundp 'font-lock-beginning-of-syntax-function)
      (set (make-local-variable
            'font-lock-beginning-of-syntax-function) nil)) )

  (when font-lock-mode
    (if (fboundp 'font-lock-flush)
        (font-lock-flush)
      (with-no-warnings (font-lock-fontify-buffer)))))

Re: rainbow-delimiters-mode, syntax-begin-function and font-lock-beginning-of-syntax-function

[Stefan Monnier via Users list for the GNU Emacs text editor]

> Have looked for `font-lock-beginning-of-syntax-function' but
> could not find information of possible obsolescence.

    % grep font-lock-beginning-of-syntax-function etc/NEWS*
    etc/NEWS.19:font-lock-beginning-of-syntax-function and font-lock-keywords-case-fold-search.
    etc/NEWS.25:Removed 'font-lock-beginning-of-syntax-function' and the SYNTAX-BEGIN
    %


-- Stefan

rainbow-delimiters-mode, syntax-begin-function and font-lock-beginning-of-syntax-function

[fatiparty--- via Users list for the GNU Emacs text editor]

Jan 25, 2022, 11:18 by help-gnu-emacs@gnu.org:

>> Have looked for `font-lock-beginning-of-syntax-function' but
>> could not find information of possible obsolescence.
>>
>
> % grep font-lock-beginning-of-syntax-function etc/NEWS*
>  etc/NEWS.19:font-lock-beginning-of-syntax-function and font-lock-keywords-case-fold-search.
>  etc/NEWS.25:Removed 'font-lock-beginning-of-syntax-function' and the SYNTAX-BEGIN
>  %
>
>
> -- Stefan
>


'syntax-begin-function' has also been declared obsolete.

Would this mean that I can remove the following pieces

  (when (bound-and-true-p syntax-begin-function)
       (syntax-ppss-flush-cache 0))
and

   (when (boundp 'syntax-begin-function)        (set (make-local-variable 'syntax-begin-function) nil)) )


----

Code:

(define-minor-mode richerenkov-minor-mode
  "Colour Brace Marks according to their depth."
  nil
  ""   ; exclude modeline indicator
  nil 
  
  (font-lock-remove-keywords nil test-font-lock)

  (when test-minor-mode
    (font-lock-add-keywords nil test-font-lock 'append)
    (set (make-local-variable 'jit-lock-contextually) t)

    ;; flush-cache to avoid old values when modifying `syntax-begin-function'
    (when (bound-and-true-p syntax-begin-function)
      (syntax-ppss-flush-cache 0))

    ;; Don't use `syntax-begin-function', as it may break the
    ;; assumption that `syntax-ppss' is equivalent to
    ;; `parse-partial-sexp' from `point-min'.

    (when (boundp 'syntax-begin-function)
      (set (make-local-variable 'syntax-begin-function) nil)) )

  (when font-lock-mode
    (if (fboundp 'font-lock-flush)
        (font-lock-flush)
      (with-no-warnings (font-lock-fontify-buffer)))))

Re: rainbow-delimiters-mode, syntax-begin-function and font-lock-beginning-of-syntax-function

[Stefan Monnier]

fatiparty@tutanota.com [2022-01-25 09:46:35] wrote:
> Jan 25, 2022, 11:18 by help-gnu-emacs@gnu.org:
>>> Have looked for `font-lock-beginning-of-syntax-function' but
>>> could not find information of possible obsolescence.
>> % grep font-lock-beginning-of-syntax-function etc/NEWS*
>>  etc/NEWS.19:font-lock-beginning-of-syntax-function and font-lock-keywords-case-fold-search.
>>  etc/NEWS.25:Removed 'font-lock-beginning-of-syntax-function' and the SYNTAX-BEGIN
>>  %
> 'syntax-begin-function' has also been declared obsolete.

"Also" isn't quite right here: the above grep shows the other var has
been *removed* which is different from being declared obsolete.

> Would this mean that I can remove the following pieces

It means those pieces of code only have an effect for packages that
still use that obsolete variable.  Whether you want to remove it or not
is a choice I can't make for you.


        Stefan

rainbow-delimiters-mode, syntax-begin-function and font-lock-beginning-of-syntax-function

[fatiparty--- via Users list for the GNU Emacs text editor]

Jan 26, 2022, 01:04 by monnier@iro.umontreal.ca:

> fatiparty@tutanota.com [2022-01-25 09:46:35] wrote:
>
>> Jan 25, 2022, 11:18 by help-gnu-emacs@gnu.org:
>>
>>>> Have looked for `font-lock-beginning-of-syntax-function' but
>>>> could not find information of possible obsolescence.
>>>>
>>> % grep font-lock-beginning-of-syntax-function etc/NEWS*
>>>  etc/NEWS.19:font-lock-beginning-of-syntax-function and font-lock-keywords-case-fold-search.
>>>  etc/NEWS.25:Removed 'font-lock-beginning-of-syntax-function' and the SYNTAX-BEGIN
>>>  %
>>>
>> 'syntax-begin-function' has also been declared obsolete.
>>

In `etc/NEWS.25' there is the comment 

** 'syntax-begin-function' is declared obsolete.

I guess that if one stops using  'syntax-begin-function', it would help remove
a dependency which would not be required in future.



> "Also" isn't quite right here: the above grep shows the other var has
> been *removed* which is different from being declared obsolete.
>
>> Would this mean that I can remove the following pieces
>>
>
> It means those pieces of code only have an effect for packages that
> still use that obsolete variable.  Whether you want to remove it or not
> is a choice I can't make for you.
>
>
>  Stefan
>
Right, if the version of emacs has `syntax-begin-function', then one could use that,
as has been done.  I am making a new version without calls to things that would,
 be obsolete, but without breaking the colouring functionality. 

There is a call to flush the cache `(syntax-ppss-flush-cache 0' with 
`(boundp 'syntax-begin-function)'.  Would this still be necessary?  

Re: rainbow-delimiters-mode, syntax-begin-function and font-lock-beginning-of-syntax-function

[Stefan Monnier]

> Right, if the version of emacs has `syntax-begin-function', then one could use that,
> as has been done.  I am making a new version without calls to things that would,
>  be obsolete, but without breaking the colouring functionality. 
>
> There is a call to flush the cache `(syntax-ppss-flush-cache 0' with 
> `(boundp 'syntax-begin-function)'.  Would this still be necessary?  

I recommend you read a bit more to understand a bit what these things do
(e.g. what does it mean for `syntax-begin-function` to be nil) because
changing code which you don't understand at all is just not fun.


        Stefan

rainbow-delimiters-mode, syntax-begin-function and font-lock-beginning-of-syntax-function

[fatiparty--- via Users list for the GNU Emacs text editor]

Jan 26, 2022, 01:55 by monnier@iro.umontreal.ca:

>> Right, if the version of emacs has `syntax-begin-function', then one could use that,
>> as has been done.  I am making a new version without calls to things that would,
>>  be obsolete, but without breaking the colouring functionality. 
>>
>> There is a call to flush the cache `(syntax-ppss-flush-cache 0' with 
>> `(boundp 'syntax-begin-function)'.  Would this still be necessary? 
>>
>
> I recommend you read a bit more to understand a bit what these things do
> (e.g. what does it mean for `syntax-begin-function` to be nil) because
> changing code which you don't understand at all is just not fun.
>
>  Stefan
>

If I completely remove the section (in Emacs 27.2):

(when (or (bound-and-true-p syntax-begin-function)
          (bound-and-true-p font-lock-beginning-of-syntax-function))

  (syntax-ppss-flush-cache 0))

(when (boundp 'syntax-begin-function)
  (set (make-local-variable 'syntax-begin-function) nil))

(when (boundp 'font-lock-beginning-of-syntax-function)
  (set (make-local-variable 'font-lock-beginning-of-syntax-function) nil))


I don't see any obvious  adverse effects.


What was syntax-begin-function obsoleted in favour of?  

rainbow-delimiters-mode, syntax-begin-function and font-lock-beginning-of-syntax-function

[fatiparty--- via Users list for the GNU Emacs text editor]

Jan 26, 2022, 01:55 by monnier@iro.umontreal.ca:

>> Right, if the version of emacs has `syntax-begin-function', then one could use that,
>> as has been done.  I am making a new version without calls to things that would,
>>  be obsolete, but without breaking the colouring functionality. 
>>
>> There is a call to flush the cache `(syntax-ppss-flush-cache 0' with 
>> `(boundp 'syntax-begin-function)'.  Would this still be necessary? 
>>
>
> I recommend you read a bit more to understand a bit what these things do
> (e.g. what does it mean for `syntax-begin-function` to be nil) because
> changing code which you don't understand at all is just not fun.
>
>  Stefan
>
syntax-begin-function is now always nil though.