Suggestion: Mapping of M-g should be goto-line

Suggestion: Mapping of M-g should be goto-line

[Jari Aalto+mail.emacs]

        M-g in current 21.3 is mapped to binding "set face default". 

        First, it is not possible to obtain any help information with
        standard binding to see what it actually does:

                C-h k M-g


        Second, it is unlikely that new Emacs user, or even
        experienced one seldom uses this keybinding. I have yet to
        find use for this functionality (= never used in my Emacs
        lifetime).

        There may be few users that may potentially find the current
        binding useful in some cases. But I hardly think those few
        users needs should be put first, when considering the
        usefulness of the current binding in large.

        If put side by side, "set face default" and goto-line, I would
        estimate that the 99 % of the cases, there is more need for
        goto-line. Therefore it would be better if the M-g was mapped
        to it. 

        Users need line information. After all, Emacs for most, is
        programming language development environment: C / C++, Python,
        Perl, Ruby etc. In time they may find and start use other
        features like Gnus, Mail and others.

        The "de facto" situation for long has been that everybody maps
        M-g to goto-line. People advice that in newsgroups, because it
        is indeed the most logical key.

        If we would take a poll in Emacs newsgroups, presumably the
        "yes" votes for goto-line would win in great majority.

        It would be better if Emacs supported this 99 % daily usage of
        M-g, instead of current not-so-useful binding.
        
        Jari

Re: Suggestion: Mapping of M-g should be goto-line

[Eli Zaretskii]

> From: jari.aalto@poboxes.com (Jari Aalto+mail.emacs)
> Date: Thu, 25 Mar 2004 08:11:24 +0200
> 
>         Second, it is unlikely that new Emacs user, or even
>         experienced one seldom uses this keybinding. I have yet to
>         find use for this functionality (= never used in my Emacs
>         lifetime).

My $0.02: I use it whenever I use Enriched Text mode.

Re: Suggestion: Mapping of M-g should be goto-line

[John Wiegley]

Eli Zaretskii <eliz@elta.co.il> writes:

>>         Second, it is unlikely that new Emacs user, or even
>>         experienced one seldom uses this keybinding. I have yet to
>>         find use for this functionality (= never used in my Emacs
>>         lifetime).
>
> My $0.02: I use it whenever I use Enriched Text mode.

I second the idea that M-g should bind to goto-line.  My guess is that
99% of users have no clue what M-g currently does.  Those using
enriched text mode could always rebind.

John

Re: Suggestion: Mapping of M-g should be goto-line

[Eric Hanchrow]

I have had bound M-g to goto-line in my .emacs for months, if not years.
-- 
Documentation exists, but never seems to be complete.
        -- Dennis Ritchie, 1972

Re: Suggestion: Mapping of M-g should be goto-line

[ams]

Since this is getting to a vote, I'll pitch in with my 0.02 groszy.
Never used goto-line (don't see why one would use it other then in
really obscure situations), use M-g far more often that that.

But on another side, the idea by Kim Storm (I think) of adding a set
of bindings for recompile, compile, and next-error wasn't a bad one.
Those are things I always wondered why they aren't bound by default.

Cheers.

Re: Suggestion: Mapping of M-g should be goto-line

[Stefan Monnier]

> But on another side, the idea by Kim Storm (I think) of adding a set
> of bindings for recompile, compile, and next-error wasn't a bad one.
> Those are things I always wondered why they aren't bound by default.

For what it's worth I always thought C-c C-c was the natural binding for
`compile' (or more generally for "here, I'm done editing, now process what
I've edited").


        Stefan

Re: Suggestion: Mapping of M-g should be goto-line

[David Kastrup]

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

> > But on another side, the idea by Kim Storm (I think) of adding a set
> > of bindings for recompile, compile, and next-error wasn't a bad one.
> > Those are things I always wondered why they aren't bound by default.
> 
> For what it's worth I always thought C-c C-c was the natural binding
> for `compile' (or more generally for "here, I'm done editing, now
> process what I've edited").

C-c C-c, however, is used by _many_ modes to mean "do the most common
thing", including modes where you want to be able to call "compile".

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: Suggestion: Mapping of M-g should be goto-line

[Stefan Monnier]

>> For what it's worth I always thought C-c C-c was the natural binding
>> for `compile' (or more generally for "here, I'm done editing, now
>> process what I've edited").

> C-c C-c, however, is used by _many_ modes to mean "do the most common
> thing", including modes where you want to be able to call "compile".

That might be, but I haven't come across such conflicts in my short life.
What conflicts did you encounter?


        Stefan

Re: Suggestion: Mapping of M-g should be goto-line

[David Kastrup]

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

> >> For what it's worth I always thought C-c C-c was the natural binding
> >> for `compile' (or more generally for "here, I'm done editing, now
> >> process what I've edited").
> 
> > C-c C-c, however, is used by _many_ modes to mean "do the most common
> > thing", including modes where you want to be able to call "compile".
> 
> That might be, but I haven't come across such conflicts in my short life.
> What conflicts did you encounter?

How about AUCTeX?  C-c C-c is the central dispatcher, and you might
want to use compile for running make-based compilations.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: Suggestion: Mapping of M-g should be goto-line

[Stefan Monnier]

>> >> For what it's worth I always thought C-c C-c was the natural binding
>> >> for `compile' (or more generally for "here, I'm done editing, now
>> >> process what I've edited").
>> 
>> > C-c C-c, however, is used by _many_ modes to mean "do the most common
>> > thing", including modes where you want to be able to call "compile".
>> 
>> That might be, but I haven't come across such conflicts in my short life.
>> What conflicts did you encounter?

> How about AUCTeX?  C-c C-c is the central dispatcher, and you might
> want to use compile for running make-based compilations.

Isn't AUCTeX's central dispatcher supposed to be a "better compile"
so that you don't need compile for those cases?  At least I can't remember
wishing to run `compile' when I was using AUCTeX.  And it can't be that
hard to add `make' to the list of AUCTeX actions, can it?


        Stefan

Re: Suggestion: Mapping of M-g should be goto-line

[David Kastrup]

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

> >> >> For what it's worth I always thought C-c C-c was the natural binding
> >> >> for `compile' (or more generally for "here, I'm done editing, now
> >> >> process what I've edited").
> >> 
> >> > C-c C-c, however, is used by _many_ modes to mean "do the most common
> >> > thing", including modes where you want to be able to call "compile".
> >> 
> >> That might be, but I haven't come across such conflicts in my short life.
> >> What conflicts did you encounter?
> 
> > How about AUCTeX?  C-c C-c is the central dispatcher, and you
> > might want to use compile for running make-based compilations.
> 
> Isn't AUCTeX's central dispatcher supposed to be a "better compile"
> so that you don't need compile for those cases?

Better?  It substitutes for most of it, yes, but if you have
processes like weaving a noweb file or other autogenerated stuff,
AUCTeX can be a bit tedious.  For example, generating index and
glossary and so on often is done by Makefiles in more complicated
projects.

> At least I can't remember wishing to run `compile' when I was using
> AUCTeX.  And it can't be that hard to add `make' to the list of
> AUCTeX actions, can it?

Well, Gnus sends mail and articles with C-c C-c, calc finished
editing, PCL-vcs aborts a job, most shell modes send an interrupt
(don't tell me you never want to use compile from a shell), and so
on.  It's not exactly the least used key combination.

In fact, I am now sending this mail with it.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: Suggestion: Mapping of M-g should be goto-line

[Stefan Monnier]

>> >> >> For what it's worth I always thought C-c C-c was the natural binding
>> >> >> for `compile' (or more generally for "here, I'm done editing, now
>> >> >> process what I've edited").
[...]
>> Isn't AUCTeX's central dispatcher supposed to be a "better compile"
>> so that you don't need compile for those cases?

> Better?  It substitutes for most of it, yes, but if you have
> processes like weaving a noweb file or other autogenerated stuff,
> AUCTeX can be a bit tedious.  For example, generating index and
> glossary and so on often is done by Makefiles in more complicated
> projects.

But people who need that can do M-x compile RET, right?
After all that's what they do already.

> Well, Gnus sends mail and articles with C-c C-c,

Yes, exactly what I said "process what I've edited".  I've never felt the
need to compile an email.

> calc finished editing,

Do you mean it just quits with C-c C-c?  Or does it take the result of
your editing and processes it?

> PCL-vcs aborts a job,

That was maybe a poor choice.

> most shell modes send an interrupt (don't tell me you never want to use
> compile from a shell), and so on.

I never want to use compile from a shell, to tell you the truth.
Why would you ever want to?

> It's not exactly the least used key combination.

Indeed and it often means "process what Ive just edited", which in an email
means "send it" and in a C buffer means "compile it".

C-c C-c is currently globally unbound and I suggest we bind it to `compile'.
Major modes would be encouraged to override it with mode-specific
implementations of the idea of "process what I've just edited", like AUCTeX
and message already do.


        Stefan

Re: Suggestion: Mapping of M-g should be goto-line

[Kim F. Storm]

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

> C-c C-c is currently globally unbound and I suggest we bind it to `compile'.

Try
        M-x recursive-edit RET C-h k C-c C-c
=>
        C-c C-c runs the command exit-recursive-edit

Quite a number of commands use recursive-edit, and it doesn't make me
feel comfortable to have two important "global" bindings on the same
key sequence.

> Major modes would be encouraged to override it with mode-specific
> implementations of the idea of "process what I've just edited", like AUCTeX
> and message already do.

For most of the existing keybindings it seems to me that C-c C-c means
"I am done with this, finalize it and pass it on, and we are both done
with it".

You can make compile fit with that definition also, but in my mind
compile is more like a "start doing something" than finalize
something.  

But I see your point, and maybe I just need change a brain pattern to
get used to this.  Actually, the more I think about it, the more I
tend to agree with you!  (That's a 180 degree turn in 10 lines :-)

--
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: Suggestion: Mapping of M-g should be goto-line

[Juanma Barranquero]

On 26 Mar 2004 12:14:49 +0100
storm@cua.dk (Kim F. Storm) wrote:

> Try
>         M-x recursive-edit RET C-h k C-c C-c
> =>
>         C-c C-c runs the command exit-recursive-edit

Uh?

emacs -q --no-site-file
M-x emacs-version => "GNU Emacs 21.3.50.1 (i386-msvc-nt5.0.2195) of 2004-03-26 on JMBARRANQUERO"
M-x recursive-edit RET C-h k C-c C-c => "C-c C-c is undefined"
C-h w exit-recursive-edit RET => "exit-recursive-edit is on C-M-c"

                                                                Juanma

Re: Suggestion: Mapping of M-g should be goto-line

[Kim F. Storm]

Juanma Barranquero <jmbarranquero@wke.es> writes:

> On 26 Mar 2004 12:14:49 +0100
> storm@cua.dk (Kim F. Storm) wrote:
> 
> > Try
> >         M-x recursive-edit RET C-h k C-c C-c
> > =>
> >         C-c C-c runs the command exit-recursive-edit
> 
> Uh?
> 
> emacs -q --no-site-file
> M-x emacs-version => "GNU Emacs 21.3.50.1 (i386-msvc-nt5.0.2195) of 2004-03-26 on JMBARRANQUERO"
> M-x recursive-edit RET C-h k C-c C-c => "C-c C-c is undefined"
> C-h w exit-recursive-edit RET => "exit-recursive-edit is on C-M-c"

Wauw...  I stand corrected :-)

I wish I had version control on my .emacs file  -- I've used that binding for eons.

In any case, I find it is a logical binding (for the same reasons as
other modes uses C-c C-c to "exit").

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: Suggestion: Mapping of M-g should be goto-line

[Juanma Barranquero]

On 26 Mar 2004 13:38:01 +0100
storm@cua.dk (Kim F. Storm) wrote:

> I wish I had version control on my .emacs file  -- I've used that binding for eons.

I do, under Subversion of course.

<aside class="sigh">
  Funny thing is, that's *exactly* a situation where arc would be better,
  as I almost dayly have to synchronize by hand my conf repository at
  home with the one at work...
</aside>

> In any case, I find it is a logical binding (for the same reasons as
> other modes uses C-c C-c to "exit").

I agree.

                                                                Juanma

Re: Suggestion: Mapping of M-g should be goto-line

[Stefan Monnier]

>> > Try
>> >         M-x recursive-edit RET C-h k C-c C-c
>> > =>
>> >         C-c C-c runs the command exit-recursive-edit

Exactly my point: yet another example of C-c C-c used for "I'm done editing,
now process it".  In the context of recursive-edit, of course this should
not be `compile'.

It seems people don't understand what I really want to say, so here's
another take on it:

1 - let's create a new command

   (defvar done-editing-now-do-it-function 'compile)
   (defun done-editing-now-do-it ()
     "Do what needs to be done with what you've just finished editing."
     (interactive)
     (call-interactively done-editing-now-do-it-function))
   (global-set-key "\C-c\C-c" 'done-editing-now-do-it)

2 - let's optimize it away:

   (global-set-key "\C-c\C-c" 'compile)

    Notice it's only an optimization that relies on the fact that the
    binding can trivially be overridden by local maps and on the fact that
    done-editing-now-do-it did not do anything more than call the
    done-editing-now-do-it-function function.  The intention is still that
    major modes (or recursive-edit or minor-modes or whatever else shows up)
    should rebind it to something more appropriate if applicable.
    In elisp-mode, I've rebound it to byte-compile-file.

I.e. this is not "a standard binding for `compile'", but "a formalization
of what C-c C-c is expected to do".


	Stefan

Re: Suggestion: Mapping of M-g should be goto-line

[Miles Bader]

On Fri, Mar 26, 2004 at 09:25:49AM -0500, Stefan Monnier wrote:
> It seems people don't understand what I really want to say, so here's
> another take on it:
> 
> 1 - let's create a new command
>
>    (global-set-key "\C-c\C-c" 'done-editing-now-do-it)
> 
> 2 - let's optimize it away:
> 
>    (global-set-key "\C-c\C-c" 'compile)
> 
>     Notice it's only an optimization that relies on the fact that the
>     binding can trivially be overridden by local maps and on the fact that
>     done-editing-now-do-it did not do anything more than call the
>     done-editing-now-do-it-function function.  The intention is still that
>     major modes (or recursive-edit or minor-modes or whatever else shows up)
>     should rebind it to something more appropriate if applicable.

I intensely dislike this sort of `ah modes can just rebind the key' way of
avoiding a level of indirection, because often I use different global
bindings that normal -- and in that case, with a variable used for
configuring things, everything works great, but if modes `override by
rebinding', then I end up having to have my own personal special cases for
every single mode that wants to do this.

A good example is `fill-paragraph-function': I _love_ this, because I use a
non-standard binding for `fill-paragaraph'; in the bad-old-days, when many
modes with paragraph-filling support would just rebind `M-q' (or whatever the
standard binding for fill-paragraph is -- I can't remember!), it was a
nightmare, but now that most instead just set fill-paragraph-function, I have
no problems at all, everything works swimmingly.

[I've always wished for a level of indirection _built into keymaps_ --
e.g. standard global bindings could be named, and then locally overridden by
name in modes.]

-Miles
-- 
Freedom's just another word, for nothing left to lose   --Janis Joplin

Re: Suggestion: Mapping of M-g should be goto-line

[Juanma Barranquero]

On Fri, 26 Mar 2004 19:07:52 -0500, Miles Bader <miles@gnu.org> wrote:

> I intensely dislike this sort of `ah modes can just rebind the key' way of
> avoiding a level of indirection, because often I use different global
> bindings that normal

<snip>

> A good example is `fill-paragraph-function': I _love_ this, because I use a
> non-standard binding for `fill-paragaraph';

I'm on your camp here regarding having explicit levels of indirection.

Still, I don't find your example compelling at all, because the given
examples are not equivalent. I can imagine a (more or less) all-purpose
`fill-paragraph', but it's less easy to imagine an all-purpose
`done-editing-now-do-it' (which, in the thread, has been shown as
standing for: compile, send message, commit change to VC, exit recursive
edit...).

In fact, I'm not convinced by the idea of generalizing C-c C-c, because
the two use cases, as pointed out before, are quite different in my view:
"done for now, so do external action and come back" (compile), and "done
for good, so do what you've got to do and let's get outta here"
(log-edit-done, for example).

                                                           /L/e/k/t/u

Re: Suggestion: Mapping of M-g should be goto-line

[Miles Bader]

Juanma Barranquero <lektu@mi.madritel.es> writes:
> > I intensely dislike this sort of `ah modes can just rebind the key' way of
> > avoiding a level of indirection, because often I use different global
> > bindings that normal
> 
> I'm on your camp here regarding having explicit levels of indirection.
> 
> Still, I don't find your example compelling at all, because the given
> examples are not equivalent. I can imagine a (more or less) all-purpose
> `fill-paragraph', but it's less easy to imagine an all-purpose
> `done-editing-now-do-it'

You may very well be right in this case; I was mostly just trying to say
that in the case where there _is_ a generic function that occasionally
is overridden by modes &c, rebinding of the default key-sequence is not
a good override mechanism.

-Miles
-- 
`...the Soviet Union was sliding in to an economic collapse so comprehensive
 that in the end its factories produced not goods but bads: finished products
 less valuable than the raw materials they were made from.'  [The Economist]

Re: Suggestion: Mapping of M-g should be goto-line

[Richard Stallman]

    [I've always wished for a level of indirection _built into keymaps_ --
    e.g. standard global bindings could be named, and then locally overridden by
    name in modes.]

We have a feature like this now: a keymap can redirect one command
to another.

Re: Suggestion: Mapping of M-g should be goto-line

[Per Abrahamsen]

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

> C-c C-c is currently globally unbound and I suggest we bind it to `compile'.

The would violate the separation of keyspace between major modes and
Emacs.  I would support changing the the major mode binding in cc-mode
to compile.  It is currently bound to "comment-region", which is a
waste, since comment-region has a global binding (M-;).

Other modes for language with a cultural tradition for using make
should make a similar binding.

Re: Suggestion: Mapping of M-g should be goto-line

[Alan Mackenzie]

On Fri, 26 Mar 2004, Per Abrahamsen wrote:

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

>> C-c C-c is currently globally unbound and I suggest we bind it to `compile'.

>The would violate the separation of keyspace between major modes and
>Emacs.  I would support changing the the major mode binding in cc-mode
>to compile.  It is currently bound to "comment-region", which is a
>waste, since comment-region has a global binding (M-;).

No it hasn't.  M-; is globally bound to `comment-dwim', an almost, but
not quite, entirely unlike operation, except when transient-mark minor
mode is enabled.  In CC Mode, C-c C-c is bound to `comment-region'.  

`compile' does not belong by default in a language mode's key binding
space, because it is not part of manipulating the text of that language.
If anything, `compile' should be in the global map, and tailored to each
mode where it is useful.  In my not so humble opinion, of course.  ;-)

Surely in this discussion, CC Mode should have priority over the use of
this binding because of its name.  :-)

-- 
Alan Mackenzie (Munich, Germany)
acm@muc.de

Re: Suggestion: Mapping of M-g should be goto-line

[Per Abrahamsen]

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

> Isn't AUCTeX's central dispatcher supposed to be a "better compile"

Yes it was.  Or rather, it was supposed to be the mode specific
version of compile.  It even uses compile for some subtasks.

I originally chose C-c C-c for that task in AUC TeX because that was
the binding I thought of as the standard for compile.

> And it can't be that hard to add `make' to the list of AUCTeX
> actions, can it?

It already is there.

Re: Suggestion: Mapping of M-g should be goto-line

[David Kastrup]

Per Abrahamsen <abraham@dina.kvl.dk> writes:

> Stefan Monnier <monnier@iro.umontreal.ca> writes:
> 
> > Isn't AUCTeX's central dispatcher supposed to be a "better compile"
> 
> Yes it was.  Or rather, it was supposed to be the mode specific
> version of compile.  It even uses compile for some subtasks.
> 
> I originally chose C-c C-c for that task in AUC TeX because that was
> the binding I thought of as the standard for compile.
> 
> > And it can't be that hard to add `make' to the list of AUCTeX
> > actions, can it?
> 
> It already is there.

Hmm.  Not in my version of AUCTeX, and that has been updated from the
CVS repository about 10 minutes ago.  And considering that I am the
current maintainer of AUCTeX, I am pretty sure that it is not checked
into a branch.

Tell me more.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: Suggestion: Mapping of M-g should be goto-line

[Per Abrahamsen]

David Kastrup <dak@gnu.org> writes:

> Per Abrahamsen <abraham@dina.kvl.dk> writes:
>
>> Stefan Monnier <monnier@iro.umontreal.ca> writes:
>> 
>> > And it can't be that hard to add `make' to the list of AUCTeX
>> > actions, can it?
>> 
>> It already is there.
>
> Hmm.  Not in my version of AUCTeX, and that has been updated from the
> CVS repository about 10 minutes ago.  And considering that I am the
> current maintainer of AUCTeX, I am pretty sure that it is not checked
> into a branch.
>
> Tell me more.

It is "TeX-run-compile", which is one of the customize options for the
"How" field of "TeX-command-list". 

Re: Suggestion: Mapping of M-g should be goto-line

[David Kastrup]

Per Abrahamsen <abraham@dina.kvl.dk> writes:

> David Kastrup <dak@gnu.org> writes:
> 
> > Per Abrahamsen <abraham@dina.kvl.dk> writes:
> >
> >> Stefan Monnier <monnier@iro.umontreal.ca> writes:
> >> 
> >> > And it can't be that hard to add `make' to the list of AUCTeX
> >> > actions, can it?
> >> 
> >> It already is there.
> >
> > Hmm.  Not in my version of AUCTeX, and that has been updated from the
> > CVS repository about 10 minutes ago.  And considering that I am the
> > current maintainer of AUCTeX, I am pretty sure that it is not checked
> > into a branch.
> >
> > Tell me more.
> 
> It is "TeX-run-compile", which is one of the customize options for the
> "How" field of "TeX-command-list". 

Pretty inconspicuous, I'd say.  Having some support function that can
be customized as a component for a "make" Command one could, if one
wanted to, customize into some variable is not exactly the same as
"it already is there" in my book, but then I am a wimp.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: Suggestion: Mapping of M-g should be goto-line

[Stefan Monnier]

>> Second, it is unlikely that new Emacs user, or even
>> experienced one seldom uses this keybinding. I have yet to
>> find use for this functionality (= never used in my Emacs
>> lifetime).

> My $0.02: I use it whenever I use Enriched Text mode.

So you mean that the current M-g binding should be moved from global-map to
enriched-mode-map ?  Agreed.


        Stefan

Re: Suggestion: Mapping of M-g should be goto-line

[Kim F. Storm]

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

> >> Second, it is unlikely that new Emacs user, or even
> >> experienced one seldom uses this keybinding. I have yet to
> >> find use for this functionality (= never used in my Emacs
> >> lifetime).
> 
> > My $0.02: I use it whenever I use Enriched Text mode.
> 
> So you mean that the current M-g binding should be moved from global-map to
> enriched-mode-map ?  Agreed.

The current M-g bindings are:

M-g d		facemenu-set-default
M-g b		facemenu-set-bold
M-g i		facemenu-set-italic
M-g l		facemenu-set-bold-italic
M-g u		facemenu-set-underline
M-g o		facemenu-set-face

What if enriched mode had these (equally logical) bindings instead:

C-c C-d		facemenu-set-default
C-c C-b		facemenu-set-bold
C-c C-i		facemenu-set-italic
C-c C-l		facemenu-set-bold-italic
C-c C-u		facemenu-set-underline
C-c C-o		facemenu-set-face

(and C-c d, C-c b, etc as well).

If other modes would like to use these bindings as well, it could be
a separate minor mode, e.g. enriched-face-mode, that could be enabled
by enriched mode by default.

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: Suggestion: Mapping of M-g should be goto-line

[Alan Mackenzie]

On 25 Mar 2004, Stefan Monnier wrote:

>>> Second, it is unlikely that new Emacs user, or even experienced one
>>> seldom uses this keybinding. I have yet to find use for this
>>> functionality (= never used in my Emacs lifetime).

>> My $0.02: I use it whenever I use Enriched Text mode.

>So you mean that the current M-g binding should be moved from global-map to
>enriched-mode-map ?  Agreed.

Please don't!  Maybe I'm one of the few who _does_ use M-g d regularly
(and occasionally M-g M-g).

I keep a log of my hacking activities (tends to be in fundamental-mode
most of the time ;-), and I frequently copy text from an Elisp buffer to
my log.  Irritatingly, the font-locking gets copied along with it, so I'm
grateful for M-g d.

However, I too think that goto-line should have a binding, and M-g is a
good binding for it.  How about moving `facemenu-keymap' to C-M-g?  C-M-g
is currently unused, I think.

>        Stefan

-- 
Alan Mackenzie (Munich, Germany)
acm@muc.de

Re: Suggestion: Mapping of M-g should be goto-line

[Danilo Segan]

Today at 7:11, Jari Aalto wrote:

>         If put side by side, "set face default" and goto-line, I would
>         estimate that the 99 % of the cases, there is more need for
>         goto-line. Therefore it would be better if the M-g was mapped
>         to it. 

Emacs tries hard to make goto-line unnecessary, AFAICT.  Can you
provide an example of when goto-line couldn't be replaced with a
better functionality (such as using M-x next-error)? 

>         Users need line information. After all, Emacs for most, is
>         programming language development environment: C / C++, Python,
>         Perl, Ruby etc. In time they may find and start use other
>         features like Gnus, Mail and others.

Exactly, and that's why the next-error and friends are so useful — I
need not worry about typing the exact line number, because I do not
care about the exact line number.  I care about the place with
certain code/error/whatever, and that's where I want to go.  Emacs
has the context available, so better make use of it.

FWIW, C-x ` is bound to next-error, and it's a big win in situations
like this.

>         The "de facto" situation for long has been that everybody maps
>         M-g to goto-line. People advice that in newsgroups, because it
>         is indeed the most logical key.
>
>         If we would take a poll in Emacs newsgroups, presumably the
>         "yes" votes for goto-line would win in great majority.

Perhaps, but that doesn't mean that it's a right thing to do.  
I seldom use current M-g binding as well, but not less than M-x
goto-line. 

Cheers,
Danilo

Re: Suggestion: Mapping of M-g should be goto-line

[Lucas]

Danilo Segan <dsegan@gmx.net> writes:

>
> FWIW, C-x ` is bound to next-error, and it's a big win in situations
> like this.
>

And, ` is on AltGR 7 here, so basically absolutely unusable on French
keyboard layouts. Of course next-error is bound to something else

But anyway, going to a specified line is still useful in some
cases. Like : "I found a bug on line 25 and 26 ...". Unless you could
provide a function to read Gnus mails, and open the specified file to
the desired line ;).


-- 
Lucas

Re: Suggestion: Mapping of M-g should be goto-line

[Juanma Barranquero]

On Thu, 25 Mar 2004 10:46:07 +0100
Danilo Segan <dsegan@gmx.net> wrote:

> Can you
> provide an example of when goto-line couldn't be replaced with a
> better functionality (such as using M-x next-error)? 

Some people, like me, doesn't do everything inside Emacs. I rarely, if
ever, use compile.el (I prefer to compile outside Emacs and search for
bugs the old way). Also, I compare sources with a visual tool and then
go to Emacs to edit, if needed. In these cases, M-x goto-line is very
useful (in fact, I bound it to H-S-left long time ago).

Also, I often process large files with Perl, get errors, and I have to
edit the file by hand, fix some formatting issue, and reprocess. M-x
goto-line is very handy for that.

> FWIW, C-x ` is bound to next-error, and it's a big win in situations
> like this.

C-x ` SPC (four keystrokes) is not comfortable to me.

> I seldom use current M-g binding as well, but not less than M-x
> goto-line. 

Jari's saying he's got the feeling that you're in the minority here.
FWIW, I'd agree. I'm certainly in the "use goto-line every day, *never*
had to use 'set face default'" camp.

                                                                Juanma

Re: Suggestion: Mapping of M-g should be goto-line

[Jari Aalto+mail.linux]

* Thu 2004-03-25 Danilo Segan <dsegan <AT> gmx.net> mail.default.spool
| Today at 7:11, Jari Aalto wrote:
| 
| >         If put side by side, "set face default" and goto-line, I would
| >         estimate that the 99 % of the cases, there is more need for
| >         goto-line. Therefore it would be better if the M-g was mapped
| >         to it. 
| 
| Emacs tries hard to make goto-line unnecessary, AFAICT.  Can you
| provide an example of when goto-line couldn't be replaced with a
| better functionality (such as using M-x next-error)? 

Only if user knows Emacs. He doesn't when he is a beginner. And even
long period of Emacs experience, people still run many, many programs
outside of emacs in their terminals.

I have 20-40 terminals open to run and monitor various
tasks. Including command 'grep' and I use it with -n to get line numbers.

It's different to have 20 terminals open in all correct directory
locations when you have wealth of bash scripts and aliases
available. Switching to Emacs to typing M-x grep + long directory names,
which I might write wrong is not what people want to do. They might, but
the terminal usually wins (think about readline macros that can be used to
manipulate string input from previous commands)-

Emacs is no substitute for all system tasks. It is a great companion,
but I believe only handful of people do _all_ from inside Emacs and
not ever open a single terminal.

So, line numbers come from everywhere. You might be running 3-5 server
screen, shared samba mounts, then you see error in next screen and want
to view it in Emacs that contains the log file or configuration file ...
the needs are many.
 
| >         Users need line information. After all, Emacs for most, is
| >         programming language development environment: C / C++, Python,
| >         Perl, Ruby etc. In time they may find and start use other
| >         features like Gnus, Mail and others.
| 
| Exactly, and that's why the next-error and friends are so useful — I
| need not worry about typing the exact line number, because I do not
| care about the exact line number.  I care about the place with
| certain code/error/whatever, and that's where I want to go.  Emacs
| has the context available, so better make use of it.
| 
| FWIW, C-x ` is bound to next-error, and it's a big win in situations
| like this.

Please remember that Emacs can't do for every programming language.
PHP displays errors in Web page, but your code is in Emacs.

        ... and you would need M-g (goto-line) all the time.

Other Web-based programming languages reports problems in the page,
not to Emacs.

| >         The "de facto" situation for long has been that everybody maps
| >         M-g to goto-line. People advice that in newsgroups, because it
| >         is indeed the most logical key.
| >
| >         If we would take a poll in Emacs newsgroups, presumably the
| >         "yes" votes for goto-line would win in great majority.
| 
| Perhaps, but that doesn't mean that it's a right thing to do.  
| I seldom use current M-g binding as well, but not less than M-x
| goto-line. 

The right thing for M-g is exactly that. To move it to a sensible
function, and goto-line has been it for more years than I can
remember.

Let's think about the "average joe" how he would see it.

Jari

-- 
http://tiny-tools.sourceforge.net/
Swatch @time   http://www.mir.com.my/iTime/itime.htm
               http://www.ryanthiessen.com/swatch/resources.htm
Use Licenses!  http://www.linuxjournal.com/article.php?sid=6225
Which Licence? http://www.linuxjournal.com/article.php?sid=4825
OSI Licences   http://www.opensource.org/licenses/

Re: Suggestion: Mapping of M-g should be goto-line

[Danilo Segan]

Hi Jari,  Lucas, Juanma,

Today at 11:32, Jari Aalto wrote:

> * Thu 2004-03-25 Danilo Segan <dsegan <AT> gmx.net> mail.default.spool
> | Today at 7:11, Jari Aalto wrote:
> | 
> | >         If put side by side, "set face default" and goto-line, I would
> | >         estimate that the 99 % of the cases, there is more need for
> | >         goto-line. Therefore it would be better if the M-g was mapped
> | >         to it. 
> | 
> | Emacs tries hard to make goto-line unnecessary, AFAICT.  Can you
> | provide an example of when goto-line couldn't be replaced with a
> | better functionality (such as using M-x next-error)? 
>
> Only if user knows Emacs. He doesn't when he is a beginner. And even
> long period of Emacs experience, people still run many, many programs
> outside of emacs in their terminals.

That's only a reason to educate them better, not to adjust behaviour.
Emacs is not too friendly to beginners at all -- they'd need to
relearn a lot of things, no matter what editor they used before (if
it wasn't Emacs).  This is only one minor point in all the effort
they'll need to expend.

At the same time, Lucas writes:
> >
> > FWIW, C-x ` is bound to next-error, and it's a big win in situations
> > like this.
> 
> And, ` is on AltGR 7 here, so basically absolutely unusable on French
> keyboard layouts. Of course next-error is bound to something else

Which would only go so far to recommend assigning M-g to next-error,
perhaps? ;)


Also, Jari mentioned this example:

> Emacs is no substitute for all system tasks. It is a great companion,
> but I believe only handful of people do _all_ from inside Emacs and
> not ever open a single terminal.
>
> So, line numbers come from everywhere. You might be running 3-5 server
> screen, shared samba mounts, then you see error in next screen and want
> to view it in Emacs that contains the log file or configuration file ...
> the needs are many.

...which complements what Juanma Barranquero wrote:

> Some people, like me, doesn't do everything inside Emacs. I rarely, if
> ever, use compile.el (I prefer to compile outside Emacs and search for
> bugs the old way). Also, I compare sources with a visual tool and then
> go to Emacs to edit, if needed. In these cases, M-x goto-line is very
> useful (in fact, I bound it to H-S-left long time ago).

I suggest you try "emacsclient -n +5 path/to/file" (of course, I've
got the alias "ec" for "emacsclient").  If you're not working in
Emacs all the time, this is a major gain -- you need not switch
between windows or even workspaces while entering a line number, and
you can look at it while you're typing it (with lots of numbers on the
screen, I find it much easier to mechanically type, instead of trying
to remember a number and enter it in Emacs directly).

This way is much better -- terminal already has the right context
(i.e. more suitable cwd), so you need not search the filename by full
path in Emacs, or switch between buffers -- you've got all the data
you need at your sight. (Of course, if the file is already opened in
a buffer, emacsclient simply Does The Right Thing.)

With all this, I rarely if ever need to use M-x goto-line.  For those
that are not willing to go the "better" path (of course, if it becomes
a consensus that this is actually "better", I'm not insisting this
is), they can always customize their keybindings. 

For "average joe", we want to make them learn the better way.

With the examples you mentioned, it seems that you haven't made use
of emacsclient so far; I recommend it, since it's been my $EDITOR for
a long time. ;)

[There's one catch to all this though: (server-start), required by
emacsclient, is not run by default.]

Cheers,
Danilo

Re: Suggestion: Mapping of M-g should be goto-line

[Lucas]

Danilo Segan <dsegan@gmx.net> writes:

> Hi Jari,  Lucas, Juanma,

Hello

>
> With the examples you mentioned, it seems that you haven't made use
> of emacsclient so far; I recommend it, since it's been my $EDITOR for
> a long time. ;)

I use it, but when i'm already in Emacs, that's not convenient. Well,
anyway M-g is bound on goto-line here. That was just an attempt to
defend the goto-line fans :)


-- 
Lucas

Re: Suggestion: Mapping of M-g should be goto-line

[Vinicius Jose Latorre]

> Danilo Segan <dsegan@gmx.net> writes:
> 
> > Hi Jari,  Lucas, Juanma,
> 
> Hello
> 
> > With the examples you mentioned, it seems that you haven't made use
> > of emacsclient so far; I recommend it, since it's been my $EDITOR for
> > a long time. ;)
> 
> I use it, but when i'm already in Emacs, that's not convenient. Well,
> anyway M-g is bound on goto-line here. That was just an attempt to
> defend the goto-line fans :)

Well, and if it's used M-g M-o or M-g M-l instead of M-g ??

Here in my job we use C-c g.


Vinicius

Re: Suggestion: Mapping of M-g should be goto-line

[Juanma Barranquero]

On Thu, 25 Mar 2004 12:23:34 +0100
Danilo Segan <dsegan@gmx.net> wrote:

> That's only a reason to educate them better, not to adjust behaviour.

Many people don't have the time to be "educated". I've introduced a few
friends to Emacs and they're using it happily, but they just won't spend
the time to learn features they feel they're not going to need.

> I suggest you try "emacsclient -n +5 path/to/file"

I don't use emacsclient, but I *do* use gnuclient.exe and gnuserv.el
(I'm on Windows), which have the same functionality. Still, I routinely
use M-x goto-line.

> For those
> that are not willing to go the "better" path (of course, if it becomes
> a consensus that this is actually "better", I'm not insisting this
> is), they can always customize their keybindings. 

Yes, that's what I've done, so I certainly won't insist on M-g being
goto-line. Still, making goto-line more accesible that "set default font"
seems sensible for all these "uneducated" people out there which don't
want to spend too much time learning Emacs, i.e., the exact people to
whom the customize stuff is addressed...

> With the examples you mentioned, it seems that you haven't made use
> of emacsclient so far; I recommend it, since it's been my $EDITOR for
> a long time. ;)

In fact, I routinely have Emacs running all the time and have an alias
"em=gnuclient -qF" so I can edit files with "em filename.ext". gnuclient
is also my editor-cmd in Subversion, so Emacs comes forward to edit
Subversion commit logs, etc. etc.

Really, I *do* know there are alternatives. I use some of them, and the
ones I don't use, is because they're not that useful/comfortable to me.
I refuse to be "educated" to do things in ways I've already discarded ;)

                                                                Juanma

Re: Suggestion: Mapping of M-g should be goto-line

[Danilo Segan]

Hi Juanma,

Today at 12:55, Juanma Barranquero wrote:

> On Thu, 25 Mar 2004 12:23:34 +0100
> Danilo Segan <dsegan@gmx.net> wrote:
>
>> That's only a reason to educate them better, not to adjust behaviour.
>
> Many people don't have the time to be "educated". I've introduced a few
> friends to Emacs and they're using it happily, but they just won't spend
> the time to learn features they feel they're not going to need.

Sorry if I sounded too harsh -- I just want the defaults to be good,
and I'm not at all claiming to know what they should be.  I'm
bringing issues and/or solutions to issues brought up in discussion,
which are more "Emacs-ey" than simply using goto-line by hand.

With that aside, your claim seems to be spurious at best: you've got
friends who "just won't spend the time to learn features they feel
they're not going to need" -- if they're not going to need them, why
would they care if goto-line is easily accessible, or not?  OTOH, if
they're going to need them, we should make them learn next-error, not
goto-line (at least IMHO, it's related to the issue of what is
officially supported/recommended way of doing things with Emacs).

>> I suggest you try "emacsclient -n +5 path/to/file"
>
> I don't use emacsclient, but I *do* use gnuclient.exe and gnuserv.el
> (I'm on Windows), which have the same functionality. Still, I routinely
> use M-x goto-line.

Thanks for mentioning that.  But two of us, _who_ do know of alternate
[that's important, since it may be the case that many people don't
know about them], and (arguably) better way to open a file at specific
point from terminal, are far from a good statistical sample (and even
our views differ).

>> For those that are not willing to go the "better" path (of course,
>> if it becomes a consensus that this is actually "better", I'm not
>> insisting this is), they can always customize their keybindings. 
>
> Yes, that's what I've done, so I certainly won't insist on M-g being
> goto-line. Still, making goto-line more accesible that "set default font"
> seems sensible for all these "uneducated" people out there which don't
> want to spend too much time learning Emacs, i.e., the exact people to
> whom the customize stuff is addressed...

I'd actually go the different route.  I'd suggest making next-error
even more accessible (eg. using M-g for that), since it's infinitely
more useful than setting default face.  With that done, one might even
put goto-line to C-x `, or something.  Lets not knowingly hide very
useful features such as next-error, and expose goto-line instead.

(If you want a shortcut to make some sense, it's easy: "error" in
Serbian is "greška" [so M-g], and you can easily remember it, right? ;)

> In fact, I routinely have Emacs running all the time and have an alias
> "em=gnuclient -qF" so I can edit files with "em filename.ext". gnuclient
> is also my editor-cmd in Subversion, so Emacs comes forward to edit
> Subversion commit logs, etc. etc.
>
> Really, I *do* know there are alternatives. I use some of them, and the
> ones I don't use, is because they're not that useful/comfortable to me.

(I'd rather put the emphasis on "I": "*I* do know there are
alternatives", since I'd like to point out that many others who
reach for goto-line probably don't)

The last time I used Windows, I found its terminal to be very
unusable.  So, I would tend to blame incomfortability on that
instead. :)

> I refuse to be "educated" to do things in ways I've already discarded ;)

For those who're already long-time users of Emacs (like probably
yourself), there's no point in "educating" -- you know how to set it
up to suit your preferences, and have probably developed your own set
of preferences over time, which differ from defaults. 

Cheers,
Danilo

Re: Suggestion: Mapping of M-g should be goto-line

[David Kastrup]

Danilo Segan <dsegan@gmx.net> writes:

> Today at 12:55, Juanma Barranquero wrote:
> 
> > On Thu, 25 Mar 2004 12:23:34 +0100
> > Danilo Segan <dsegan@gmx.net> wrote:
> >
> >> That's only a reason to educate them better, not to adjust
> >> behaviour.
> >
> > Many people don't have the time to be "educated". I've introduced
> > a few friends to Emacs and they're using it happily, but they just
> > won't spend the time to learn features they feel they're not going
> > to need.
> 
> Sorry if I sounded too harsh -- I just want the defaults to be good,

M-g for goto-line is a good default.  "Educating" users is not the
task of an editor, and certainly not by willful omission of features.
Next thing you'll propose educating users about key combinations by
removing the menus.  And educating them about M-x delete-file RET by
making shell-mode barf on rm.

That's not the way to go about it.  Education of users is by making
the available information more accessible, not by hiding away all
other possibilities and making them inconvenient.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: Suggestion: Mapping of M-g should be goto-line

[Danilo Segan]

Today at 14:43, David Kastrup wrote:

> Danilo Segan <dsegan@gmx.net> writes:
>> 
>> Sorry if I sounded too harsh -- I just want the defaults to be good,
>
> M-g for goto-line is a good default.  

Such a blank statement is not really something I'd consider a good
analysis.  Certainly, you're an established Emacs hacker and
power-user (at least compared to me), so you do have some knowledge of
what is good and what is not good for Emacs.

Yet, the point you seem to be missing is that M-g is also a good
default for infinitely many other things.  And that's why I'm making
such a fuss about it.

> "Educating" users is not the task of an editor, and certainly not by
> willful omission of features.  Next thing you'll propose educating
> users about key combinations by removing the menus.  And educating
> them about M-x delete-file RET by making shell-mode barf on rm.

Ah, so I see.  Let's put cua-mode as a default; after all, majority
of potential Emacs users lurk there.  Lets not "educate" users that
C-c is not for copying with the lame excuse that it's better suited
for many other shortcuts.

Or, is in fact Emacs already doing what I'm "hereticaly" proposing?
Educating users with a different way of working?  Yeah, I guessed
so.  Those who do not want to be "educated", need to load all sort
of stuff (like M-x viper or cua-mode), in order to use what they
already know.

I didn't propose removing goto-line function, but rather, if we're
looking for improvements, lets make improvements where they matter
as well (perhaps where they matter even more).  And Kim Storm
expanded that point in the area where it also seems very much
relevant.

> That's not the way to go about it.  Education of users is by making
> the available information more accessible, not by hiding away all
> other possibilities and making them inconvenient.

Please, tell me how did you come up with this?  Did what I "proposed"
make goto-line in any way more inconvenient compared to what we have
now?

I repeatedly argued only for having next-error *more* visible than
goto-line, not for obscuring goto-line at all, on the assumption that
it's supposed to be more useful.  Juanma doesn't agree with this
assumption, so he naturally doesn't agree that next-error should
receive more exposure.  It's up to Emacs developers to decide whether
my assumption is correct, since they worked on next-error
functionality, and they know whether it was supposed to be more useful
(FAQ entry seems to indicate it was)--perhaps it failed?

Since you seem to be pretty much concerned with my usage of word
"educate", I'll point out that I'm using it in the sense of "make
aware of features in Emacs" (whether by making them more accessible
from UI, writing about them in more appropriate places in manual,
advertising them, or whatever), not only by forcing users to learn
current behaviour.  Perhaps I chose the wrong word.  If so, I'm deeply
sorry about it.

Cheers,
Danilo

Re: Suggestion: Mapping of M-g should be goto-line

[Ted Lemon]

I'm a little late in the game in replying to this particular thread, 
but I'd just like to throw in a word here for not changing the default 
bindings all the time.   Every time I get a new version of Emacs, I 
find that a bunch of bindings have changed.   If you want different 
bindings, please hack them into your .emacs file rather than forcing 
all the other emacs users to hack their .emacs files with every new 
release.

Re: Suggestion: Mapping of M-g should be goto-line

[Juanma Barranquero]

On Thu, 25 Mar 2004 13:30:18 +0100
Danilo Segan <dsegan@gmx.net> wrote:

> Sorry if I sounded too harsh

Not at all. My criticism, if any, is directed to the idea that
"educating" is always the best answer. People often don't want to be
educated, and the reasons (lack of time, interest, whatever) are as
subjective as perfectly legitimate.

> -- I just want the defaults to be good,
> and I'm not at all claiming to know what they should be.  I'm
> bringing issues and/or solutions to issues brought up in discussion,
> which are more "Emacs-ey" than simply using goto-line by hand.

I, in fact, agree with you. We differ in that I think goto-line is a
perfectly reasonable thing to do in many situations, and its relation
with compiling, etc. is only marginal.

> With that aside, your claim seems to be spurious at best: you've got
> friends who "just won't spend the time to learn features they feel
> they're not going to need" -- if they're not going to need them, why
> would they care if goto-line is easily accessible, or not?

I was referring to things like compile.el; in a few cases, they chose
the same path I do (not because of my influence, just because they
didn't want to spend time reading the manuals): compile in another
console window, use Emacs to edit, fix, save, and back to the console
to compile again. Certainly they do use goto-line; in fact, binding it
to a key was one of the first things they asked me to do (the other one
is setup an f-key to do bs-cycle-next, they won't touch switch-to-buffer
and list-buffers with a ten feet pole).

> But two of us, _who_ do know of alternate [...]
> are far from a good statistical sample (and even
> our views differ).

Sure.

> Lets not knowingly hide very
> useful features such as next-error, and expose goto-line instead.

The falacy here is that you consider next-error much more useful than
goto-line. I don't. I'd bet for users in general, many of which will
never compile a thing, going to a line is orders of magnitude more usual
than searching for an error.

I have a friend (a bookstore owner) who manages all client orders through
an ASCII/Perl/MultiEdit setup (MultiEdit is a programmers' editor). He
doesn't program at all; he writes orders in ascii files, process them
with Perl scripts, send the resulting request to his suppliers, and on
arrival of the goodies, he manually edits the ascii files to remove
items and classifies the items in boxes for his clients. He won't know
what next-error does, but he uses (the MultiEdit equivalent of) goto-line
many times a day. As you've said earlier, you and I and most people
around here are not statistically normal Emacs users (I think).

> (If you want a shortcut to make some sense, it's easy: "error" in
> Serbian is "greška" [so M-g], and you can easily remember it, right? ;)

Not a good mnemonic, "gresca" in Spanish means something totally
different, I'm afraid ;)

> since I'd like to point out that many others who
> reach for goto-line probably don't)

Or, perhaps, it's just that there are not alternatives for a simple
task: going to a line. Not an error line, not a line containing some
text, just to the line numbered X. 

> you know how to set it
> up to suit your preferences, and have probably developed your own set
> of preferences over time, which differ from defaults. 

Sure.

                                                                Juanma

Re: Suggestion: Mapping of M-g should be goto-line

[Danilo Segan]

Today at 14:53, Juanma Barranquero wrote:

>> Lets not knowingly hide very
>> useful features such as next-error, and expose goto-line instead.
>
> The falacy here is that you consider next-error much more useful than
> goto-line. I don't. I'd bet for users in general, many of which will
> never compile a thing, going to a line is orders of magnitude more usual
> than searching for an error.

Yeah, I did make all conclusions on the assumption that next-error is
more usable then goto-line.  If that assumption fails, so do all the
conclusions I made ;)

Apart from programming-like stuff, I really cannot imagine lots of
examples that really need goto-line.  And what I can imagine, it can
be automated, or it resembles programming a lot (like TeX/dvi).
next-error is not used only for programming (eg. po-mode for
translating program UIs via gettext PO files also makes use of it, and
it has nothing to do with programming).

>> since I'd like to point out that many others who
>> reach for goto-line probably don't)
>
> Or, perhaps, it's just that there are not alternatives for a simple
> task: going to a line. Not an error line, not a line containing some
> text, just to the line numbered X. 

I still cannot imagine a real world example where such a thing would
be _commonly_ used, but I'll blame my programming-deviation on that. ;)

I think I'll ease on the spamming of emacs-devel for a while :)

Cheers,
Danilo

Re: Suggestion: Mapping of M-g should be goto-line

[Juanma Barranquero]

On Thu, 25 Mar 2004 15:49:19 +0100
Danilo Segan <dsegan@gmx.net> wrote:

> Apart from programming-like stuff, I really cannot imagine lots of
> examples that really need goto-line.

`next-error' is a particular case of the general one: processing an
input file and getting some info related to specific lines in the input
(be an error, the matching of a regular expression, or whatever). As
has already been said, Emacs can not embrace every single use of the
general feature "going to a line of a file for whatever reason". If it
could, there wouldn't be a `goto-line'.

> I still cannot imagine a real world example where such a thing would
> be _commonly_ used, but I'll blame my programming-deviation on that. ;)

I've already put a few examples of real-life, both mine and from people
I've directly observed.

> I think I'll ease on the spamming of emacs-devel for a while :)

Yeah, I think we've pretty much beaten the dead equine to a pulp... :)

                                                                Juanma

Re: Suggestion: Mapping of M-g should be goto-line

[Kim F. Storm]

Juanma Barranquero <jmbarranquero@wke.es> writes:

> Or, perhaps, it's just that there are not alternatives for a simple
> task: going to a line. Not an error line, not a line containing some
> text, just to the line numbered X. 

M-x 123 RET

Argh -- that cannot be done in lisp :-(


-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: Suggestion: Mapping of M-g should be goto-line

[Stefan Monnier]

>> Or, perhaps, it's just that there are not alternatives for a simple
>> task: going to a line. Not an error line, not a line containing some
>> text, just to the line numbered X. 

> M-x 123 RET

> Argh -- that cannot be done in lisp :-(

Of course it can be.  M-x right now is written in C, but it's pretty easy
to rewrite it in Lisp (without any changes to the C code).

Along the same lines, I presonally use C-s 123 (which gives me incremental
line number search, not that the incremental nature of it is a feature in
this case, more like a side effect of hooking into isearch).

Obviously I agree that goto-line is a common need since I went to the
trouble of writing my C-s hack to do the equivalent.  Before that I had
M-g bound to goto-line but my fingers often instinctively hit C-s when
I was "searching for line number 123", so I figured I should adapt Emacs
to my fingers rather than the other way around.

In my book the default M-g bindings are "never used other than to test
whether they still work".  Especially since the `face' part of it (arguably
the main part) simply doesn't work in any buffer that uses font-lock.


        Stefan

Re: Suggestion: Mapping of M-g should be goto-line

[Per Abrahamsen]

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

> In my book the default M-g bindings are "never used other than to test
> whether they still work".  

I do very occasionally use it to _\bu_\bn_\bd_\be_\br_\bl_\bi_\bn_\be stuff in Gnus, mostly when
I want to confuse everybody with my amazing geek powers to use an
ancient, forgotten and unsupported Usenet feature.

Re: Suggestion: Mapping of M-g should be goto-line

[David Kastrup]

Per Abrahamsen <abraham@dina.kvl.dk> writes:

> Stefan Monnier <monnier@iro.umontreal.ca> writes:
> 
> > In my book the default M-g bindings are "never used other than to test
> > whether they still work".  
> 
> I do very occasionally use it to _\bu_\bn_\bd_\be_\br_\bl_\bi_\bn_\be stuff in Gnus, mostly when
> I want to confuse everybody with my amazing geek powers to use an
> ancient, forgotten and unsupported Usenet feature.

Actually, writing stuff like _this_ will properly underline in most
readers.  I am not so sure about your backspaced stuff above.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: Suggestion: Mapping of M-g should be goto-line

[Jari Aalto+mail.linux]

* Thu 2004-03-25 Danilo Segan <dsegan <AT> gmx.net> mail.default.spool
| That's only a reason to educate them better, ...

That is not the correct way. I believe that if there is functionality
that is not appropriate for the 99 % of the users, it should be
changed - not matter what other "education" might then be.

But you missed the point. Emacs is a companion. It should be a good
one.  Right now M-g is not a good companion. It could be much better
with goto-line.

| I suggest you try "emacsclient -n +5 path/to/file"
| ... With all this, I rarely if ever need to use M-x goto-line.  

For you maybe. I believe the more user friendly Emacs is, the better
people get a hand on it. Why do you think there is vi(1) people
that never touched emacs? Or Windows users that use programmer's 
file editor instead of Emacs?

Little things can make a difference sometimes. M-g could play a tiny
part towards it.

| For "average joe", we want to make them learn the better way.

Sure, they can add line to .emacs to map goto-line. Everybody
can. Millions of users can. But that's unproductive and unnecessary.


Jari

-- 
http://tiny-tools.sourceforge.net/
Swatch @time   http://www.mir.com.my/iTime/itime.htm
               http://www.ryanthiessen.com/swatch/resources.htm
Use Licenses!  http://www.linuxjournal.com/article.php?sid=6225
Which Licence? http://www.linuxjournal.com/article.php?sid=4825
OSI Licences   http://www.opensource.org/licenses/

Re: Suggestion: Mapping of M-g should be goto-line

[Danilo Segan]

Today at 13:25, Jari Aalto wrote:

> * Thu 2004-03-25 Danilo Segan <dsegan <AT> gmx.net> mail.default.spool
> | That's only a reason to educate them better, ...
>
> That is not the correct way. I believe that if there is functionality
> that is not appropriate for the 99 % of the users, it should be
> changed - not matter what other "education" might then be.

I'm not arguing for keeping M-g bound to face changing functions.  I'm
arguing for not using goto-line more than next-error and other
mechanisms, since they're *easier* to use.

> But you missed the point. Emacs is a companion. It should be a good
> one.  Right now M-g is not a good companion. It could be much better
> with goto-line.

Exactly, and it would be even better if it was bound to next-error.
You're not bringing any points that my view is not a valid one, or
usage patterns where using goto-line is easier than next-error, or
"emacsclient [+LINE[:COLUMN]]".  Visual tools should provide a way to
enter any editor at "current" line (like DVI viewers do), and
creating web pages is only a very tiny subset of what people do with
Emacs today.

> | I suggest you try "emacsclient -n +5 path/to/file"
> | ... With all this, I rarely if ever need to use M-x goto-line.  
>
> For you maybe. I believe the more user friendly Emacs is, the better
> people get a hand on it. 

This seems strange.  Are you actually claiming that if I compile a
program in one xterm, and get an error like

   some-main-file.c:655: error, this is error

it's easier for me to remember filename (main.c), line number (655),
switch to Emacs (which might hide the current xterm, or which might
be in the different workspace), find-file main.c (for which I usually
need to type couple of path components as well), type
whatever-the-shortcut for goto-line (eg. M-g), and enter the line 
number (provided I didn't forget it this far :)?  Instead of the simple

   emacsclient -n +655 some-main-file.c

while reading both the line number and filename on the screen (so no
need to remember them), and making use of name completion in the
terminal?

How could the first approach using goto-line be "more user friendly"
than the latter?  If that's your point, I strongly disagree.  And no,
I'm not talking only about me, I'm talking about everyone using a
terminal outside Emacs for such tasks.

> Why do you think there is vi(1) people that never touched emacs?

Some, at least, do it for religious reasons ;)  vi also has a
completely different philosophy, and all the things that stand for
Emacs, also stand for vi, and maybe even more so -- it's even harder
for newbies to use, so I don't really understand your bringing it up
(and those who want may try M-x viper instead).  I surely don't
think Emacs should try to replicate "user-friendliness" of vi ;)

> Or Windows users that use programmer's file editor instead of Emacs?

Because Emacs is entirely different from how other Windows programs
behave?  C-x, C-c don't do the tasks people got used them to do in
Windows environment?

> Little things can make a difference sometimes. M-g could play a tiny
> part towards it.

Yeah, but I'm still not convinced that binding goto-line there is
that tiny part towards it.

> | For "average joe", we want to make them learn the better way.
>
> Sure, they can add line to .emacs to map goto-line. Everybody
> can. Millions of users can. But that's unproductive and unnecessary.

I'm not that insane to claim that adding a line to .emacs is the
"better way" if a feature is to be used by most of the users.  I'm
trying to point out that there's such a need for goto-line only
because other Emacs features which are the right way are not exposed
enough: so, we need to expose them instead, not to expose goto-line.

Re: Suggestion: Mapping of M-g should be goto-line

[Jari Aalto]

* Thu 2004-03-25 Danilo Segan <dsegan <AT> gmx.net> mail.default.spool
| Exactly, and it would be even better if it was bound to next-error ...

[snip]

And your argument to help handling PHP and other web programming
languages errors - which print error line numbers in web page - was
....?

next-error is useless in those an in many other cases with shell
tools, other programs, other utilities spit out line numbers.

M-g is it, goto-line it it.

Using next-line is not the solution. It some weird way of seeing it
"all done Emacs way". Th reality is not  "all Emacs way", although
we might want it.

| ... there's such a need for goto-line only
| because other Emacs features which are the right way are not exposed
| enough: so, we need to expose them instead, not to expose goto-line.

Wrong assumption. You assume that Emacs has all the answers. It hasn't
User got a problem. And he wants Emacs to answer it. next-error won't
do it, but goto-line will
        
Imagine those web programming error lines in Web pages. Imagine lint
tools, commercial ones, free ones, lines of obscure error message
lines with line numbers. Imagine legacy of other sysadm tools; home
grown and others. Picture it. It's not "all Emacs" or even "educate
users to use Emacs better".

Emacs will never be able to support all those, every variation, every
new tool, every user script. But M-g, goto-line would be there. Ah,
how relaxing that would be.
 
Jari

-- 
http://tiny-tools.sourceforge.net/
Swatch @time   http://www.mir.com.my/iTime/itime.htm
               http://www.ryanthiessen.com/swatch/resources.htm
Use Licenses!  http://www.linuxjournal.com/article.php?sid=6225
Which Licence? http://www.linuxjournal.com/article.php?sid=4825
OSI Licences   http://www.opensource.org/licenses/

Re: Suggestion: Mapping of M-g should be goto-line

[David Kastrup]

Danilo Segan <dsegan@gmx.net> writes:

> Today at 13:25, Jari Aalto wrote:
> 
> > * Thu 2004-03-25 Danilo Segan <dsegan <AT> gmx.net> mail.default.spool
> > | That's only a reason to educate them better, ...
> >
> > That is not the correct way. I believe that if there is functionality
> > that is not appropriate for the 99 % of the users, it should be
> > changed - not matter what other "education" might then be.
> 
> I'm not arguing for keeping M-g bound to face changing functions.  I'm
> arguing for not using goto-line more than next-error and other
> mechanisms, since they're *easier* to use.
> 
> > But you missed the point. Emacs is a companion. It should be a good
> > one.  Right now M-g is not a good companion. It could be much better
> > with goto-line.
> 
> Exactly, and it would be even better if it was bound to next-error.

No.  Emacs keybinding are, with few exceptions, designed to be
mnemonic.  In contrast to vi, we take RSI as the lesser evil than
contorting your brain to remember a keybinding.

M-g is mnemonic for goto-line.  It isn't for next-error.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: Suggestion: Mapping of M-g should be goto-line

[Alan Shutko]

Danilo Segan <dsegan@gmx.net> writes:

> How could the first approach using goto-line be "more user friendly"
> than the latter?  If that's your point, I strongly disagree.  And no,
> I'm not talking only about me, I'm talking about everyone using a
> terminal outside Emacs for such tasks.

emacsclient requires lots of keystrokes.  Even if you alias it, it
requires an open terminal on the same machine you're running Emacs. 

One common use case I have is reading (often really large) java log
files on another machine and checking on java exceptions.  next-error
won't work for those even if I transfer the log file and force it
into compilation-mode.  There is some code which will parse java
exceptions and jump similar to next-error, but I'd still have to
transfer the file, which I don't want to do (since it is big, remote,
and has lines appended to it while I'm looking at it).

Or, getting a backtrace from a web app in the browser.  I could cut
and paste it all into a buffer and fake it up appropriately, or I
could just use goto-line.

Or, another use-case: getting line numbers from a java debugger
running in a graphical window.  Sure, I could pop to a console
(bringing a third application into the mix) and type "emacsclient -n
+2323 LongDirectoryHeirarchy/Whatever.java", or I could flip to
Emacs, load it using file-cache to find the file, and quickly go to
the file.

Or, someone calls me saying "I'm having trouble understanding line
124 of this file, could you explain it?

Emacs's other features are not the solution to these issues.  I've
been using Emacs for about 10 years now and have a pretty fair handle
on what it does, and have coded things to make things easier, and
really, sometimes goto-line is the simplest answer.

-- 
Alan Shutko <ats@acm.org> - I am the rocks.
Man invented Alcohol. God invented Grass. Who do you trust?

Re: Suggestion: Mapping of M-g should be goto-line

[David Kastrup]

Danilo Segan <dsegan@gmx.net> writes:

> For "average joe", we want to make them learn the better way.

No.  We want them to be able to use Emacs productively.  One does not
do this by artificial deficiencies.  That an experienced user will
often be able to avoid using goto-line my more involved procedures
does not mean that it does not have its place.  And in particular,
the user will not magically discover useful alternatives just because
goto-line is hard to use.

The way to teach people is to make the information for better and
more complex alternatives better accessible, not by sabotaging use of
simpler commands.

I certainly count as a power user with Emacs, and even I use M-x
goto-line more often than, uh, set-face-whatever.  And "M-g" for
goto-line is certainly much more mnemonic that "face-whatever".

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: Suggestion: Mapping of M-g should be goto-line

[Danilo Segan]

Today at 13:27, David Kastrup wrote:

> Danilo Segan <dsegan@gmx.net> writes:
>
>> For "average joe", we want to make them learn the better way.
>
> No.  We want them to be able to use Emacs productively.  

I thought "better way" is a way of being more productive.  They can
be 'productive' also if they count lines one-by-one, and jump to the
wanted line that way, but it's certainly not a "better way" (except in
some situations where we might proclaim that this will cause them to
review lots of previously written text, so it's "better" ;).

> One does not do this by artificial deficiencies.  That an
> experienced user will often be able to avoid using goto-line my more
> involved procedures does not mean that it does not have its place.
> And in particular, the user will not magically discover useful
> alternatives just because goto-line is hard to use.

Indeed.  That's why useful alternatives need to be better exposed.
If you read my mails, you'd notice that that is what I actually asked 
for.  You may notice that I mentioned that (server-start) would be
better as a default (there're probably security and other
implications, so I'm not saying to take it as is now), and that I
recommended M-g for next-error, and C-x ` for goto-line, because the 
former is (or at least should be, IMO) more useful.

> The way to teach people is to make the information for better and
> more complex alternatives better accessible, not by sabotaging use of
> simpler commands.

Exactly.  Can you please point out where did I suggest anything
remotely similar to what you're implying here?

> I certainly count as a power user with Emacs, and even I use M-x
> goto-line more often than, uh, set-face-whatever.  And "M-g" for
> goto-line is certainly much more mnemonic that "face-whatever".

I never said that face-* belongs there.  I specifically said that it
may be dumped altogether, and some modes (like enriched) could
bring it back.  That still doesn't mean that goto-line is a good
function for that keybinding.  Better, yes, but can we do even 
better still?  That's what I asked, and that's what I'm asking.

As Lupus pointed out, "C-x `" is hard to type on at least French
keyboard, so that's another reason for binding next-error to another
key such as M-g.

[Emacs FAQ already documents some of this in the question "How can
I go to a certain line given its number?" -- perhaps it would need an
update?]

Cheers,
Danilo

Re: Suggestion: Mapping of M-g should be goto-line

[David Kastrup]

Danilo Segan <dsegan@gmx.net> writes:

> Today at 13:27, David Kastrup wrote:
> 
> > Danilo Segan <dsegan@gmx.net> writes:
> >
> >> For "average joe", we want to make them learn the better way.
> >
> > No.  We want them to be able to use Emacs productively.  
> 
> I thought "better way" is a way of being more productive.

You are weaseling.  We are talking about "make them learn the better
way", and doing that the hard way, by obstructing other options.  And
that is only a way of being more productive if
a) they did not actually need the method you are obstructing
b) they don't need the functionality so often that the cost of
learning is not higher than the cost of doing it in the simplistic
way.

We don't gain anything by making Emacs unusable for people that don't
bow to our idea of what they should be forced to learn.

> They can be 'productive' also if they count lines one-by-one, and
> jump to the wanted line that way, but it's certainly not a "better
> way"

So should we try somehow to make it impossible for users to count
lines one-by-one if they want to?

> > One does not do this by artificial deficiencies.  That an
> > experienced user will often be able to avoid using goto-line my
> > more involved procedures does not mean that it does not have its
> > place.  And in particular, the user will not magically discover
> > useful alternatives just because goto-line is hard to use.
> 
> Indeed.  That's why useful alternatives need to be better exposed.

But this is not what we are talking about right now.  We are talking
about the M-g keybinding.

>  If you read my mails, you'd notice that that is what I actually
> asked for.  You may notice that I mentioned that (server-start)
> would be better as a default

That will buy the user exactly squat, since he still needs to read up
on Emacs-server in order to use the functionality.  So even if it is a
good idea for every user to have (server-start) in his .emacs file, it
is useless to do this automatically, since its presence in the default
Emacs startup will not make the user _use_ this feature.

> (there're probably security and other implications, so I'm not
> saying to take it as is now), and that I recommended M-g for
> next-error, and C-x ` for goto-line, because the former is (or at
> least should be, IMO) more useful.

But M-g is much more mnemonic for goto-line.  I don't mind finding a
better keybinding than C-x ` for next-error, but M-g is an obvious
choice for goto-line.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: Suggestion: Mapping of M-g should be goto-line

[Kim F. Storm]

Danilo Segan <dsegan@gmx.net> writes:

> Hi Jari,  Lucas, Juanma,

> > And, ` is on AltGR 7 here, so basically absolutely unusable on French
> > keyboard layouts. Of course next-error is bound to something else
> 
> Which would only go so far to recommend assigning M-g to next-error,
> perhaps? ;)


This brings me to another pet of mine ... 

         M-x compile and M-x recompile

What about using a new "compile prefix" C-x c with bindings like:

C-x c c => compile
C-x c r => recompile
C-x c n => next error
C-x c p => prev error
C-x c g => grep

and:

C-x c l => goto-line

..

Unfortunately, using C-x c has a high danger of someone typing C-x C-c
which is probably why it isn't used :-(

But C-x C-c has always been too risky for my fingers, so I moved
that to C-x C-c C-x a LOOOOOONG time ago.

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: Suggestion: Mapping of M-g should be goto-line

[Danilo Segan]

Today at 14:35, Kim F. Storm wrote:

> This brings me to another pet of mine ... 
>
>          M-x compile and M-x recompile
>
> What about using a new "compile prefix" C-x c with bindings like:
>
> C-x c c => compile
> C-x c r => recompile
> C-x c n => next error
> C-x c p => prev error
> C-x c g => grep
>
> and:
>
> C-x c l => goto-line

Yeah, that would be very nice, IMO.

> Unfortunately, using C-x c has a high danger of someone typing C-x C-c
> which is probably why it isn't used :-(

If it turns out that M-g is currently very much unused, how good (or
bad) would it be to use that instead of the (perhaps more logical)
proposal you made?

Eg.
M-g c => compile (M-g c read as "Go Compile!" ;)
M-g r => recompile
M-g n => next-error
M-g l => goto-line (read as "Goto Line")
...


Cheers,
Danilo

Re: Suggestion: Mapping of M-g should be goto-line

[Juri Linkov]

Danilo Segan <dsegan@gmx.net> writes:
> Today at 14:35, Kim F. Storm wrote:
>> What about using a new "compile prefix" C-x c with bindings like:
>>
>> C-x c c => compile
>> C-x c r => recompile
>> C-x c n => next error
>> C-x c p => prev error
>> C-x c g => grep
>>
>> and:
>>
>> C-x c l => goto-line
>
> If it turns out that M-g is currently very much unused, how good (or
> bad) would it be to use that instead of the (perhaps more logical)
> proposal you made?
>
> Eg.
> M-g c => compile (M-g c read as "Go Compile!" ;)
> M-g r => recompile
> M-g n => next-error
> M-g l => goto-line (read as "Goto Line")
> ...

These key bindings are the most convenient.
And then please add these too:

M-g p => previous-error
M-g N => compilation-next-file
M-g P => compilation-previous-file

-- 
Juri Linkov
http://www.jurta.org/emacs/

Re: Suggestion: Mapping of M-g should be goto-line

[Richard Stallman]

I don't think key bindings are needed for M-x compile.  One doesn't
type it all that often.

A better key binding for next-error, and some binding for
previous-error, could be useful.  Sometimes one needs to type them
many times for one compilation.

Re: Suggestion: Mapping of M-g should be goto-line

[Joachim Nilsson]

On 03/27/04 06:52, Richard Stallman wrote:
> I don't think key bindings are needed for M-x compile.  One doesn't
> type it all that often.

With all due respect to all your years of hacking, but
that is a use-case question.

Remainder of this followup concerns bindings in C-mode.

I was spoon-fed C on Borland-C 2.0 IDE, so I have compile
bound to C-F9 (plus many other neat bindings from that IDE),
and I hit that sequence too much every day.  Much thanks to
gcc not being able to bail out in time and giving me followup
errors to something early in a file and partly also due to
not having a neat bindings for next/prev-error.

The strongest point, however, to me using compile this often
is simple laziness.  I know I have a crazy quick CPU and the
sloppiness in my style often suggest I've missed semicolons
and other stuff.  So I (and many more dumb-nuts like me)
rather fix one error at a time and then recompile.  Actually,
this "technique" is often taught to us in school with the
words; "Aw, don't trust that compiler output, it is probably
something else -- fix the first one and rebuild instead".
I know, "then we should fix the compiler" ...

Fwiw, I'd like it very much to have C-c C-c (or C-c C-f,
like in LaTeX-mode) be bound to compile (for the current
major mode).


Regards
  /Joachim

Re: Suggestion: Mapping of M-g should be goto-line

[Jari Aalto+mail.linux]

* 2004-03-25 no-spam <AT> cua.dk (Kim Storm) gmane.emacs.devel
* <http://groups.google.com/groups?oi=djq&as_umsgid=%3Cm31xnht5fq.fsf@kfs-l.imdomain.dk>
| Danilo Segan <dsegan <AT> gmx.net> writes:
| 
| > Hi Jari,  Lucas, Juanma,
| 
| > > And, ` is on AltGR 7 here, so basically absolutely unusable on French
| > > keyboard layouts. Of course next-error is bound to something else
| > 
| > Which would only go so far to recommend assigning M-g to next-error,
| > perhaps? ;)
| 
| 
| This brings me to another pet of mine ... 
| 
|          M-x compile and M-x recompile
| 
| What about using a new "compile prefix" C-x c with bindings like:
| 
| C-x c c => compile
| C-x c r => recompile
| C-x c n => next error
| C-x c p => prev error
| C-x c g => grep
| 

Excellent idea

| 
| C-x c l => goto-line

But not this. The whole point of M-g being so useful for goto-line
is its sing keys nature. If it were put to C-x c l ir to any other
key that requires "more keystrokes", it just don't cut the idea.

It's not just question of Being gogo-line mapping, it really needs to
be M-g. The de facto, that it has been for years in the user
community. Let's follow what's already there.

I would hate wasting keystrokes to reach other than M-g in a 
PHP debugging sessions - staring those Web page errors.

It's fortunately Alt-g in my keyboard. For other systems I warrant
goto-line to a function key of its own. It's really essential for
fluent programming.

Jari

-- 
http://tiny-tools.sourceforge.net/
Swatch @time   http://www.mir.com.my/iTime/itime.htm
               http://www.ryanthiessen.com/swatch/resources.htm
Use Licenses!  http://www.linuxjournal.com/article.php?sid=6225
Which Licence? http://www.linuxjournal.com/article.php?sid=4825
OSI Licences   http://www.opensource.org/licenses/

Re: Suggestion: Mapping of M-g should be goto-line

[David Kastrup]

jari.aalto@poboxes.com (Jari Aalto+mail.linux) writes:

> * 2004-03-25 no-spam <AT> cua.dk (Kim Storm) gmane.emacs.devel
> * <http://groups.google.com/groups?oi=djq&as_umsgid=%3Cm31xnht5fq.fsf@kfs-l.imdomain.dk>
>
> | C-x c c => compile
> | C-x c r => recompile
> | C-x c n => next error
> | C-x c p => prev error
> | C-x c g => grep
> | 
> 
> Excellent idea

Since C-x c is too close to C-x C-c, how about mixing the proposals
into an easily accessable binding?

C-c C-c C-c => compile
C-c C-c C-r => recompile
C-c C-c C-n => next error
C-c C-c C-p => previous error
C-c C-c Uhh... do we really need a case for grep?

Anyhow, I have already been musing about having a special keymap
active in AUCTeX after C-c C-c so that C-c C-c C-c will do the "most
common thing".  Could help avoid confusion.

I do agree that C-c C-c is a bad choice in PCL-CVS and should
probably be replaced by C-c C-k so that conflict is not much of a
nuisance.

Shell-mode is a different issue.  I am not sure that you won't ever
want to use compile or even next error in it.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: Suggestion: Mapping of M-g should be goto-line

[Stefan Monnier]

> Shell-mode is a different issue.  I am not sure that you won't ever
> want to use compile or even next error in it.

This is not relevant w.r.t my proposal since I don't want C-c C-c to be "the
binding for `compile'".


        Stefan

Re: Suggestion: Mapping of M-g should be goto-line

[Jari Aalto]

* 2004-03-26 David Kastrup <dak <AT> gnu.org> mail.default.spool
| jari.aalto <AT> poboxes.com (Jari Aalto+mail.linux) writes:
| 
| > * 2004-03-25 no-spam <AT> cua.dk (Kim Storm) gmane.emacs.devel
| > * <http://groups.google.com/groups?oi=djq&as_umsgid=%3Cm31xnht5fq.fsf@kfs-l.imdomain.dk>
| >
| > | C-x c c => compile
| > | C-x c r => recompile
| > | C-x c n => next error
| > | C-x c p => prev error
| > | C-x c g => grep
| > | 
| > 
| > Excellent idea
| 
| Since C-x c is too close to C-x C-c, how about mixing the proposals
| into an easily accessable binding?
| 
| C-c C-c C-c => compile
| C-c C-c C-r => recompile
| C-c C-c C-n => next error
| C-c C-c C-p => previous error
| C-c C-c Uhh... do we really need a case for grep?

Really, if there are any more C-c C- C- C- key combinations, my hand
will retire soon. The first proposal of "C-c c" prefix is much more
friendly to wrists.

Jari

Re: Suggestion: Mapping of M-g should be goto-line

[David Kastrup]

Jari Aalto <jari.aalto@poboxes.com> writes:

> * 2004-03-26 David Kastrup <dak <AT> gnu.org> mail.default.spool
> | jari.aalto <AT> poboxes.com (Jari Aalto+mail.linux) writes:
> | 
> | > * 2004-03-25 no-spam <AT> cua.dk (Kim Storm) gmane.emacs.devel
> | > * <http://groups.google.com/groups?oi=djq&as_umsgid=%3Cm31xnht5fq.fsf@kfs-l.imdomain.dk>
> | >
> | > | C-x c c => compile
> | > | C-x c r => recompile
> | > | C-x c n => next error
> | > | C-x c p => prev error
> | > | C-x c g => grep
> | > | 
> | > 
> | > Excellent idea
> | 
> | Since C-x c is too close to C-x C-c, how about mixing the proposals
> | into an easily accessable binding?
> | 
> | C-c C-c C-c => compile
> | C-c C-c C-r => recompile
> | C-c C-c C-n => next error
> | C-c C-c C-p => previous error
> | C-c C-c Uhh... do we really need a case for grep?
> 
> Really, if there are any more C-c C- C- C- key combinations, my hand
> will retire soon. The first proposal of "C-c c" prefix is much more
> friendly to wrists.

I can't remember any such proposal, and it would not go through,
anyway, since C-c letter combinations are reserved for the user.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: Suggestion: Mapping of M-g should be goto-line

[Miles Bader]

On Fri, Mar 26, 2004 at 10:29:08PM +0100, David Kastrup wrote:
> > Really, if there are any more C-c C- C- C- key combinations, my hand
> > will retire soon. The first proposal of "C-c c" prefix is much more
> > friendly to wrists.
> 
> I can't remember any such proposal, and it would not go through,
> anyway, since C-c letter combinations are reserved for the user.

In any case Jari's comment seems just plainly wrong -- it's _much_ easier to
type multiple keystroke bindings if all keys in the sequence use the same
modifier: e.g., to type C-c C-c C-c, just (1) hold down the the control key
with one finger, and (2) hit `c' three times with the other finger.

By contrast, e.g. the binding `C-c c c' is _harder_ to type, because the
release of the control key is must be made precisely between the first and
second keystrokes, which in my experience slows down typing considerably.

[BTW, I don't like the `C-c C-c'-as-a-prefix suggestion though -- bindings
shouldn't be used as prefixes in some modes and as non-prefixes in others.]

-Miles
-- 
o The existentialist, not having a pillow, goes everywhere with the book by
  Sullivan, _I am going to spit on your graves_.

Re: Suggestion: Mapping of M-g should be goto-line

[Jari Aalto+mail.emacs]

* Fri 2004-03-26 Miles Bader <miles <AT> gnu.org> gmane.emacs.devel
* <http://groups.google.com/groups?oi=djq&as_umsgid=%3C20040327001641.GD26429@fencepost>
| On Fri, Mar 26, 2004 at 10:29:08PM +0100, David Kastrup wrote:
| 
| > > Really, if there are any more C-c C- C- C- key combinations, my hand
| > > will retire soon. The first proposal of "C-c c" prefix is much more
| > > friendly to wrists.
| > 
| > I can't remember any such proposal, and it would not go through,
| > anyway, since C-c letter combinations are reserved for the user.
| 
| In any case Jari's comment seems just plainly wrong -- it's _much_ easier to
| type multiple keystroke bindings if all keys in the sequence use the same
| modifier: e.g., to type C-c C-c C-c, just (1) hold down the the control key
| with one finger, and (2) hit `c' three times with the other finger.

This assumes, that one indeed uses two hand on keyboard.

I use left hand on keyboard and right hand on mouse almost 90 % of the
time. I have found it more efficient and faster.  In cases, where it's
impossible to reach keys with left hand I have to shift both hands to
keyboard.

So, which one is "easier" depends how one uses it.

To throw another perspective, think about disabled persons which may
not have good hand coordination or strength. Keeping keys down is
harder that using non-modifier keys:

        C-c c c    vs.  C-c C-c C-c  

Jari

Re: Suggestion: Mapping of M-g should be goto-line

[David Kastrup]

jari.aalto@poboxes.com (Jari Aalto+mail.emacs) writes:

> * Fri 2004-03-26 Miles Bader <miles <AT> gnu.org>

> | In any case Jari's comment seems just plainly wrong -- it's _much_
> | easier to type multiple keystroke bindings if all keys in the
> | sequence use the same modifier: e.g., to type C-c C-c C-c, just
> | (1) hold down the the control key with one finger, and (2) hit `c'
> | three times with the other finger.
> 
> This assumes, that one indeed uses two hand on keyboard.

Not at all.  I can easily type C-c C-c with the left hand.  In
fact, even when I am typing with both hands, I use just the left hand
for C-c C-c and it is quite fast.

> I use left hand on keyboard and right hand on mouse almost 90 % of
> the time. I have found it more efficient and faster.  In cases,
> where it's impossible to reach keys with left hand I have to shift
> both hands to keyboard.

But it isn't with C-c.

> To throw another perspective, think about disabled persons which may
> not have good hand coordination or strength. Keeping keys down is
> harder that using non-modifier keys:
> 
>         C-c c c    vs.  C-c C-c C-c  

We should cater for the ergonomics of the most common users first.
Disabled persons will rebind their keys for their most common
applications, anyway.

We should try to cater for their needs as long as it does not
negatively impact more common users.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: Suggestion: Mapping of M-g should be goto-line

[Miles Bader]

On Sun, Mar 28, 2004 at 07:37:24PM +0200, David Kastrup wrote:
> > | In any case Jari's comment seems just plainly wrong -- it's _much_
> > | easier to type multiple keystroke bindings if all keys in the
> > | sequence use the same modifier: e.g., to type C-c C-c C-c, just
> > | (1) hold down the the control key with one finger, and (2) hit `c'
> > | three times with the other finger.
> > 
> > This assumes, that one indeed uses two hand on keyboard.
> 
> Not at all.  I can easily type C-c C-c with the left hand.  In
> fact, even when I am typing with both hands, I use just the left hand
> for C-c C-c and it is quite fast.

Heh, good point -- indeed, C-c is one of the easiest modifier bidnings to
type...

-Miles
-- 
`To alcohol!  The cause of, and solution to,
 all of life's problems' --Homer J. Simpson

Re: Suggestion: Mapping of M-g should be goto-line

[Miles Bader]

On Sun, Mar 28, 2004 at 07:02:16PM +0200, Jari Aalto+mail.emacs wrote:
> | In any case Jari's comment seems just plainly wrong -- it's _much_ easier
> | to type multiple keystroke bindings if all keys in the sequence use the
> | same modifier: e.g., to type C-c C-c C-c, just (1) hold down the the
> | control key with one finger, and (2) hit `c' three times with the other
> | finger.
 > 
> This assumes, that one indeed uses two hand on keyboard.
> 
> I use left hand on keyboard and right hand on mouse almost 90 % of the
> time. I have found it more efficient and faster.

You are an _extremely_ rare case then, I think -- even quite bad typists
usually use two hands when necessary to type emacs modifier bindings.  To
not do so makes _many_ bindings in emacs hard to type.

It's obvious that your experience is inappropriate for judging emacs
bindings.  [Indeed, harmful: what you find good, most people would probably
find bad.]

-Miles
-- 
`Cars give people wonderful freedom and increase their opportunities.
 But they also destroy the environment, to an extent so drastic that
 they kill all social life' (from _A Pattern Language_)

Re: Suggestion: Mapping of M-g should be goto-line

[David Kastrup]

jari.aalto@poboxes.com (Jari Aalto+mail.emacs) writes:

>         The "de facto" situation for long has been that everybody
>         maps M-g to goto-line. People advice that in newsgroups,
>         because it is indeed the most logical key.

Seconded.  If no other binding can be found, we can override it in
enriched text mode with a local binding, but as a global binding,
goto-line is certainly much more generally useful.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: Suggestion: Mapping of M-g should be goto-line

[Jari Aalto+mail.emacs]

* 2004-03-25 David Kastrup <dak <AT> gnu.org> mail.default.spool
| jari.aalto <AT> poboxes.com (Jari Aalto+mail.emacs) writes:
| 
| >         The "de facto" situation for long has been that everybody
| >         maps M-g to goto-line. People advice that in newsgroups,
| >         because it is indeed the most logical key.
| 
| Seconded.  If no other binding can be found, we can override it in
| enriched text mode with a local binding, but as a global binding,
| goto-line is certainly much more generally useful.


If possible, we really want M-g, because of "g", goto. Using any other
key would counter what people have been already using. And it should
be one key.  If it were two keys, people would never use it. They
would map it right back to "M-g".

Jari

Re: Suggestion: Mapping of M-g should be goto-line

[Joachim Nilsson]

On 03/25/04 13:27, Jari Aalto+mail.emacs wrote:
> * 2004-03-25 David Kastrup <dak <AT> gnu.org> mail.default.spool
> | jari.aalto <AT> poboxes.com (Jari Aalto+mail.emacs) writes:
> | >         The "de facto" situation for long has been that everybody
> | >         maps M-g to goto-line. People advice that in newsgroups,
> | >         because it is indeed the most logical key.
> | Seconded.  If no other binding can be found, we can override it in
> | enriched text mode with a local binding, but as a global binding,
> | goto-line is certainly much more generally useful.
> If possible, we really want M-g, because of "g", goto. Using any other
> key would counter what people have been already using. And it should
> be one key.  If it were two keys, people would never use it. They
> would map it right back to "M-g".

I too agree that a keybinding for goto-line is useful.  Like the
original poster I used M-g as well, but since 21.3 I have a new
setup where I use M-# instead.  To me that is equally intuitive,
but clashes with calc-dispatch -- so whatever is decided the risk
of a key already being bound is always present in Emacs.

Personally I've actually started getting used to "M-g *" for setting
faces when I write in LaTeX.  I find it very useful actually.

My reason for having a keybinding at all for goto-line is that I find
myself often supporting code over the telephone and I don't think we
have a mode for that yet in Emacs. :)


Regards
  /Joachim

Re: Suggestion: Mapping of M-g should be goto-line

[David Kastrup]

Joachim Nilsson <joachim.nilsson@vmlinux.org> writes:

> On 03/25/04 13:27, Jari Aalto+mail.emacs wrote:
> > * 2004-03-25 David Kastrup <dak <AT> gnu.org>

> > | Seconded.  If no other binding can be found, we can override it in
> > | enriched text mode with a local binding, but as a global binding,
> > | goto-line is certainly much more generally useful.
> > If possible, we really want M-g, because of "g", goto. Using any other
> > key would counter what people have been already using. And it should
> > be one key.  If it were two keys, people would never use it. They
> > would map it right back to "M-g".
> 
> I too agree that a keybinding for goto-line is useful.  Like the
> original poster I used M-g as well, but since 21.3 I have a new
> setup where I use M-# instead.  To me that is equally intuitive,
> but clashes with calc-dispatch -- so whatever is decided the risk
> of a key already being bound is always present in Emacs.

Some vague memory made me start up that other editor.  And guess what:

    M-g runs `goto-line'

    `goto-line' is an interactive compiled Lisp function
      -- loaded from "/usr/share/xemacs-21.4.12/lisp/simple.elc"
    (goto-line LINE)

    Documentation:
    Goto line LINE, counting from line 1 at beginning of buffer.

Since there is considerable consensus here that M-g is a reasonable
binding for goto-line, I would like to add that not choosing a
different binding would probably also be appreciated by those that
want or need to cross between Emacs incarnations.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: Suggestion: Mapping of M-g should be goto-line

[Per Abrahamsen]

Joachim Nilsson <joachim.nilsson@vmlinux.org> writes:

> Personally I've actually started getting used to "M-g *" for setting
> faces when I write in LaTeX.  I find it very useful actually.

Excellent!

When I preach Emacs to the heathens, my main point is that Emacs gives
you an environment that allows you to use the same keys to issue
*semantically* equivalent operations, even though those operations
have different textual representations depending on what kind of text
your editing.

The canonical example is "M-;", to comment out stuff.

So even though M-g currently is rarely used, I believe there should be
*some* common keybinding for setting fonts for the current mode.  It
doesn't has to be M-g, but it should be there.  It is what Emacs is
all about.  At least, as I preach the gospel.

Re: Suggestion: Mapping of M-g should be goto-line

[Kim F. Storm]

David Kastrup <dak@gnu.org> writes:

> jari.aalto@poboxes.com (Jari Aalto+mail.emacs) writes:
> 
> >         The "de facto" situation for long has been that everybody
> >         maps M-g to goto-line. People advice that in newsgroups,
> >         because it is indeed the most logical key.
> 
> Seconded.  If no other binding can be found, we can override it in
> enriched text mode with a local binding, but as a global binding,
> goto-line is certainly much more generally useful.


BTW, Gnus rebinds M-g:

M-g runs the command gnus-summary-rescan-group


-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: Suggestion: Mapping of M-g should be goto-line

[David Kastrup]

no-spam@cua.dk (Kim F. Storm) writes:

> David Kastrup <dak@gnu.org> writes:
> 
> > jari.aalto@poboxes.com (Jari Aalto+mail.emacs) writes:
> > 
> > >         The "de facto" situation for long has been that everybody
> > >         maps M-g to goto-line. People advice that in newsgroups,
> > >         because it is indeed the most logical key.
> > 
> > Seconded.  If no other binding can be found, we can override it in
> > enriched text mode with a local binding, but as a global binding,
> > goto-line is certainly much more generally useful.
> 
> 
> BTW, Gnus rebinds M-g:
> 
> M-g runs the command gnus-summary-rescan-group

I see no problem with that: in the summary buffer you would not want
to jump to a particular line, anyway.  It is a generated buffer
without external access that would make line numbers a useful metric.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: Suggestion: Mapping of M-g should be goto-line

[Kim F. Storm]

David Kastrup <dak@gnu.org> writes:

> no-spam@cua.dk (Kim F. Storm) writes:
> 
> > BTW, Gnus rebinds M-g:
> > 
> > M-g runs the command gnus-summary-rescan-group
> 
> I see no problem with that: in the summary buffer you would not want
> to jump to a particular line, anyway.  It is a generated buffer
> without external access that would make line numbers a useful metric.

It is no problem if M-g is goto-line.

My point was that it would be a problem if M-g runs next-error -- which
should really be available in ALL modes.

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: Suggestion: Mapping of M-g should be goto-line

[Richard Stallman]

I want Emacs to move in the direction of doing word processing.  It
may take years, but we will get there.  Then commands to specify faces
will become important, and will need a good key binding.

I chose the M-g binding for that reason, and the reason continues to
have force.  So I don't intend to change that binding.

Re: Suggestion: Mapping of M-g should be goto-line

[Jérôme Marant]

Richard Stallman <rms@gnu.org> writes:

> I want Emacs to move in the direction of doing word processing.  It
> may take years, but we will get there.  Then commands to specify faces
> will become important, and will need a good key binding.
>
> I chose the M-g binding for that reason, and the reason continues to
> have force.  So I don't intend to change that binding.

What if users prefer M-g to run goto-line?

-- 
Jérôme Marant

Re: Suggestion: Mapping of M-g should be goto-line

[David Kastrup]

jmarant@nerim.net (Jérôme Marant) writes:

> Richard Stallman <rms@gnu.org> writes:
> 
> > I want Emacs to move in the direction of doing word processing.
> > It may take years, but we will get there.  Then commands to
> > specify faces will become important, and will need a good key
> > binding.
> >
> > I chose the M-g binding for that reason, and the reason continues
> > to have force.  So I don't intend to change that binding.
> 
> What if users prefer M-g to run goto-line?

They are free to rebind it on their own.  That's not a concern.

A concern however is that M-g _is_ frequently rebound by users and
installations to goto-line, and so maybe we should spend some thought
about a recommendation where the font selection commands should go on
installations that rebind M-g, to avoid more inconsistency in actual
installations than can be avoided.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: Suggestion: Mapping of M-g should be goto-line

[Jérôme Marant]

David Kastrup <dak@gnu.org> writes:
 
>> What if users prefer M-g to run goto-line?
>
> They are free to rebind it on their own.  That's not a concern.
>
> A concern however is that M-g _is_ frequently rebound by users and
> installations to goto-line, and so maybe we should spend some thought
> about a recommendation where the font selection commands should go on
> installations that rebind M-g, to avoid more inconsistency in actual
> installations than can be avoided.

If anyone usually rebind it, it would make sense to make it a default
binding, wouldn't it? When you need to fire up Emacs where you don't
have your .emacs around, having such default bindings is generaly
not a bad idea.

-- 
Jérôme Marant

http://marant.org

Re: Suggestion: Mapping of M-g should be goto-line

[David Kastrup]

Richard Stallman <rms@gnu.org> writes:

> I want Emacs to move in the direction of doing word processing.  It
> may take years, but we will get there.  Then commands to specify
> faces will become important, and will need a good key binding.
> 
> I chose the M-g binding for that reason, and the reason continues to
> have force.  So I don't intend to change that binding.

I can't see the particular mnemonic value of that binding.  Even if we
stipulate the importance of its function, it's not very in line with
Emacs' tradition of having bindings one can easily remember.

And apart from not being easy to remember, it's not that easy to type
either: if you are not using a keyboard with Meta key, M-g M-g is
quite inconvenient to type, and if you are using a keyboard with a
Meta key, M-g b where you have to let go of the Meta key in the right
moment is inconvenient to type.

AUCTeX has for its font selection commands C-c C-f C-t (for example,
for switching to typewriter).  There are a few differences (for
example, if a transient region is active, the font selection command
applies to that region), but the main point is
a) it is easy to remember
b) it is easy to type
Even if the M-g binding would work under AUCTeX, I doubt that anyone
would prefer to use it.

Of course, assigning M-g to goto-line is not something we should do
before we have found a better place for the font selection commands.
If we manage to find such a place, however (one that is both easy to
type and associate with fonts or faces), there would not be much
sense not changing into a setup that would be convenient for all
users, including those that work with text processing.

And don't tell me I am not concerned with them: I use Emacs almost
exclusively for text processing and am maintainer of the LaTeX WYSIWYG
extension preview-latex for it as well as AUCTeX.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: Suggestion: Mapping of M-g should be goto-line

[Richard Stallman]

    > I chose the M-g binding for that reason, and the reason continues to
    > have force.  So I don't intend to change that binding.

    I can't see the particular mnemonic value of that binding.

M-g is not mnemonic.  It is not a very good binding, but it seemed to
be the best that was available.  If you can suggest a better binding
for the text property commands, I could agree to moving them.

    AUCTeX has for its font selection commands C-c C-f C-t (for example,
    for switching to typewriter).

That's okay for AUCTeX because it is a major mode, but it would not be
allowed for a global binding.  Anyway, it is one extra character,
which is not very nice.  Anyone using Emacs to edit a document with
faces would probably use these commands very often.

Re: Suggestion: Mapping of M-g should be goto-line

[Kim F. Storm]

Richard Stallman <rms@gnu.org> writes:

>     > I chose the M-g binding for that reason, and the reason continues to
>     > have force.  So I don't intend to change that binding.
> 
>     I can't see the particular mnemonic value of that binding.
> 
> M-g is not mnemonic.  It is not a very good binding, but it seemed to
> be the best that was available.  If you can suggest a better binding
> for the text property commands, I could agree to moving them.

If we look for a binding for "text processor" modes only, it would be
more logical to have the face bindings on, e.g.

        C-c C-b => bold
        C-c C-a => italic   (easier to type that C-c C-i)
        C-c C-d => default
        etc.

These are currently unbound in text-mode.

This may (haven't checked) conflict with existing bindings in other
text-related modes eg. enriched-text-mode or tex-mode, and also be
shadowed by C-c bindings in other modes derived from text-mode.

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: Suggestion: Mapping of M-g should be goto-line

[Miles Bader]

On Sun, Mar 28, 2004 at 07:03:49PM +0200, Kim F. Storm wrote:
> If we look for a binding for "text processor" modes only, it would be
> more logical to have the face bindings on, e.g.
> 
>         C-c C-b => bold
>         C-c C-a => italic   (easier to type that C-c C-i)
>         C-c C-d => default
>         etc.

That's something to be wary of though, as it uses up so many bindings.

It _might_ be appropriate for the most common faces in a pure word-processor
mode, but I'd be afraid of it interfering with many other natural bindings in
a mode that tried to add other functionality.

-Miles
-- 
.Numeric stability is probably not all that important when you're guessing.

Re: Suggestion: Mapping of M-g should be goto-line

[Per Abrahamsen]

storm@cua.dk (Kim F. Storm) writes:

>         C-c C-b => bold

This is "run command on region" in AUC TeX.

>         C-c C-d => default

This is "save current document" in AUC TeX.

I don't believe you should steal so many useful keybindings for simple
font manipulation commands.  Users of any kind of higher level
word-processing would need such commands only rarely.  Instead they
will use commands such as "mark this text as program code".  And even
users of a pure "What You See Is All You Got" would hopefully not make
font switching that common.

Re: Suggestion: Mapping of M-g should be goto-line

[David Kastrup]

Per Abrahamsen <abraham@dina.kvl.dk> writes:

> storm@cua.dk (Kim F. Storm) writes:
> 
> >         C-c C-b => bold
> 
> This is "run command on region" in AUC TeX.
> 
> >         C-c C-d => default
> 
> This is "save current document" in AUC TeX.
> 
> I don't believe you should steal so many useful keybindings for
> simple font manipulation commands.

Actually is _is_ simple commands that should get the useful
keybindings.  The complex ones can be called explicitly.

But if we really want to get into mode-surpassing text processing and
markup, it is quite clear that the current keybindings leave us almost
no place to go.  That is why I suggested opening this particular can
of worms together with the "Super" key.  Placing text processing
functions on Super keys will not take away the currently alloted
keybinding space, and with an entry shortcut like M-s, one could even
manipulate stuff on pure text terminals (where the need for font
manipulation of the text may be less, but not completely absent).

Current "industry standard" keyboards have a key that is used as
Super-modifier in recent X servers.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: Suggestion: Mapping of M-g should be goto-line

[Joachim Nilsson]

On 03/27/04 06:52, Richard Stallman wrote:
> I want Emacs to move in the direction of doing word processing.  It
> may take years, but we will get there.  Then commands to specify faces
> will become important, and will need a good key binding.
> I chose the M-g binding for that reason, and the reason continues to
> have force.  So I don't intend to change that binding.

The TeXmacs people chose M-A-b, M-A-i, etc. for bold and italic resp.


  /Joachim

Re: Suggestion: Mapping of M-g should be goto-line

[Richard Stallman]

    The TeXmacs people chose M-A-b, M-A-i, etc. for bold and italic resp.

I don't think it is possible to type M-A- characters on most
keyboards.  How would you do it?  There is an Emacs command that
applies the Alt prefix to the following character, but it is
cumbersome.  (I don't remember it because I've never used it.)

Re: Suggestion: Mapping of M-g should be goto-line

[Joachim Nilsson]

On 03/28/04 06:25, Richard Stallman wrote:
>     The TeXmacs people chose M-A-b, M-A-i, etc. for bold and italic resp.
> I don't think it is possible to type M-A- characters on most
> keyboards.  How would you do it?

I don't know of other environments than Sun and COTS PC's.
On the PC's I run there is _always_ a Windows-key on the
left-hand side of the keyboard between Ctrl and Alt. This key
in XFree86 is mapped to Alt, whereas the actual Alt is mapped
to Meta (or if it is the other way around).

I'm not suggesting this is the best solution, but it actually
works - and might be suitable for us that want to see Emacs
as a useful environment for both documenting (LaTeX/TeXinfo/RTF)
and programming (M-g => goto-line). I know I do.


  /Joachim

Re: Suggestion: Mapping of M-g should be goto-line

[David Kastrup]

Joachim Nilsson <joachim.nilsson@vmlinux.org> writes:

> On 03/28/04 06:25, Richard Stallman wrote:
> >     The TeXmacs people chose M-A-b, M-A-i, etc. for bold and italic resp.
> > I don't think it is possible to type M-A- characters on most
> > keyboards.  How would you do it?
> 
> I don't know of other environments than Sun and COTS PC's.
> On the PC's I run there is _always_ a Windows-key on the
> left-hand side of the keyboard between Ctrl and Alt. This key
> in XFree86 is mapped to Alt, whereas the actual Alt is mapped
> to Meta (or if it is the other way around).
> 
> I'm not suggesting this is the best solution, but it actually
> works - and might be suitable for us that want to see Emacs
> as a useful environment for both documenting (LaTeX/TeXinfo/RTF)
> and programming (M-g => goto-line). I know I do.

This is usually maps as the Super-modifier in Emacs, and as opposed to
the Meta-Key it does not have a keyboard workaround.  If we wanted to
place font selection commands on the Super-key, that is a widely free
area yet.  It would be an idea to open that for text processing
functions in general.  However, we would still need to have a
reasonably uncomplicated other method for keying stuff like this in
even on text terminals: a text terminal is sufficient for, say,
writing LaTeX texts, and this should remain so.

If we declare the Super-Domain open for text processing, the obvious
keyboard translation for "Super" would be M-s, and the previous
function "center-line" could be placed on M-s c as it can be seen as a
text processing function (in LaTeX mode, this probably should place a
"center" environment instead).

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: Suggestion: Mapping of M-g should be goto-line

[Robert J. Chassell]

    > I don't know of other environments than Sun and COTS PC's.
    > On the PC's I run there is _always_ a Windows-key ...

My commercial of the shelf personal computer does not have a key
labeled `Windows'.  

It does have two keys labeled `Alt' and two keys labeled `Ctrl'.  I
have have rebound the right hand pair to Super and Hyper respectively.
It might make sense to add more bindings to the Super key.

However, I do not know how easy it would be to ensure a Super binding
on all the keyboards that Emacs people use.  My impression is that
Emacs does not have to do anything about keyboards in which the Meta
key is misleadingly labeled Alt, except that the documentation needs
to explain that the key label is misleading.  

For a Super keybinding on my keyboard, some Emacs auto-keybinding
program would have detect that it has two keys labeled `Alt', that
they are different, and that the right handed one should be rebound to
Super.

On a keyboard with an Alt or Meta key that is misleadingly labeled
`Windows', the Emacs keybinding program must rebind it to Super, but
only if there is more than one Meta key.

Perhaps David Kastrup is right to suggest that by default, a `Meta-s'
prefix be provided for people who lack a Super key, and that default
alternative be a part of all the documentation, so that when you read
about centering a line you see this:

    (emacs)Fill Commands

    The command `S-s' (`center-line') (or if you lack a Super key,
    `M-s c') centers the current line ...

or when your are reading about setting a face, you see this:

    (emacs)Format Faces

    `S-b' (or if you lack a Super key, `M-s g b')
         Set the region, or the next inserted character, to the `bold' face
         (`facemenu-set-bold').

-- 
    Robert J. Chassell                         Rattlesnake Enterprises
    http://www.rattlesnake.com                  GnuPG Key ID: 004B4AC8
    http://www.teak.cc                             bob@rattlesnake.com

Re: Suggestion: Mapping of M-g should be goto-line

[Stefan Monnier]

> I don't know of other environments than Sun and COTS PC's.
> On the PC's I run there is _always_ a Windows-key on the
> left-hand side of the keyboard between Ctrl and Alt. This key
> in XFree86 is mapped to Alt, whereas the actual Alt is mapped
> to Meta (or if it is the other way around).

Given the mess that Alt/Meta/mod1/mod2 are under X11, many people configure
they keyboard to only have Meta (or only Alt) so that applications
behave consistently.


        Stefan

Re: Suggestion: Mapping of M-g should be goto-line

[Miles Bader]

On Sun, Mar 28, 2004 at 03:32:37PM +0200, Joachim Nilsson wrote:
> >I don't think it is possible to type M-A- characters on most
> >keyboards.  How would you do it?
>
> I'm not suggesting this is the best solution, but it actually works

It's inappropriate to use this style for anything but optional short-cuts,
because there _are_ many cases where meta/alt functionality is messed up;
it's not at all rare under X to find someone has mapped all their `alt'
modifiers to meta, or the reverse (though probably an emacs user would have
done the former).

-Miles
-- 
In New York, most people don't have cars, so if you want to kill a person, you
have to take the subway to their house.  And sometimes on the way, the train
is delayed and you get impatient, so you have to kill someone on the subway.
  [George Carlin]

Re: Suggestion: Mapping of M-g should be goto-line

[Richard Stallman]

    I don't know of other environments than Sun and COTS PC's.
    On the PC's I run there is _always_ a Windows-key on the
    left-hand side of the keyboard between Ctrl and Alt. 

Not all PCs have such a key.  My laptop does not have one.  

Re: Suggestion: Mapping of M-g should be goto-line

[Kim F. Storm]

jari.aalto@poboxes.com (Jari Aalto+mail.emacs) writes:

>         The "de facto" situation for long has been that everybody maps
>         M-g to goto-line. People advice that in newsgroups, because it
>         is indeed the most logical key.
> 
>         If we would take a poll in Emacs newsgroups, presumably the
>         "yes" votes for goto-line would win in great majority.
> 
>         It would be better if Emacs supported this 99 % daily usage of
>         M-g, instead of current not-so-useful binding.

It occurred to me that you can have your cake and eat it too:

Try this little trick:

(defun goto-line-piggyback ()
  (interactive)
  (goto-line
   (string-to-int
    (read-string "Goto line: "
		 (substring (this-command-keys) -1)))))

(dolist (digit '(?1 ?2 ?3 ?4 ?5 ?6 ?7 ?8 ?9))
  (define-key facemenu-keymap (vector digit) 'goto-line-piggyback))


Eval and try M-g 123 RET


Or course, the feedback in the minibuffer when you hit M-g is far from
perfect, and C-h k M-g still gives odd results...

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk