Re: [Emacs-diffs] master aae9ac2: Avoid an infloop in shr when filling text with :align-to properties

Re: [Emacs-diffs] master aae9ac2: Avoid an infloop in shr when filling text with :align-to properties

[Stefan Monnier]

Hi Lars,

I hope you had a nice vacation.

> @@ -2363,6 +2364,7 @@ flags that control whether to collect or render objects."
>  	(car (window-text-pixel-size nil (point-min) (point-max)))))))
>  
>  (defun shr-render-td (dom width fill)
> +  (setq d dom)
>    (let ((cache (intern (format "shr-td-cache-%s-%s" width fill))))
>      (or (dom-attr dom cache)
>  	(and fill

Really?


        Stefan

Re: [Emacs-diffs] master aae9ac2: Avoid an infloop in shr when filling text with :align-to properties

[Lars Ingebrigtsen]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

> Really?

My mind must still be on vacation; fixed.

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

Re: master aae9ac2: Avoid an infloop in shr when filling text with :align-to properties

[Lars Ingebrigtsen]

larsi@gnus.org (Lars Ingebrigtsen) writes:

> +  (setq d dom)

This reminds me: Aren't there any good ways to get at variables in the
debugger?  If you have

;;; -*- lexical-binding: t -*-

(defun foo (bar)
  (let ((zot 1))
    (debug)))

then you can `e bar RET' to display the value of `bar'.  But if you say
`(setq a bar)', you get

Debugger entered--Lisp error: (void-variable bar)

What's up with that?

Furthermore, if you `v' stack line, you get:

  foo(t)
    bar = t

That's nice, but again there's no way to get at that value that I'm
aware of.  This only matters when you have long (and possibly
unprintable) values, of course, but then you have to resort to putting
setqs in the source code, which sucks.

Could we add a command that, when on a variable value, copies the value
into...  uhm...  a variable prompted for by the command?  Sounds pretty
trivial to implement.

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

Re: master aae9ac2: Avoid an infloop in shr when filling text with :align-to properties

[Andreas Schwab]

On Aug 23 2019, Lars Ingebrigtsen <larsi@gnus.org> wrote:

> larsi@gnus.org (Lars Ingebrigtsen) writes:
>
>> +  (setq d dom)
>
> This reminds me: Aren't there any good ways to get at variables in the
> debugger?  If you have
>
> ;;; -*- lexical-binding: t -*-
>
> (defun foo (bar)
>   (let ((zot 1))
>     (debug)))
>
> then you can `e bar RET' to display the value of `bar'.  But if you say
> `(setq a bar)', you get
>
> Debugger entered--Lisp error: (void-variable bar)

Worksforme.  You have to eval it in the stack frame, of course.

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 7578 EB47 D4E5 4D69 2510  2552 DF73 E780 A9DA AEC1
"And now for something completely different."

Re: master aae9ac2: Avoid an infloop in shr when filling text with :align-to properties

[Lars Ingebrigtsen]

Andreas Schwab <schwab@linux-m68k.org> writes:

> Worksforme.  You have to eval it in the stack frame, of course.

What do you mean by "in the stack frame"?  With point on a specific
line?

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

Re: master aae9ac2: Avoid an infloop in shr when filling text with :align-to properties

[Andreas Schwab]

On Aug 23 2019, Lars Ingebrigtsen <larsi@gnus.org> wrote:

> Andreas Schwab <schwab@linux-m68k.org> writes:
>
>> Worksforme.  You have to eval it in the stack frame, of course.
>
> What do you mean by "in the stack frame"?  With point on a specific
> line?

With e (setq a bar) RET.

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 7578 EB47 D4E5 4D69 2510  2552 DF73 E780 A9DA AEC1
"And now for something completely different."

Re: master aae9ac2: Avoid an infloop in shr when filling text with :align-to properties

[Lars Ingebrigtsen]

Andreas Schwab <schwab@linux-m68k.org> writes:

>> What do you mean by "in the stack frame"?  With point on a specific
>> line?
>
> With e (setq a bar) RET.

*scratches head*  Now it works for me, too.  I was pretty sure I was
 using the `e' command, but perhaps muscle memory made me use `M-:'
 instead? 

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

Re: master aae9ac2: Avoid an infloop in shr when filling text with :align-to properties

[Stefan Monnier]

> *scratches head*  Now it works for me, too.  I was pretty sure I was
>  using the `e' command, but perhaps muscle memory made me use `M-:'
>  instead? 

That would be my guess: it happens to me all the time.


        Stefan

Re: master aae9ac2: Avoid an infloop in shr when filling text with :align-to properties

[Richard Stallman]

[[[ To any NSA and FBI agents reading my email: please consider    ]]]
[[[ whether defending the US Constitution against all enemies,     ]]]
[[[ foreign or domestic, requires you to follow Snowden's example. ]]]

  > > *scratches head*  Now it works for me, too.  I was pretty sure I was
  > >  using the `e' command, but perhaps muscle memory made me use `M-:'
  > >  instead? 

  > That would be my guess: it happens to me all the time.

This must be a common mistake.
Should the debugger rebind M-: to be like e?
Make it ask for confirmation?

-- 
Dr Richard Stallman
President, Free Software Foundation (https://gnu.org, https://fsf.org)
Internet Hall-of-Famer (https://internethalloffame.org)

RE: master aae9ac2: Avoid an infloop in shr when filling text with :align-to properties

[Drew Adams]

>>> I  was pretty sure I was using the `e' command, but perhaps
>>> muscle memory made me use `M-:' instead?
>> That would be my guess: it happens to me all the time.
> 
> This must be a common mistake.
> Should the debugger rebind M-: to be like e?

If you mean rebind `M-:' to do what `e' does then
no, definitely not, IMO.  Users need to be able to
use _both_ `e' and `M-:' in the debugger buffer,
to get their different behaviors.

> Make it ask for confirmation?

Dunno whether you mean `e' or `M-:' (in debugger),
but I'd again say no.  It's perfectly normal to use
both `M-:' and `e' in the debugger, for different
purposes.  I don't see why either should ask for
confirmation.

When you use `e' we already use a prompt that should
make it clear that `e' is being used (provided you
read the prompt):

  Eval in stack frame:

If we think that's not obvious/clear enough, perhaps
adding some flash (color? ding? mode-line indicator?)
would help.

That might not help if someone uses `M-:' by mistake,
but if a user gets used to expecting it for `e' then
its lack might be noticeable enough.

Another possibility is to change the prompt for `M-:'
when in the debugger, to make it clear that the eval
takes place in the outside-debugger context, not in
the debugger stack frame.  But that introduces a
different prompt for `M-:' depending on the context,
which is maybe not so good.

Do I think this is really a problem that needs
fixing?  No.  But those are suggestions that occur
to me, in case others think it is.

Re: master aae9ac2: Avoid an infloop in shr when filling text with :align-to properties

[Lars Ingebrigtsen]

Richard Stallman <rms@gnu.org> writes:

> This must be a common mistake.
> Should the debugger rebind M-: to be like e?
> Make it ask for confirmation?

I think it's somewhat useful to be able to `M-:' as normal in the
debugging window (in addition to `e', which is probably what you want
most of the time).  So rebinding doesn't seem like the best solution.
Perhaps.

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no