Re: trunk r117960: * etc/TODO: Add a few entries, remove others, expand some

Re: trunk r117960: * etc/TODO: Add a few entries, remove others, expand some

[Dmitry Gutov]

Hi Stefan,

Any particular reason for the removal of this item?

> -** Compute the list of active keymaps *after* reading the first event.

After all, at the end of bug#10459, you've called the current solution
an ugly hack.

Cheers,
Dmitry.

Re: trunk r117960: * etc/TODO: Add a few entries, remove others, expand some

[Stefan Monnier]

> Any particular reason for the removal of this item?

AFAIK it is now fixed.

>> -** Compute the list of active keymaps *after* reading the first event.
> After all, at the end of bug#10459, you've called the current solution
> an ugly hack.

That was not a solution in src.keyboard.c but a workaround in Toby's
Elisp code, IIUC.  Since then the C code has been changed (see
src/keyboard.c:9310:

	  if (NILP (first_event))
	    {
	      first_event = key;
	      /* Even if first_event does not specify a particular
		 window/position, it's important to recompute the maps here
		 since a long time might have passed since we entered
		 read_key_sequence, and a timer (or process-filter or
		 special-event-map, ...) might have switched the current buffer
		 or the selected window from under us in the mean time.  */
	      if (fix_current_buffer
		  && (XBUFFER (XWINDOW (selected_window)->contents)
		      != current_buffer))
		Fset_buffer (XWINDOW (selected_window)->contents);
	      current_binding = active_maps (first_event);
	    }

-- Stefan

Re: trunk r117960: * etc/TODO: Add a few entries, remove others, expand some

[Dmitry Gutov]

On 09/28/2014 08:24 PM, Stefan Monnier wrote:
>> Any particular reason for the removal of this item?
>
> AFAIK it is now fixed.

Oh, that's great, thanks. Should the bugs be closed, then?

And the change be mentioned in NEWS? So that ELisp coders know starting 
with which version of Emacs they don't need the workaround anymore.

>>> -** Compute the list of active keymaps *after* reading the first event.
>> After all, at the end of bug#10459, you've called the current solution
>> an ugly hack.
>
> That was not a solution in src.keyboard.c but a workaround in Toby's
> Elisp code, IIUC.  Since then the C code has been changed (see
> src/keyboard.c:9310:

I just meant that as long as the recommended approach for how to do that 
desired thing was a hack, the TODO item was still valid.

> 	  if (NILP (first_event))
> 	    {
> 	      first_event = key;
> 	      /* Even if first_event does not specify a particular
> 		 window/position, it's important to recompute the maps here
> 		 since a long time might have passed since we entered
> 		 read_key_sequence, and a timer (or process-filter or
> 		 special-event-map, ...) might have switched the current buffer
> 		 or the selected window from under us in the mean time.  */
> 	      if (fix_current_buffer
> 		  && (XBUFFER (XWINDOW (selected_window)->contents)
> 		      != current_buffer))
> 		Fset_buffer (XWINDOW (selected_window)->contents);
> 	      current_binding = active_maps (first_event);
> 	    }

Cool, it seems to have been changed right after the 24.3 release.

Re: trunk r117960: * etc/TODO: Add a few entries, remove others, expand some

[Stefan Monnier]

>>> Any particular reason for the removal of this item?
>> AFAIK it is now fixed.
> Oh, that's great, thanks.  Should the bugs be closed, then?

I think so, tho we should double check that the corresponding bugs are
really fixed.

> And the change be mentioned in NEWS? So that ELisp coders know starting with
> which version of Emacs they don't need the workaround anymore.

I think it's too specific to merit mention in NEWS.  NEWS is already
pretty boringly detailed for the average user reading it.

> Cool, it seems to have been changed right after the 24.3 release.

Right, it might very well be that it's fixed in 24.4, indeed.


        Stefan

Re: trunk r117960: * etc/TODO: Add a few entries, remove others, expand some

[Dmitry Gutov]

On 09/30/2014 04:28 AM, Stefan Monnier wrote:

>> And the change be mentioned in NEWS? So that ELisp coders know starting with
>> which version of Emacs they don't need the workaround anymore.
>
> I think it's too specific to merit mention in NEWS.  NEWS is already
> pretty boringly detailed for the average user reading it.

If you say so. I'm not sure the average users are supposed to read the 
"Lisp Changes" section, though.

Another change I've just found out about is that 
`condition-case-unless-debug' now works in post-command-hook functions.

I.e. if you `toggle-debug-on-error' and then trigger an error inside 
`condition-case-unless-debug', the backtrace will pop up. And the 
function does not get removed from `post-command-hook' either way. Which 
is very nice all around.

But NEWS just mentions backtraces from failing timers.

Re: trunk r117960: * etc/TODO: Add a few entries, remove others, expand some

[Stefan Monnier]

> Another change I've just found out about is that
> `condition-case-unless-debug' now works in post-command-hook functions.
>
> I.e. if you `toggle-debug-on-error' and then trigger an error inside
> `condition-case-unless-debug', the backtrace will pop up. And the function
> does not get removed from `post-command-hook' either way. Which is very nice
> all around.

I'm not aware of such a change.


        Stefan

Re: trunk r117960: * etc/TODO: Add a few entries, remove others, expand some

[Dmitry Gutov]

On 10/05/2014 05:54 AM, Stefan Monnier wrote:

> I'm not aware of such a change.

I thought it might have been accidental.

Try this:

(defun foo-post-command ()
   (condition-case-unless-debug err
       (error "hello")
     (error nil)))

(add-hook 'post-command-hook 'foo-post-command)

With debug-on-error disabled, it does nothing. With it enabled, it'll 
pop up backtrace after each command.

Re: trunk r117960: * etc/TODO: Add a few entries, remove others, expand some

[Stefan Monnier]

> I thought it might have been accidental.

Yes, it's probably accidental.


        Stefan