Idea for compilation mode

Idea for compilation mode

[Richard Stallman]

It occurs to me that it might be an improvement to make M-x compile
parse the output as it arrives.  Currently there is a long delay in
parsing it the first time if there are many error messages, or many
error messages above point, and that can be annoying.  Parsing the
error messages as they come in would avoid the delay.

(I think that 15 years ago there was a good reason for the current
implementation, but I am not sure any more what it was.)

Re: Idea for compilation mode

[Stephan Stahl]

Richard Stallman <rms@gnu.org> wrote:

> It occurs to me that it might be an improvement to make M-x compile
> parse the output as it arrives.  Currently there is a long delay in
> parsing it the first time if there are many error messages, or many
> error messages above point, and that can be annoying.  Parsing the
> error messages as they come in would avoid the delay.

If someone changes this maybe it would be a good idea to change some other 
things to. Some menu entries (recompile, first error, previous error) could be 
made availible via keyboard.

An other feature i would like to see is to seperate errors and warnings. C-' 
could be used for "next-warning" and C-M-n / C-M-p for show next/previous 
warning.

And there also seems to be a small bug in compilation-next-error / compilation-
previous-error. When compilation has finished and i use M-< to go to the start 
of the buffer and then type M-n point goes to the second error not as one would 
expect to the first. Same thing for M-> and M-p. (GNU Emacs 21.3.1 (i386-msvc-
nt5.1.2600) of 2003-03-28 on buffy)

-- 
Stephan Stahl

Re: Idea for compilation mode

[David Kastrup]

Stephan Stahl <stl@isogmbh.de> writes:

> Richard Stallman <rms@gnu.org> wrote:
> 
> > It occurs to me that it might be an improvement to make M-x
> > compile parse the output as it arrives.  Currently there is a long
> > delay in parsing it the first time if there are many error
> > messages, or many error messages above point, and that can be
> > annoying.  Parsing the error messages as they come in would avoid
> > the delay.
> 
> If someone changes this maybe it would be a good idea to change some
> other things to. Some menu entries (recompile, first error, previous
> error) could be made availible via keyboard.

I want to point out that the next-error binding on C-x ` is awful to
type on most international keyboards because ` is typically a dead
key and only available with shift, to boot.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: Idea for compilation mode

[Miles Bader]

David.Kastrup@t-online.de (David Kastrup) writes:
> I want to point out that the next-error binding on C-x ` is awful to
> type on most international keyboards because ` is typically a dead
> key and only available with shift, to boot.

I've always use the old gosmacs binding for this -- `C-x C-n' -- which I
find very nice (even on a normal ascii keyboard, C-x ` is rather awkward
to type).

-Miles
-- 
Yo mama's so fat when she gets on an elevator it HAS to go down.

Re: Idea for compilation mode

[Kai Großjohann]

Stephan Stahl <stl@isogmbh.de> writes:

> If someone changes this maybe it would be a good idea to change some
> other things to. Some menu entries (recompile, first error, previous
> error) could be made availible via keyboard.

C-- C-x ` does like previous-error.

-- 
This line is not blank.

Re: Idea for compilation mode

[Frank Schmitt]

David.Kastrup@t-online.de (David Kastrup) writes:

> I want to point out that the next-error binding on C-x ` is awful to
> type on most international keyboards because ` is typically a dead
> key and only available with shift, to boot.

Strong ack.

-- 
WANTED: A nice signature
REWARD: none

Re: Idea for compilation mode

[Stephan Stahl]

Kai Großjohann <kai.grossjohann@gmx.net> wrote:

> C-- C-x ` does like previous-error.

Yes but i think in compile mode there are plenty of non used C-c keys.
Wouldn't it be nice to have C-c n for next error, C-c p for previous error, C-c 
a for first and C-c e for last error or something like that? As other said C-x 
` is difficult to type not to mention C-- C-x `.

And maybe some similar keys could be used for next/previous warning which i 
think would be very nice to have in compile mode.

-- 
Stephan Stahl

Re: Idea for compilation mode

[Kai Großjohann]

Stephan Stahl <stl@isogmbh.de> writes:

> Kai Großjohann <kai.grossjohann@gmx.net> wrote:
>
>> C-- C-x ` does like previous-error.
>
> Yes but i think in compile mode there are plenty of non used C-c keys.

Ah, you mean in the *compilation* buffer!  I'm sorry, I missed that.

I'd prefer `n' and `p' in that buffer.  In compilation-minor-mode,
other keys might have to be used.
-- 
This line is not blank.

Re: Idea for compilation mode

[Stephan Stahl]

Kai Großjohann <kai.grossjohann@gmx.net> wrote:

> Stephan Stahl <stl@isogmbh.de> writes:
> 
> > Yes but i think in compile mode there are plenty of non used C-c keys.
> 
> Ah, you mean in the *compilation* buffer!  I'm sorry, I missed that.
> 
> I'd prefer `n' and `p' in that buffer.  In compilation-minor-mode,
> other keys might have to be used.

Well sorry for not being clear enough.
I think that is a good idea. What do you think about extending the compile 
thing to treat errors and warnings different? I know that depends just on 
compilation-error-regexp-alist but when someone is already adding some keys 
maybe it would be worth to add the same for warnings to. That way one can look 
quickly for error in the compilation and when neccessary for warnings to get 
the code clean.

-- 
Stephan Stahl

Re: Idea for compilation mode

[Stefan Monnier]

> I think that is a good idea. What do you think about extending the compile 
> thing to treat errors and warnings different?

It's been on my todo list for ever.  I doubt I'll ever get to
working on it, tho, so I wish someone would do it for me ;-)


	Stefan

Re: Idea for compilation mode

[Kevin Rodgers]

Miles Bader wrote:

> David.Kastrup@t-online.de (David Kastrup) writes:
> 
>>I want to point out that the next-error binding on C-x ` is awful to
>>type on most international keyboards because ` is typically a dead
>>key and only available with shift, to boot.
>>
> 
> I've always use the old gosmacs binding for this -- `C-x C-n' -- which I
> find very nice (even on a normal ascii keyboard, C-x ` is rather awkward
> to type).

Excellent -- I've had this in my .emacs forever anyway:

(put 'set-goal-column 'disabled t)

And for symmetry, I'll bind `C-x C-p' to previous-error, then move
set-goal-column to `C-x N' and mark-page to `C-x P'.

-- 
<a href="mailto:&lt;kevin.rodgers&#64;ihs.com&gt;">Kevin Rodgers</a>

Re: Idea for compilation mode

[Richard Stallman]

    If someone changes this maybe it would be a good idea to change some other 
    things to. Some menu entries (recompile, first error, previous error) could be 
    made availible via keyboard.

These two issues are unrelated.  If you suggest specific bindings,
I can think about them.

David Kastrup wrote:

    I want to point out that the next-error binding on C-x ` is awful to
    type on most international keyboards because ` is typically a dead
    key and only available with shift, to boot.

I agree it is inconvenient, but we were already running out of keys
when I chose it.  What other key sequence would you suggest?

Miles wrote:

    I've always use the old gosmacs binding for this -- `C-x C-n' -- which I
    find very nice (even on a normal ascii keyboard, C-x ` is rather awkward
    to type).

I didn't put it there because I carried forward the original Emacs
meaning of C-x C-n.  However, it could well be that next-error is
used more than set-goal-column.  So perhaps that binding should be changed.

However, if next-error is C-x C-n, previous-error can't be C-x C-p.
It would be better to find a good pair of keys for next-error and
previous-error.

Re: Idea for compilation mode

[Miles Bader]

Richard Stallman <rms@gnu.org> writes:
>     I've always use the old gosmacs binding for this -- `C-x C-n' -- which I
>     find very nice (even on a normal ascii keyboard, C-x ` is rather awkward
>     to type).
> 
> I didn't put it there because I carried forward the original Emacs
> meaning of C-x C-n.  However, it could well be that next-error is
> used more than set-goal-column.  So perhaps that binding should be changed.
> 
> However, if next-error is C-x C-n, previous-error can't be C-x C-p.
> It would be better to find a good pair of keys for next-error and
> previous-error.

Hmm, well I of course use `C-x C-p' for previous-error, but I see that the
default binding is `mark-page'; is that a widely used command (I don't
think I use any page commands at all, except for forward/backward-page)?

I think that next/previous-error are quite widely used (not just
compilation, but grep, etc. also), and so deserve rather easy-to-type
bindings.

-Miles
-- 
80% of success is just showing up.  --Woody Allen

Re: Idea for compilation mode

[David Kastrup]

Miles Bader <miles@lsi.nec.co.jp> writes:

> Richard Stallman <rms@gnu.org> writes:

> >     I've always use the old gosmacs binding for this -- `C-x C-n'
> >     -- which I find very nice (even on a normal ascii keyboard,
> >     C-x ` is rather awkward to type).
> > 
> > I didn't put it there because I carried forward the original Emacs
> > meaning of C-x C-n.  However, it could well be that next-error is
> > used more than set-goal-column.  So perhaps that binding should be
> > changed.
> > 
> > However, if next-error is C-x C-n, previous-error can't be C-x
> > C-p.  It would be better to find a good pair of keys for
> > next-error and previous-error.
> 
> Hmm, well I of course use `C-x C-p' for previous-error, but I see that the
> default binding is `mark-page'; is that a widely used command (I don't
> think I use any page commands at all, except for forward/backward-page)?
> 
> I think that next/previous-error are quite widely used (not just
> compilation, but grep, etc. also), and so deserve rather
> easy-to-type bindings.

I agree here.  Previously we were talking about polls about
unused/obsoletable functions/Lisp files.  Storing those in an
"obsolete" directory or whatever else does not impact general Emacs
performance.

In contrast, unused keybindings hog user interface resources.  A lot
of bindings come from a time where there were few functions yet
available, so one could assign keybindings rather freely to them.
When new, much more convenient and more often used functions get
concocted, there are no easily accessible bindings left.  It would
probably be a stretch to make a connection to major Emacs developers
(like RMS and Ben Wing) suffering from RSI, but I think some
keybindings of Emacs might be worth reconsideration, and this might
require some polling.

We might need something similar to "disabled" commands for
keybindings, so that users get clued in about when some favorite
keybinding of theirs might be replaced in a future version, and maybe
also tell the user when an alternative binding already exists.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: Idea for compilation mode

[Stephan Stahl]

Richard Stallman <rms@gnu.org> wrote:

> These two issues are unrelated.  If you suggest specific bindings,
> I can think about them.

I was thinking of C-c r for recompile, C-c c for compile and C-c g or C-c s for 
compilation-mode-grep.

>     I've always use the old gosmacs binding for this -- `C-x C-n' -- which I
>     find very nice (even on a normal ascii keyboard, C-x ` is rather awkward
>     to type).
> 
> I didn't put it there because I carried forward the original Emacs
> meaning of C-x C-n.  However, it could well be that next-error is
> used more than set-goal-column.  So perhaps that binding should be changed.
> 
> However, if next-error is C-x C-n, previous-error can't be C-x C-p.
> It would be better to find a good pair of keys for next-error and
> previous-error.

I am not sure how often C-x C-p (mark-page) is used but maybe it could be moved 
to C-x p which is free at the moment and would be similar to C-x h (mark-whole-
buffer).

Then C-x C-p and C-x C-n could be used for next previous error if
set-goal-column is not so important to be on C-x C-n. But as someone pointed 
out a negative argument to next-error goes backward so maybe C-x C-n can be 
left untouched.

-- 
Stephan Stahl

Re: Idea for compilation mode

[Andreas Schwab]

Stephan Stahl <stl@isogmbh.de> writes:

|> Richard Stallman <rms@gnu.org> wrote:
|> 
|> > These two issues are unrelated.  If you suggest specific bindings,
|> > I can think about them.
|> 
|> I was thinking of C-c r for recompile, C-c c for compile and C-c g or C-c s for 
|> compilation-mode-grep.

C-c <letter> is reserved for the user.

Andreas.

-- 
Andreas Schwab, SuSE Labs, schwab@suse.de
SuSE Linux AG, Deutschherrnstr. 15-19, D-90429 Nürnberg
Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: Idea for compilation mode

[Richard Stallman]

    Yes but i think in compile mode there are plenty of non used C-c keys.

C-x ` can be used in all modes; you don't have to go to the
compilation buffer to use it.

Re: Idea for compilation mode

[Richard Stallman]

    I think that is a good idea. What do you think about extending the compile 
    thing to treat errors and warnings different? I know that depends just on 
    compilation-error-regexp-alist but when someone is already adding some keys 
    maybe it would be worth to add the same for warnings to. That way one can look 
    quickly for error in the compilation and when neccessary for warnings to get 
    the code clean.

It sounds like a good feature.  I would suggest having a variable
that controls whether warnings are ignored.

Re: Idea for compilation mode

[Stephan Stahl]

Andreas Schwab <schwab@suse.de> wrote:

> Stephan Stahl <stl@isogmbh.de> writes:
> 
> |> I was thinking of C-c r for recompile, C-c c for compile and C-c g or C-c
> s for 
> |> compilation-mode-grep.
> 
> C-c <letter> is reserved for the user.

I did not know that. Then what about C-c C-r, C-c C-c, C-c C-g keys (in the 
compilation buffer). They are free, don't hurt anyone and i think they are 
better than M-x recom-TAB RET (recompile) for example.
Well they are easier to hit then i thought :-).



Richard Stallman <rms@gnu.org> wrote:

>     Yes but i think in compile mode there are plenty of non used C-c keys.
> 
> C-x ` can be used in all modes; you don't have to go to the
> compilation buffer to use it.

Richard Stallman <rms@gnu.org> wrote:

>     I think that is a good idea. What do you think about extending the
>     compile thing to treat errors and warnings different? I know that depends
>     just on compilation-error-regexp-alist but when someone is already adding
>     some keys maybe it would be worth to add the same for warnings to. That
>     way one can look quickly for error in the compilation and when neccessary
>     for warnings to get the code clean.
> 
> It sounds like a good feature.  I would suggest having a variable
> that controls whether warnings are ignored.

My point was not to replace C-x ` but to add some keybindings to make compile 
mode better. When compiling i think one is often in the compile buffer and 
could surely use this bindings.
As someothers pointed out C-x ` is very difficult to hit even on normal 
keyboards so some alternative would be nice even if it is just in the compile 
buffer.

That way there could even be a binding for toggling that variable or different 
keys for errors and warnings. Maybe C-c C-w to toggle warnings of/off. C-M-p 
and C-M-n for previous and next warning.

-- 
Stephan Stahl

Re: Idea for compilation mode

[Richard Stallman]

    I am not sure how often C-x C-p (mark-page) is used but maybe it could be moved 
    to C-x p which is free at the moment and would be similar to C-x h (mark-whole-
    buffer).

That seems like a plausible change.

Re: Idea for compilation mode

[Andreas Schwab]

Stephan Stahl <stl@isogmbh.de> writes:

|> My point was not to replace C-x ` but to add some keybindings to make compile 
|> mode better. When compiling i think one is often in the compile buffer and 
|> could surely use this bindings.

When you are inside the compilation buffer you can hit RET to jump to an
error.

Andreas.

-- 
Andreas Schwab, SuSE Labs, schwab@suse.de
SuSE Linux AG, Deutschherrnstr. 15-19, D-90429 Nürnberg
Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: Idea for compilation mode

[Stephan Stahl]

Andreas Schwab <schwab@suse.de> wrote:

> Stephan Stahl <stl@isogmbh.de> writes:
> 
> |> My point was not to replace C-x ` but to add some keybindings to make
> compile 
> |> mode better. When compiling i think one is often in the compile buffer and
> 
> |> could surely use this bindings.
> 
> When you are inside the compilation buffer you can hit RET to jump to an
> error.

When i first replied to this thread my idea was only to add a few key bindings 
to compile mode. Like C-c C-r for recompile, C-c C-g for grep and C-c C-c for 
compile. And to extend compile mode to handle errors and warnings different and 
to provide key bindings for those.

Someone else started that C-x ` -replacement-thing. I do not have much of a 
opinion about that. It was suggestest to use C-x C-n instead if only few people 
want to use its current binding set-goal-column. I added that it then would be 
nice to use C-x C-p  for previous-error. C-x C-p has a binding at the moment 
(mark-page) which could/should be moved to C-x p which is free at the moment 
and which would then be similar to C-x h (mark-whole-buffer).

So i have to say thanks for pointing out to use RET but i know that allready. 
Maybe my previous messages were not clear enough. Sorry for the confusion.

-- 
Stephan Stahl

Re: Idea for compilation mode

[Stefan Monnier]

> I did not know that. Then what about C-c C-r, C-c C-c, C-c C-g keys (in the 

C-h is for `abort' so it's better to avoid it in a key-binding.
I suggest you read the elisp manual section about coding conventions.


	Stefan

Re: Idea for compilation mode

[Lute Kamstra]

"Stefan Monnier" <monnier+gnu/emacs@rum.cs.yale.edu> writes:

>> I did not know that. Then what about C-c C-r, C-c C-c, C-c C-g keys (in the 
>
> C-h is for `abort' so it's better to avoid it in a key-binding.
  ^^^
  C-g


Lute.

Re: Idea for compilation mode

[Miles Bader]

On Thu, Jun 05, 2003 at 04:00:03PM +0200, Lute Kamstra wrote:
> > C-h is for `abort' so it's better to avoid it in a key-binding.
>   ^^^
>   C-g

You know that weird Stefan, he's rebound all his keys in funny ways...

-Miles
-- 
"1971 pickup truck; will trade for guns"

Re: Idea for compilation mode

[Stefan Monnier]

> > > C-h is for `abort' so it's better to avoid it in a key-binding.
> >   ^^^
> >   C-g
> 
> You know that weird Stefan, he's rebound all his keys in funny ways...

Nah, my hand was hit by one of those bursts of alpha thingies.


	Stefan

Re: Idea for compilation mode

[Miles Bader]

Richard Stallman <rms@gnu.org> writes:
>     I am not sure how often C-x C-p (mark-page) is used but maybe it
>     could be moved to C-x p which is free at the moment and would be
>     similar to C-x h (mark-whole- buffer).
> 
> That seems like a plausible change.

Yeah, that's a good solution -- not only is `C-x p' slightly more
intuitive because it's a bit closer to `C-x h', but also previous-error
makes better use of the fact that `C-x C-p' is easy to type repeatedly
(because you can just hold down the control key and type `x p x p ...').
There's obviously little point in using mark-page more than once in row. :-)

There seems to be something of a consensus developing in this thread for
using `C-x C-n' and `C-x C-p' for next/previous-error, moving mark-page
to `C-x p', and simply removing the default binding for set-goal-column.

Are there any _objections_ to such a change?

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

Re: Idea for compilation mode

[David Kastrup]

Miles Bader <miles@lsi.nec.co.jp> writes:

> There seems to be something of a consensus developing in this thread
> for using `C-x C-n' and `C-x C-p' for next/previous-error, moving
> mark-page to `C-x p', and simply removing the default binding for
> set-goal-column.
> 
> Are there any _objections_ to such a change?

Just remember to update manual and reference card, too.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: Idea for compilation mode

[Stephan Stahl]

Miles Bader <miles@lsi.nec.co.jp> wrote:

> There seems to be something of a consensus developing in this thread for
> using `C-x C-n' and `C-x C-p' for next/previous-error, moving mark-page
> to `C-x p', and simply removing the default binding for set-goal-column.

If set-goal-column should be on a key maybe C-x , can be used. Thats not far 
away from the old C-x C-n and C-x . which is directly beside would be somehow 
similar in function (set-fill-prefix).
I think we should stop soon or we'll end up remapping all keys :-).

> Are there any _objections_ to such a change?

I don't have any.

-- 
Stephan Stahl

Re: Idea for compilation mode

[David Kastrup]

Stephan Stahl <stl@isogmbh.de> writes:

> Miles Bader <miles@lsi.nec.co.jp> wrote:
> 
> > There seems to be something of a consensus developing in this
> > thread for using `C-x C-n' and `C-x C-p' for next/previous-error,
> > moving mark-page to `C-x p', and simply removing the default
> > binding for set-goal-column.
> 
> If set-goal-column should be on a key maybe C-x , can be used. Thats
> not far away from the old C-x C-n and C-x . which is directly beside
> would be somehow similar in function (set-fill-prefix).  I think we
> should stop soon or we'll end up remapping all keys :-).

Which would not be the worst thing.  As I said, it would be a viable
task for the TODO list to go through the existing, historically grown
keymappings with a fine comb, try to compare the probable usefulness
of mappings with their mnemonic characteristics and their awkwardness
(taking non-American keyboard layouts somewhat into mind as well) or
non-existence, and try to develop something better suited for
accessing the currently available functionality.

Language reform processes suffer a similar combination of growing
necessity, getting rid of obsolescent features that have fallen into
disuse, resistance from people accustomed to the old bindings, polling
people for what they commonly use for circumventing perceived
shortcomings and so on.

As I said, TODO list material.  A good task, but needing considerable
effort if done right and thoroughly.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: Idea for compilation mode

[Stephan Stahl]

David Kastrup <David.Kastrup@t-online.de> wrote:

> Stephan Stahl <stl@isogmbh.de> writes:
> 
> > If set-goal-column should be on a key maybe C-x , can be used. Thats
> > not far away from the old C-x C-n and C-x . which is directly beside
> > would be somehow similar in function (set-fill-prefix).  I think we
> > should stop soon or we'll end up remapping all keys :-).
> 
> Which would not be the worst thing.  As I said, it would be a viable
> task for the TODO list to go through the existing, historically grown
> keymappings with a fine comb, try to compare the probable usefulness
> of mappings with their mnemonic characteristics and their awkwardness
> (taking non-American keyboard layouts somewhat into mind as well) or

Taking non-American keyboard layouts into mind seems like a very complicated 
thing. I myself use a us-international layout because most others layouts i 
have tried are horrible to use with applications like emacs or maybe vi.
For example the german keyboard has not two alt keys but one alt and one altGr 
keys. To produce ] one would have to type altGr-9. It is not possible to type
C-] (aka ESC) on a german keyboard because it would require both ctrl and altGr 
to be pressed. At least i never found a keyboard / operating system where it is 
possible. Besides it would be very hard to hit all three keys at once.
I belive most other european keyboard layouts are somewhat similar in that 
aspect.

-- 
Stephan Stahl

Re: Idea for compilation mode

[Juanma Barranquero]

On Fri,  6 Jun 2003 14:01:52 +0200
Stephan Stahl <stl@isogmbh.de> wrote:

> Taking non-American keyboard layouts into mind seems like a very complicated 
> thing.

OTOH...

> I myself use a us-international layout because most others layouts i 
> have tried are horrible to use with applications like emacs or maybe vi.

...some of us use non-american keyboard layouts and Emacs. I've used it
with a Brazilian Portuguese layout and I'm now working with a Spanish
one.

> For example the german keyboard has not two alt keys but one alt and one altGr 
> keys.

Here's the same.

> To produce ] one would have to type altGr-9.

I do AltGr +. And I don't find it particularly cumbersome (yes, I
program in C, C++, Perl and other languages where [] are heavily used).

> It is not possible to type
> C-] (aka ESC) on a german keyboard because it would require both ctrl and altGr 
> to be pressed.

To do C-] I have to type AltGr Right-Ctrl +, in that precise order. And
left ctrl won't do :)

I've also remapped C-Ñ to insert ~ (although I can get it too with AltGr
4), and AltGr e to insert €.

> Besides it would be very hard to hit all three keys at once.

Do you do C-] very often?


                                                                Juanma

Re: Idea for compilation mode

[Stephan Stahl]

Juanma Barranquero <jmbarranquero@laley.wke.es> wrote:

> Stephan Stahl <stl@isogmbh.de> wrote:
> 
> > To produce ] one would have to type altGr-9.
> 
> I do AltGr +. And I don't find it particularly cumbersome (yes, I
> program in C, C++, Perl and other languages where [] are heavily used).

I am not sure where + is on a spanish keyboard but i find AtlGr-9 is very hard 
to hit on a german keyboard. AlrGr is on the right side and 9 too so one would 
have to completly leave the homerow with the right hand. very bad. Or use two 
hands in which case you have to find the homerow for two hands.. very very bad.

> > It is not possible to type
> > C-] (aka ESC) on a german keyboard because it would require both ctrl and
> > altGr to be pressed.
> 
> To do C-] I have to type AltGr Right-Ctrl +, in that precise order. And
> left ctrl won't do :)

I have tried it and you are right thanks! I would have never tried this because 
normaly i press the modifier with one hand and the key with the other.
Then again RightCtrl-AltGr-9 is almost imposible to do when typing fluidly?

> > Besides it would be very hard to hit all three keys at once.
> 
> Do you do C-] very often?

Well actually that should have been C-[ which i often use when Meta / Alt is 
not availible when using some bad graphical ssh client. But the same goes for
{ [ ] } \ ~  which are on 7 8 9... Those are used more often in emacs.

C-c C-\ c-backslash-region
C-M-\ indent-region
M-} dired-next-marked-file
M-{ dired-prev-marked-file
M-\ delete-horizontal-space
C-\ toggle-input-method
M-~ not-modified

come to mind.

Or imagine someone using viper mode. I like C-[ aka ESC much more on an 
american keyboard layout than ESC on any other keyboard. But on a german 
keyboard both are not very good IMO.

That is just to point out someone who wants to look for better bindings and 
wants to keep non american keyboards in mind has a very difficult job to do.

-- 
Stephan Stahl

Re: Idea for compilation mode

[David Kastrup]

Stephan Stahl <stl@isogmbh.de> writes:

> David Kastrup <David.Kastrup@t-online.de> wrote:
> 
> > Stephan Stahl <stl@isogmbh.de> writes:
> > 
> > > If set-goal-column should be on a key maybe C-x , can be used. Thats
> > > not far away from the old C-x C-n and C-x . which is directly beside
> > > would be somehow similar in function (set-fill-prefix).  I think we
> > > should stop soon or we'll end up remapping all keys :-).
> > 
> > Which would not be the worst thing.  As I said, it would be a viable
> > task for the TODO list to go through the existing, historically grown
> > keymappings with a fine comb, try to compare the probable usefulness
> > of mappings with their mnemonic characteristics and their awkwardness
> > (taking non-American keyboard layouts somewhat into mind as well) or
> 
> Taking non-American keyboard layouts into mind seems like a very
> complicated thing.

It's not something you could hope to achieve completely, but there
are keys that are notably difficult to type on a lot of keyboards. `
and @, for example.  Not using those for the most common
functionalities can help quite a bit.

> I myself use a us-international layout because most others layouts i
> have tried are horrible to use with applications like emacs or maybe
> vi.  For example the german keyboard has not two alt keys but one
> alt and one altGr keys. To produce ] one would have to type
> altGr-9. It is not possible to type C-] (aka ESC) on a german
> keyboard because it would require both ctrl and altGr to be
> pressed.

C-] is not ESC, but GS.  C-[ is ESC, and pressing both ctrl and altGr
is quite possible on a German keyboard.

> At least i never found a keyboard / operating system where
> it is possible. Besides it would be very hard to hit all three keys
> at once.

Why would you need to do that?  First you press the modifiers, then
you press the key itself.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: Idea for compilation mode

[Stephan Stahl]

David Kastrup <David.Kastrup@t-online.de> wrote:

> Stephan Stahl <stl@isogmbh.de> writes:
> 
> C-] is not ESC, but GS.  C-[ is ESC, and pressing both ctrl and altGr
> is quite possible on a German keyboard.

You are right with both.
My fingers just hit the wrong key i meant C-[.
But it is not enough that you can press ctrl and altGr together. As Juanma 
Barranquero pointed out it has to be in a particular order so that C-[ works. 
Remember [ is produced by AltGr-9 so for C-[ on a german keyboard one would 
have to press rightCtrl-AltGr-9. Not matter what other say i find it almost 
impossible to do.

> > At least i never found a keyboard / operating system where
> > it is possible. Besides it would be very hard to hit all three keys
> > at once.
> 
> Why would you need to do that?  First you press the modifiers, then
> you press the key itself.

Well not exactly at once but together. (I should speak and read more englisch 
to avoid those anoying mistakes which just confuse others :-).
I think it is very hard to press rightCtrl-AltGr-9 together to get C-[. But 
that was just a example see my other mail for more.

-- 
Stephan Stahl

Re: Idea for compilation mode

[Juanma Barranquero]

On Fri,  6 Jun 2003 15:40:04 +0200, Stephan Stahl <stl@isogmbh.de> wrote:

> I am not sure where + is on a spanish keyboard but i find AtlGr-9 is very hard 
> to hit on a german keyboard. AlrGr is on the right side and 9 too so one would 
> have to completly leave the homerow with the right hand.

Second row is QWERTYUIOP`+ so yes, + is on the right. I do AltGr + all
the time with the right hand and I don't find it uncomfortable.

> Then again RightCtrl-AltGr-9 is almost imposible to do when typing fluidly?

Yes. But I've never had the need of typing C-] fluidly, usually my slow
way has been enough :)

> C-c C-\ c-backslash-region    => C-c AltGr Right-Ctrl \
> C-M-\ indent-region           => AltGr Right-Ctrl Alt \
> M-} dired-next-marked-file 
> M-{ dired-prev-marked-file
> M-\ delete-horizontal-space   => AltGr Alt \
> C-\ toggle-input-method       => AltGr Right-Ctrl \
> M-~ not-modified              => AltGr Alt 4 4

So I can type most of them (I've not found equivalent combinations for
M-{}).

> That is just to point out someone who wants to look for better bindings and 
> wants to keep non american keyboards in mind has a very difficult job to do.

And my point is that some of us just *use* a national keyboard, and
we're able to make do, if sometimes in a very convoluted way
(fortunately, most of these keys I don't use, and the ones I use I can
rebind to Hyper-something, where Hyper is my left "window" key).

So national keyboards is something that we must live with, and finding
easier keybindings for some of those commands would be *good*.

                                                           /L/e/k/t/u

Re: Idea for compilation mode

[Frank Schmitt]

Stephan Stahl <stl@isogmbh.de> writes:

> C-M-\ indent-region

This one is really especially horrible to type.

-- 
WANTED: A nice signature
REWARD: none

Re: Idea for compilation mode

[Kai Großjohann]

Stephan Stahl <stl@isogmbh.de> writes:

> My point was not to replace C-x ` but to add some keybindings to make compile 
> mode better. When compiling i think one is often in the compile buffer and 
> could surely use this bindings.

M-n and M-p move between errors in the *compilation* buffer.
-- 
This line is not blank.

Re: Idea for compilation mode

[Richard Stallman]

I'm not interested in making any major changes in Emacs key bindings
for the sake of non-US keyboards.  It would be too much trouble.
However, it might perhaps make sense to support optional alternate
binding themes for some of these layouts.  Those could be maintained
by users of those keyboard layouts, perhaps using key-translation-map
to make them independent of the rest of Emacs.  That way, they won't
be a maintenance burden.

Re: Idea for compilation mode

[David Kastrup]

Richard Stallman <rms@gnu.org> writes:

> I'm not interested in making any major changes in Emacs key bindings
> for the sake of non-US keyboards.  It would be too much trouble.

Naturally.  It is just a thing to keep in mind for creating keymaps
that some keys are often worse to reach than others, when switching to
a different language.  ` is an example for a key that is horrible to
type on almost any keyboard except US ones, and so it should be
avoided for important key bindings like next-error.

> However, it might perhaps make sense to support optional alternate
> binding themes for some of these layouts.  Those could be maintained
> by users of those keyboard layouts, perhaps using
> key-translation-map to make them independent of the rest of Emacs.
> That way, they won't be a maintenance burden.

That would be overkill, I guess.  Most keybindings for Emacs are
designed not for ergonomicity, but for their mnemonic value.  For
example, C-b C-f C-p C-n make just as little sense on a Dvorak
keyboard as they do on an American.  This is in stark difference to
the bindings of vi, where the cursor movements hjkl very much depend
on a Qwert-based keyboard layout in order to work.

In my opinion, national keyboard themes would not be worth the trouble
for Emacs, but keeping in mind keys like "`@\" and a few others as
being awkward in particular in connection with modifier keys will be a
boon for many users.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: Idea for compilation mode

[Juanma Barranquero]

On Sat, 07 Jun 2003 21:09:01 -0400, Richard Stallman <rms@gnu.org> wrote:

> I'm not interested in making any major changes in Emacs key bindings
> for the sake of non-US keyboards.

I was not proposing major changes, of course. Perhaps alternate
keybindings for the more difficult ones to type today (M-C-\, things
with ` and stuff like that).

> However, it might perhaps make sense to support optional alternate
> binding themes for some of these layouts.

Yes, that'd be great.

                                                           /L/e/k/t/u

Re: Idea for compilation mode

[Stephan Stahl]

Juanma Barranquero <lektu@terra.es> wrote:

> > C-M-\ indent-region           => AltGr Right-Ctrl Alt \

I would need a third hand to do that one :-).

> And my point is that some of us just *use* a national keyboard, and
> we're able to make do, if sometimes in a very convoluted way
> (fortunately, most of these keys I don't use, and the ones I use I can
> rebind to Hyper-something, where Hyper is my left "window" key).
> 
> So national keyboards is something that we must live with, and finding
> easier keybindings for some of those commands would be *good*.

It is a good point. Since i use a us style keyboard layout maybe some other 
could bring up some ideas for new bindings?

It would also be nice if scustomizable key bindings were availible so emacs 
could use custom-themes for national keyboards.

-- 
Stephan Stahl

Re: Idea for compilation mode

[Stephan Stahl]

Juanma Barranquero <lektu@terra.es> wrote:

> On Sat, 07 Jun 2003 21:09:01 -0400, Richard Stallman <rms@gnu.org> wrote:
> 
> > I'm not interested in making any major changes in Emacs key bindings
> > for the sake of non-US keyboards.
> 
> I was not proposing major changes, of course. Perhaps alternate
> keybindings for the more difficult ones to type today (M-C-\, things
> with ` and stuff like that).

A list of difficult keys and proposals for alternatives would be nice.
I think
~ ` @ ^ |
are some of those troublemakers. Any ideas for new bindings?

-- 
Stephan Stahl

Re: Idea for compilation mode

[David Kastrup]

Stephan Stahl <stl@isogmbh.de> writes:

> Juanma Barranquero <lektu@terra.es> wrote:
> 
> > On Sat, 07 Jun 2003 21:09:01 -0400, Richard Stallman <rms@gnu.org> wrote:
> > 
> > > I'm not interested in making any major changes in Emacs key bindings
> > > for the sake of non-US keyboards.
> > 
> > I was not proposing major changes, of course. Perhaps alternate
> > keybindings for the more difficult ones to type today (M-C-\, things
> > with ` and stuff like that).
> 
> A list of difficult keys and proposals for alternatives would be nice.
> I think
> ~ ` @ ^ |
> are some of those troublemakers. Any ideas for new bindings?

At least in German, the dead ones are the more problematic ones, ~ `,
maybe ^.  ~ and ` are rarely used in programming, too.  | is not that
bad and in the usual keybindings it has mnemonic value for pipes.

At least M-C-` and M-C-~ are not currently bound...

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: Idea for compilation mode

[Juanma Barranquero]

On 11 Jun 2003 10:51:08 +0200
David.Kastrup@t-online.de (David Kastrup) wrote:

`> At least in German, the dead ones are the more problematic ones, ~ `,
> maybe ^.

Yes. Every combination of C and/or M involving one of \ ` ^ [ ] { } | @
# ~ is difficult to type in a Spanish (and I suppose, any non-US)
keyboard.

> At least M-C-` and M-C-~ are not currently bound...

And fortunately so. Both are four-finger wonders...


                                                                Juanma