Describing local-map/keymap in overlays/text properties

Describing local-map/keymap in overlays/text properties

[Lars Magne Ingebrigtsen]

When I hit `C-h b' to see the current key bindings, this apparently
doesn't include bindings from text properties/overlays, but only the
bindings done by the mode.  Is this the way it's supposed to be?  Or a
bug?  Is there a different command to see text property key bindings?

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

Re: Describing local-map/keymap in overlays/text properties

[Andreas Schwab]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> Is there a different command to see text property key bindings?

C-u C-x =

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: Describing local-map/keymap in overlays/text properties

[Lars Ingebrigtsen]

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

>> Is there a different command to see text property key bindings?
>
> C-u C-x =

That's kinda obscure.  Anybody mind if I alter `C-h m' and `C-h b' to
display text prop keymaps (first)?

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

Re: Describing local-map/keymap in overlays/text properties

[Eli Zaretskii]

> From: Lars Ingebrigtsen <larsi@gnus.org>
> Date: Sat, 06 Jul 2013 17:23:38 +0200
> Cc: emacs-devel@gnu.org
> 
> Anybody mind if I alter `C-h m' and `C-h b' to display text prop
> keymaps (first)?

There could be gazillions of them in a buffer.  Which ones will you
display, and won't that make the *Help* buffer, which is already too
large, even more so?

Re: Describing local-map/keymap in overlays/text properties

[Lars Ingebrigtsen]

Eli Zaretskii <eliz@gnu.org> writes:

> There could be gazillions of them in a buffer.

We'd only display the map under point.

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

Re: Describing local-map/keymap in overlays/text properties

[Eli Zaretskii]

> From: Lars Ingebrigtsen <larsi@gnus.org>
> Cc: schwab@linux-m68k.org,  emacs-devel@gnu.org
> Date: Sat, 06 Jul 2013 17:56:59 +0200
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> > There could be gazillions of them in a buffer.
> 
> We'd only display the map under point.

Then how is that different from "C-u C-x ="?

Re: Describing local-map/keymap in overlays/text properties

[Lars Ingebrigtsen]

Eli Zaretskii <eliz@gnu.org> writes:

> Then how is that different from "C-u C-x ="?

It's different in that people hit `C-h b' to learn what keystrokes are
available to them.  `C-u C-x =' is a relatively obscure command, and
having to use both commands for doing something as basic as this is
pretty nonsensical.

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

Re: Describing local-map/keymap in overlays/text properties

[Eli Zaretskii]

> From: Lars Ingebrigtsen <larsi@gnus.org>
> Cc: schwab@linux-m68k.org,  emacs-devel@gnu.org
> Date: Sat, 06 Jul 2013 18:13:17 +0200
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> > Then how is that different from "C-u C-x ="?
> 
> It's different in that people hit `C-h b' to learn what keystrokes are
> available to them.  `C-u C-x =' is a relatively obscure command, and
> having to use both commands for doing something as basic as this is
> pretty nonsensical.

Pushing the normal output of "C-h b" further down the buffer doesn't
make sense, either.  How about a new "C-h" command, or perhaps an
argument to "C-h b" that would trigger this new display?

Re: Describing local-map/keymap in overlays/text properties

[Lars Ingebrigtsen]

Eli Zaretskii <eliz@gnu.org> writes:

> Pushing the normal output of "C-h b" further down the buffer doesn't
> make sense, either.  How about a new "C-h" command, or perhaps an
> argument to "C-h b" that would trigger this new display?

If you're pushing `C-h b', you want to know what commands are available
for you.  That's the point of the command.  The commands that are
available in the local text properties are the ones that matter to you
most, because you moved the cursor there.

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

Re: Describing local-map/keymap in overlays/text properties

[Eli Zaretskii]

> From: Lars Ingebrigtsen <larsi@gnus.org>
> Cc: schwab@linux-m68k.org,  emacs-devel@gnu.org
> Date: Sat, 06 Jul 2013 18:31:20 +0200
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> > Pushing the normal output of "C-h b" further down the buffer doesn't
> > make sense, either.  How about a new "C-h" command, or perhaps an
> > argument to "C-h b" that would trigger this new display?
> 
> If you're pushing `C-h b', you want to know what commands are available
> for you.  That's the point of the command.  The commands that are
> available in the local text properties are the ones that matter to you
> most, because you moved the cursor there.

Would "C-u C-h b" be good enough?  I'm not sure it's right to show
this information by default.

Re: Describing local-map/keymap in overlays/text properties

[Lars Magne Ingebrigtsen]

After all this discussion, it turns out that there's already code for
outputting this data in `describe-buffer-bindings':

      start1 = get_local_map (BUF_PT (XBUFFER (buffer)),
			      XBUFFER (buffer), Qkeymap);
      if (!NILP (start1))
	{
	  describe_map_tree (start1, 1, shadow, prefix,
			     "\f\n`keymap' Property Bindings", nomenu,
			     0, 0, 0);
	  shadow = Fcons (start1, shadow);
	}

I only found this out after implementing just this again, in exactly the
same manner.  :-/

However!  It doesn't work.  get_local_map seems to return Qnil always
when fed Qkeymap.  And I don't quite understand why.  I must be
overlooking something pretty obvious again...

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

Re: Describing local-map/keymap in overlays/text properties

[Lars Magne Ingebrigtsen]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> However!  It doesn't work.  get_local_map seems to return Qnil always
> when fed Qkeymap.  And I don't quite understand why.  I must be
> overlooking something pretty obvious again...

D'oh.  Again.  The reason this doesn't work for me is that I've only
tested it in the Gnus article buffer, which does mighty voodoo to
emulate commands from the summary buffer.  And that gets in the way of
`C-h b' on text properties.

Never mind...  nothing to see here...

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

Re: Describing local-map/keymap in overlays/text properties

[Stephen Leake]

Lars Ingebrigtsen <larsi@gnus.org> writes:

> Eli Zaretskii <eliz@gnu.org> writes:
>
>> Pushing the normal output of "C-h b" further down the buffer doesn't
>> make sense, either.  How about a new "C-h" command, or perhaps an
>> argument to "C-h b" that would trigger this new display?
>
> If you're pushing `C-h b', you want to know what commands are available
> for you.  That's the point of the command.  The commands that are
> available in the local text properties are the ones that matter to you
> most, because you moved the cursor there.

+1

-- 
-- Stephe

RE: Describing local-map/keymap in overlays/text properties

[Drew Adams]

> > Pushing the normal output of "C-h b" further down the buffer doesn't
> > make sense, either.  How about a new "C-h" command, or perhaps an
> > argument to "C-h b" that would trigger this new display?
> 
> If you're pushing `C-h b', you want to know what commands are available
> for you.  That's the point of the command.  The commands that are
> available in the local text properties are the ones that matter to you
> most, because you moved the cursor there.

No, that's one use case of `C-h b', and yes, it ignores text-property
bindings.

If you make `C-h b' take into account `point' then it no longer shows
you binding information about the current context (buffer) in general.

If `a' is bound in the current buffer/mode to `self-insert-command',
but there is a text-property binding of `a' to `ignore', then `C-h b'
output will show only the latter.  You lose the general info.

Or if you choose to show both, and the fact that the text-property
bindings shadow other bindings, then why privilege text-property
bindings this way?  We don't do that in `C-h b' for other bindings
that override bindings.

`C-h b' does not currently tell you, for example, that `a' is bound
in the `global-map' to `self-insert-command' AND that it is bound in
the current local keymap to `foobar' AND that it is bound in the
currently active minor mode `toto' to `toto-go-fish'... AND give you
information about the various binding shadowings.  It simply gives
you the current effective binding of `a'.

Re: Describing local-map/keymap in overlays/text properties

[James Cloos]

>>>>> "LI" == Lars Ingebrigtsen <larsi@gnus.org> writes:

LI> Andreas Schwab <schwab@linux-m68k.org> writes:
>>> Is there a different command to see text property key bindings?
>> 
>> C-u C-x =

LI> That's kinda obscure.  Anybody mind if I alter `C-h m' and `C-h b' to
LI> display text prop keymaps (first)?

I wouldn't call C-u C-x = exotic, but when this came up for me the other
day, it never occurred to me that asking emacs about the character under
point would be a useful method to discover the bindings then in effect.

I tried C-h b.  I might have tried C-h m, except that when C-h b didn't
contain the info I needed, I presumed C-h m also wouldn't.

So please do update C-h m and C-h b to display text prop keymaps first.

(I'd also move Key translations last.)

-JimC
-- 
James Cloos <cloos@jhcloos.com>         OpenPGP: 1024D/ED7DAEA6

Re: Describing local-map/keymap in overlays/text properties

[Andreas Schwab]

James Cloos <cloos@jhcloos.com> writes:

> (I'd also move Key translations last.)

That would be wrong, since they have the highest priority.

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: Describing local-map/keymap in overlays/text properties

[James Cloos]

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

JC>> (I'd also move Key translations last.)

AS> That would be wrong, since they have the highest priority.

OK.  One'll just need to remember to next-page past it, rather
than always relying on scroll-up....

(Even after more than a quarter-century of Emacs C-x [ and C-x ] are not
integrated in to my muscle memory.)

-JimC
-- 
James Cloos <cloos@jhcloos.com>         OpenPGP: 1024D/ED7DAEA6

RE: Describing local-map/keymap in overlays/text properties

[Drew Adams]

> >> Is there a different command to see text property key bindings?
> >
> > C-u C-x =
> 
> That's kinda obscure.  Anybody mind if I alter `C-h m' and `C-h b'
> to display text prop keymaps (first)?

Yes, I mind.  (But please submit an enhancement request via
`report-emacs-bug', if you like.)

`C-h m' is already very long, even for a short major-mode doc string.
This does not belong there at all, IMO.

`C-h b' is another story, but it refers to the current context (e.g.
buffer) generally, not only to a specific position.

(I assume you are speaking generally, and not just about the doc
for some particular mode.)

Better than messing with `C-h m' or `C-h b' might be to provide a
new command that lists only the text-property bindings at point.

(But that could also be said for other parts of the `C-u C-x =' output:
it is a mix of different kinds of info.  One could imagine several
separate commands for the different kinds of info.)

My general reaction to your proposal is YAGNI.  `C-u C-x =' is fine.