Different fontification in temp buffer

Different fontification in temp buffer

[Daniele Nicolodi]

Hello,

I'm hacking on a minor mode and I'm writing unit tests for the
fontification feature. I am encountering a strange issue: the
fontification is different if execute in a temp buffer or in a regular
buffer. In particular:

  (with-temp-buffer
    (insert string)
    (fundamental-mode)
    (beancount-mode)
    (font-lock-ensure)
    (buffer-string))

and

  (with-current-buffer (generate-new-buffer "*test*")
    (insert string)
    (fundamental-mode)
    (beancount-mode)
    (font-lock-ensure)
    (buffer-string))

strangely result in two different results. Why is it so?

Thank you.

Cheers,
Dan

Re: Different fontification in temp buffer

[Daniele Nicolodi]

On 19/01/2019 17:48, Daniele Nicolodi wrote:
> Hello,
> 
> I'm hacking on a minor mode and I'm writing unit tests for the
> fontification feature. I am encountering a strange issue: the
> fontification is different if execute in a temp buffer or in a regular
> buffer. In particular:
> 
>   (with-temp-buffer
>     (insert string)
>     (fundamental-mode)
>     (beancount-mode)
>     (font-lock-ensure)
>     (buffer-string))
> 
> and
> 
>   (with-current-buffer (generate-new-buffer "*test*")
>     (insert string)
>     (fundamental-mode)
>     (beancount-mode)
>     (font-lock-ensure)
>     (buffer-string))
> 
> strangely result in two different results. Why is it so?

Investigating further, the difference between the two cases is that
with-temp-buffer creates a buffer that does not keep undo information.
Surely enough, if I modify the second code to do the same (using a
buffer name starting with a space character) I obtain the same behavior.

What's left to understand is why this results in two different syntax
table being applied to the buffer.

Cheers,
Dan

Re: Different fontification in temp buffer

[Noam Postavsky]

On Sat, 19 Jan 2019 at 21:14, Daniele Nicolodi <daniele@grinta.net> wrote:
>
> On 19/01/2019 17:48, Daniele Nicolodi wrote:

> > I'm hacking on a minor mode and I'm writing unit tests for the
> > fontification feature. I am encountering a strange issue: the
> > fontification is different if execute in a temp buffer or in a regular
> > buffer.

> Investigating further, the difference between the two cases is that
> with-temp-buffer creates a buffer that does not keep undo information.

It's not about undo info, but rather that font-lock mode doesn't turn
on in temp buffers.

See in the definition of font-lock-mode:
http://git.savannah.gnu.org/cgit/emacs.git/tree/lisp/font-core.el?h=emacs-26.1#n135

  ;; Don't turn on Font Lock mode if we don't have a display (we're running a
  ;; batch job) or if the buffer is invisible (the name starts with a space).
  (when (or noninteractive (eq (aref (buffer-name) 0) ?\s))
    (setq font-lock-mode nil))

Re: Different fontification in temp buffer

[Daniele Nicolodi]

Hi Noam,

On 20/01/2019 06:57, Noam Postavsky wrote:
> On Sat, 19 Jan 2019 at 21:14, Daniele Nicolodi <daniele@grinta.net> wrote:
>>
>> On 19/01/2019 17:48, Daniele Nicolodi wrote:
> 
>>> I'm hacking on a minor mode and I'm writing unit tests for the
>>> fontification feature. I am encountering a strange issue: the
>>> fontification is different if execute in a temp buffer or in a regular
>>> buffer.
> 
>> Investigating further, the difference between the two cases is that
>> with-temp-buffer creates a buffer that does not keep undo information.
> 
> It's not about undo info, but rather that font-lock mode doesn't turn
> on in temp buffers.
> 
> See in the definition of font-lock-mode:
> http://git.savannah.gnu.org/cgit/emacs.git/tree/lisp/font-core.el?h=emacs-26.1#n135
> 
>   ;; Don't turn on Font Lock mode if we don't have a display (we're running a
>   ;; batch job) or if the buffer is invisible (the name starts with a space).
>   (when (or noninteractive (eq (aref (buffer-name) 0) ?\s))
>     (setq font-lock-mode nil))

I'm explicitly turning on font-lock mode with (font-lock-ensure). The
buffer is indeed fontified, but apparently with a different syntax
table. However, I don't fully understand what beancount-mode does with
the syntax table, thus I cannot exclude a bug there.

Cheers,
Dan

Re: Different fontification in temp buffer

[Noam Postavsky]

On Sun, 20 Jan 2019 at 10:55, Daniele Nicolodi <daniele@grinta.net> wrote:

> > http://git.savannah.gnu.org/cgit/emacs.git/tree/lisp/font-core.el?h=emacs-26.1#n135
> >
> >   ;; Don't turn on Font Lock mode if we don't have a display (we're running a
> >   ;; batch job) or if the buffer is invisible (the name starts with a space).
> >   (when (or noninteractive (eq (aref (buffer-name) 0) ?\s))
> >     (setq font-lock-mode nil))
>
> I'm explicitly turning on font-lock mode with (font-lock-ensure). The
> buffer is indeed fontified, but apparently with a different syntax
> table. However, I don't fully understand what beancount-mode does with
> the syntax table, thus I cannot exclude a bug there.

As far as I can tell, font-lock-ensure doesn't actually turn on
font-lock mode, it just goes ahead and fontifies the buffer anyway.
Perhaps beancount-mode's fontification functions don't quite work
properly in that case.

Re: Different fontification in temp buffer

[Daniele Nicolodi]

On 20/01/2019 10:56, Noam Postavsky wrote:
> On Sun, 20 Jan 2019 at 10:55, Daniele Nicolodi <daniele@grinta.net> wrote:
> 
>>> http://git.savannah.gnu.org/cgit/emacs.git/tree/lisp/font-core.el?h=emacs-26.1#n135
>>>
>>>   ;; Don't turn on Font Lock mode if we don't have a display (we're running a
>>>   ;; batch job) or if the buffer is invisible (the name starts with a space).
>>>   (when (or noninteractive (eq (aref (buffer-name) 0) ?\s))
>>>     (setq font-lock-mode nil))
>>
>> I'm explicitly turning on font-lock mode with (font-lock-ensure). The
>> buffer is indeed fontified, but apparently with a different syntax
>> table. However, I don't fully understand what beancount-mode does with
>> the syntax table, thus I cannot exclude a bug there.
> 
> As far as I can tell, font-lock-ensure doesn't actually turn on
> font-lock mode, it just goes ahead and fontifies the buffer anyway.
> Perhaps beancount-mode's fontification functions don't quite work
> properly in that case.

Ah! I didn't know that there is a difference between fontifying a buffer
and enabling font-lock mode. I'll try to understand the differences.

Thanks!

Cheers,
Dan

Re: Different fontification in temp buffer

[Stefan Monnier]

> fontification is different if execute in a temp buffer or in a regular
> buffer. In particular:

Sounds like a bug.  Please report it (and describe the actual difference
you see).


        Stefan