can shr-start go away?

can shr-start go away?

[Mike Kupfer]

shr.el has

  (defvar shr-start nil)

But the only place in Emacs that checks the value of shr-start is
shr-tag-a, and shr-tag-a overrides the defvar via a "let" declaration.
shr-tag-img can set shr-start, but it doesn't use shr-start itself.

Am I missing something?  (I'm still something of a Lisp newbie.)

If shr-start can indeed go away, I can do that as part of the fix for
bug#60423.

thanks,
mike

Re: can shr-start go away?

[Andreas Schwab]

On Nov 11 2024, Mike Kupfer wrote:

> shr-tag-img can set shr-start, but it doesn't use shr-start itself.

shr-tag-img sets it for shr-tag-a.

-- 
Andreas Schwab, SUSE Labs, schwab@suse.de
GPG Key fingerprint = 0196 BAD8 1CE9 1970 F4BE  1748 E4D4 88E3 0EEA B9D7
"And now for something completely different."

Re: can shr-start go away?

[Mike Kupfer]

Andreas Schwab wrote:

> On Nov 11 2024, Mike Kupfer wrote:
> 
> > shr-tag-img can set shr-start, but it doesn't use shr-start itself.
> 
> shr-tag-img sets it for shr-tag-a.

That might have been the intent at one time.  But shr-tag-a has

  (let ((url (dom-attr dom 'href))
        (title (dom-attr dom 'title))
	(start (point))
	shr-start)

Doesn't that create a new binding for shr-start, which means that
shr-tag-a will not see whatever shr-tag-img did?  (Apologies if this
seems like a dumb question.  Like I wrote, I'm still something of a Lisp
newbie.)

The code has been like this for several releases.  It even precedes the
conversion of shr.el to lexical binding.

Maybe I should ask about this on ding?

mike

Re: can shr-start go away?

[Andreas Schwab]

On Nov 12 2024, Mike Kupfer wrote:

> That might have been the intent at one time.  But shr-tag-a has
>
>   (let ((url (dom-attr dom 'href))
>         (title (dom-attr dom 'title))
> 	(start (point))
> 	shr-start)

It has always been that way.

> Doesn't that create a new binding for shr-start, which means that
> shr-tag-a will not see whatever shr-tag-img did?

No, shr-start has always been declared as special.

-- 
Andreas Schwab, SUSE Labs, schwab@suse.de
GPG Key fingerprint = 0196 BAD8 1CE9 1970 F4BE  1748 E4D4 88E3 0EEA B9D7
"And now for something completely different."

Re: can shr-start go away?

[Mike Kupfer]

Andreas Schwab wrote:

> No, shr-start has always been declared as special.

Okay, I think I understand the Lisp mechanics now.  The 'let' in
shr-tag-a creates a new dynamic binding for shr-start.  So as long as
shr-tag-img is called before shr-tag-a returns (i.e., in the context of
shr-tag-a's 'let' form), shr-tag-a will see what shr-tag-img did to
shr-start.  Is that correct?

thanks,
mike

Re: can shr-start go away?

[Andreas Schwab]

On Nov 12 2024, Mike Kupfer wrote:

> Okay, I think I understand the Lisp mechanics now.  The 'let' in
> shr-tag-a creates a new dynamic binding for shr-start.  So as long as
> shr-tag-img is called before shr-tag-a returns (i.e., in the context of
> shr-tag-a's 'let' form), shr-tag-a will see what shr-tag-img did to
> shr-start.  Is that correct?

Yes, that's what dynamic binding is all about.

-- 
Andreas Schwab, SUSE Labs, schwab@suse.de
GPG Key fingerprint = 0196 BAD8 1CE9 1970 F4BE  1748 E4D4 88E3 0EEA B9D7
"And now for something completely different."