page-at-a-time output for M-x shell

page-at-a-time output for M-x shell

[Francis Moreau]

Hello,

M-x term has a nice feature 'page-at-a-time' which I'd like to have in
M-x shell specially because commands like

   $ cmd-with-a-lof-of-output | less
   WARNING: terminal is not fully functional
    -  (press RETURN)

doesn't work properly.

Does anybody know some trick here ?

Thanks

Re: page-at-a-time output for M-x shell

[Pascal J. Bourguignon]

Francis Moreau <francis.moro@gmail.com> writes:

> Hello,
>
> M-x term has a nice feature 'page-at-a-time' which I'd like to have in
> M-x shell specially because commands like
>
>    $ cmd-with-a-lof-of-output | less
>    WARNING: terminal is not fully functional
>     -  (press RETURN)
>
> doesn't work properly.
>
> Does anybody know some trick here ?

I've got:

----(~/.emacs_bash)-----------------------------------------------------
unset LC_MONETARY
unset LC_NUMERIC
unset LC_MESSAGES
unset LC_COLLATE
unset LC_CTYPE
unset LC_TIME
export LC_CTYPE=en_US.UTF-8 

export PAGER=cat 
export TERM=emacs 
alias less=cat 
alias more=cat 
stty rows 10000 columns 80 
echo '# welcome to emacs shell' 
------------------------------------------------------------------------

Then I use PgUp and PgDn after a cmd-with-a-lot-of-output


-- 
__Pascal Bourguignon__

Re: page-at-a-time output for M-x shell

[Peter Dyballa]

Am 11.07.2009 um 15:14 schrieb Francis Moreau:

> Does anybody know some trick here ?


I'm leaving away the ' | less' part.

--
Greetings

   Pete

Who the fsck is "General Failure," and why is he reading my disk?

Re: page-at-a-time output for M-x shell

[Peter Dyballa]

Am 11.07.2009 um 15:56 schrieb Pascal J. Bourguignon:

> Then I use PgUp and PgDn after a cmd-with-a-lot-of-output


Well, PgUp and PgDn make it necessary that you read. GNU Emacs can it  
better and faster! So I just let it search for something when I've  
found some time…

--
Greetings

   Pete

They're putting dimes in the hole in my head to see the change in me.

Re: page-at-a-time output for M-x shell

[Sumit Narayan]

I usually use C-r to return to the beginning of the previous command.
Works fine on shell and eshell. C-e will take you back to the prompt.

Sumit.



On Sat, Jul 11, 2009 at 10:04 AM, Peter Dyballa<Peter_Dyballa@web.de> wrote:
>
> Am 11.07.2009 um 15:56 schrieb Pascal J. Bourguignon:
>
>> Then I use PgUp and PgDn after a cmd-with-a-lot-of-output
>
>
> Well, PgUp and PgDn make it necessary that you read. GNU Emacs can it better
> and faster! So I just let it search for something when I've found some time…
>
> --
> Greetings
>
>  Pete
>
> They're putting dimes in the hole in my head to see the change in me.
>
>
>
>
>
>

Re: page-at-a-time output for M-x shell

[Pascal J. Bourguignon]

Peter Dyballa <Peter_Dyballa@Web.DE> writes:

> Am 11.07.2009 um 15:56 schrieb Pascal J. Bourguignon:
>
>> Then I use PgUp and PgDn after a cmd-with-a-lot-of-output
>
>
> Well, PgUp and PgDn make it necessary that you read. 

Which is what the OP wanted to do, otherwise he would have used |grep
'target' instead of |less.

> GNU Emacs can it
> better and faster! So I just let it search for something when I've
> found some time…


-- 
__Pascal Bourguignon__

Re: page-at-a-time output for M-x shell

[Francis Moreau]

On Jul 11, 4:19 pm, Peter Dyballa <Peter_Dyba...@Web.DE> wrote:
> Am 11.07.2009 um 15:14 schrieb Francis Moreau:
>
> > Does anybody know some trick here ?
>
> I'm leaving away the ' | less' part.
>

Hm, I'm probably missing your point, but I don't see how does that
solve my issue...

I'd want to use a pager with 'M-x shell' but ' | $PAGER' doesn't work
nicely in emacs shell. So if I loose the '| $PAGER' part, I loose the
functionnality I want to have as well...

Re: page-at-a-time output for M-x shell

[Francis Moreau]

Hello,

On Jul 11, 3:56 pm, p...@informatimago.com (Pascal J. Bourguignon)
wrote:

[...]

> Then I use PgUp and PgDn after a cmd-with-a-lot-of-output

That's not what I want because it's not convenient otherwise PAGER in
all environment would be almost useless.

Re: page-at-a-time output for M-x shell

[Teemu Likonen]

On 2009-07-11 06:14 (-0700), Francis Moreau wrote:

> M-x term has a nice feature 'page-at-a-time' which I'd like to have in
> M-x shell specially because commands like
>
>    $ cmd-with-a-lof-of-output | less
>    WARNING: terminal is not fully functional
>     -  (press RETURN)
>
> doesn't work properly.
>
> Does anybody know some trick here ?

Emacs shell is a "dump" terminal which doesn't support (all) terminal
control codes. You need to use "M-x term" if you want a proper terminal
(but even that is not as good as the good old XTerm).

Re: page-at-a-time output for M-x shell

[Richard Riley]

Teemu Likonen <tlikonen@iki.fi> writes:

> On 2009-07-11 06:14 (-0700), Francis Moreau wrote:
>
>> M-x term has a nice feature 'page-at-a-time' which I'd like to have in
>> M-x shell specially because commands like
>>
>>    $ cmd-with-a-lof-of-output | less
>>    WARNING: terminal is not fully functional
>>     -  (press RETURN)
>>
>> doesn't work properly.
>>
>> Does anybody know some trick here ?
>
> Emacs shell is a "dump" terminal which doesn't support (all) terminal
> control codes. You need to use "M-x term" if you want a proper terminal
> (but even that is not as good as the good old XTerm).

http://www.emacswiki.org/emacs/download/multi-term.el

Re: page-at-a-time output for M-x shell

[Peter Dyballa]

Am 12.07.2009 um 14:20 schrieb Francis Moreau:

> I'd want to use a pager with 'M-x shell'

Why? Is it so hard for you to understand that GNU Emacs is no  
terminal emulation?

--
Greetings

   Pete

I wouldn't recommend sex, drugs or insanity for everyone, but they've  
always worked for me.
				– Hunter S. Thompson

Re: page-at-a-time output for M-x shell

[Richard Riley]

Peter Dyballa <Peter_Dyballa@Web.DE> writes:

> Am 12.07.2009 um 14:20 schrieb Francis Moreau:
>
>> I'd want to use a pager with 'M-x shell'
>
> Why? Is it so hard for you to understand that GNU Emacs is no  
> terminal emulation?

I'm not sure I understand you.

You can use a pager such as less in a multi-term. Why is he so wrong to
enquire about just that?

Re: page-at-a-time output for M-x shell

[Thierry Volpiatto]

Hi all,
If you use eshell you can call easily term for the commands that need it:

,----
| (when (require 'em-term)
|   (dolist (i '("kop" "ledger" "mc" "htop"))
|     (add-to-list 'eshell-visual-commands i)))
`----

This code add the commands kop, ledger, mc, htop to
`eshell-visual-commands'.
That mean when you run one of these commands in eshell, they are
executed in term.
At the end the term buffer is killed and you are back in eshell.

Richard Riley <rileyrgdev@gmail.com> writes:

> Teemu Likonen <tlikonen@iki.fi> writes:
>
>> On 2009-07-11 06:14 (-0700), Francis Moreau wrote:
>>
>>> M-x term has a nice feature 'page-at-a-time' which I'd like to have in
>>> M-x shell specially because commands like
>>>
>>>    $ cmd-with-a-lof-of-output | less
>>>    WARNING: terminal is not fully functional
>>>     -  (press RETURN)
>>>
>>> doesn't work properly.
>>>
>>> Does anybody know some trick here ?
>>
>> Emacs shell is a "dump" terminal which doesn't support (all) terminal
>> control codes. You need to use "M-x term" if you want a proper terminal
>> (but even that is not as good as the good old XTerm).
>
> http://www.emacswiki.org/emacs/download/multi-term.el
>

-- 
A + Thierry Volpiatto
Location: Saint-Cyr-Sur-Mer - France

Re: page-at-a-time output for M-x shell

[Pascal J. Bourguignon]

Francis Moreau <francis.moro@gmail.com> writes:

> Hello,
>
> On Jul 11, 3:56 pm, p...@informatimago.com (Pascal J. Bourguignon)
> wrote:
>
> [...]
>
>> Then I use PgUp and PgDn after a cmd-with-a-lot-of-output
>
> That's not what I want because it's not convenient otherwise PAGER in
> all environment would be almost useless.

If your purpose is to use a pager, then use M-x term RET C-c C-q 
instead.

On the other hand, if your purpose is to read the output of a command,
then use M-x shell and PgUp / PgDn.

-- 
__Pascal Bourguignon__

Re: page-at-a-time output for M-x shell

[Francis Moreau]

On Jul 12, 3:45 pm, Peter Dyballa <Peter_Dyba...@Web.DE> wrote:
> Am 12.07.2009 um 14:20 schrieb Francis Moreau:
>
> > I'd want to use a pager with 'M-x shell'
>
> Why? Is it so hard for you to understand that GNU Emacs is no  
> terminal emulation?
>

Peter,

From what I saw, your posts are usually helpful but here I must admit
that it's totaly pointeless and useless.

If you don't know or if emacs don't know, then simply say it, I can
certainly live with that.

Is Gnu Emacs a news reader ? a erc client ? ...
No, but guess what, some people wrote some nice package to extend
emacs functionnality and they work.

Oh some other people extend emacs to add shell functionality in
Emacs ! And why is it so hard to understand that a pager feature would
be nice in this package ?

Some other Emacs commands already do that, M-x shell get the output of
the command and put it in the buffer, so I would guess that it's not
that hard to tell Emacs to display one page of output at time but it's
just a guess...

Cheers

Re: page-at-a-time output for M-x shell

[Francis Moreau]

On Jul 12, 4:12 pm, Richard Riley <rileyrg...@gmail.com> wrote:
>
> You can use a pager such as less in a multi-term. Why is he so wrong to
> enquire about just that?

I don't know either.

One weird thing, is that 'M-x term' is a term emulation where less(1)
works and it has the 'C-c C-q' one-page-at-a-time thing as well...

M-x shell, has neither.

Re: page-at-a-time output for M-x shell

[Teemu Likonen]

On 2009-07-12 10:40 (-0700), Francis Moreau wrote:

> Some other Emacs commands already do that, M-x shell get the output of
> the command and put it in the buffer, so I would guess that it's not
> that hard to tell Emacs to display one page of output at time but it's
> just a guess...

Please understand that "less" needs a _terminal_. Emacs's shell-mode is
not a terminal so "less" doesn't work. If the programs you use needs
terminal features then you must use a terminal, such as "M-x term".

Re: page-at-a-time output for M-x shell

[Peter Dyballa]

Am 12.07.2009 um 19:40 schrieb Francis Moreau:

> And why is it so hard to understand that a pager feature would
> be nice in this package ?


Because I never needed it, not even when working 20 years ago at an  
alpha-numeric hardware terminal. And I have no imagination to invent  
circumstances that I would need it.

If the text cursor is at the end of the *shell* buffer and is pushed  
away by fast effervescent output, faster than you can read, it's one  
command to return to the prompt. It's easier to send the command and  
make the cursor go up a line.


GNU Emacs is no terminal emulation. It does not need a pager. These  
programmes were developed for hardware terminals from 20 or more  
years ago. There is no sense in keeping them today, in a time of  
feature full terminal emulations. They can be abolished together with  
atom reactors and bombs. And your splendid idea.

--
Greetings

   Pete

We also sponsor National Invisible Chronic Illness Awareness Week  
annually in September.
Join the millions

Re: page-at-a-time output for M-x shell

[tomas]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Sun, Jul 12, 2009 at 09:41:47PM +0200, Peter Dyballa wrote:
>
> Am 12.07.2009 um 19:40 schrieb Francis Moreau:
>
>> And why is it so hard to understand that a pager feature would
>> be nice in this package ?

[...]

> GNU Emacs is no terminal emulation. It does not need a pager. These 
> programmes were developed for hardware terminals from 20 or more years ago. 
> There is no sense in keeping them today, in a time of feature full terminal 
> emulations. They can be abolished together with atom reactors and bombs. 

Completely agree on atom reactors and bombs ;-)

But you can pry my 'less' from my cold, dead fingers...

How are you supposed to page around a hundreds-of-megabytes heavy log
file across a ssh connection: by first pulling everything before the pos
you want to jump to?

Besides, less is faster than anything else around (with exception of
possibly 'more', but this one feels really archaic, even to me).

> And your splendid idea.

Hm. See above. And besides, Emacs buffers can't cope with sizes a pager
would take quite happily (unless you go to 64 bit platforms, where I
would expect the difference to be irrelevant).

Regards
- -- tomás
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)

iD8DBQFKWsNlBcgs9XrR2kYRAqBbAJsFPj5kmpM3EkVEludEHZdU3hqZnACfUonm
94rk5EYnFc7nTcQmnUgobMQ=
=kmRy
-----END PGP SIGNATURE-----

Re: page-at-a-time output for M-x shell

[Francis Moreau]

On Jul 12, 9:41 pm, Peter Dyballa <Peter_Dyba...@Web.DE> wrote:
> Am 12.07.2009 um 19:40 schrieb Francis Moreau:
>
> > And why is it so hard to understand that a pager feature would
> > be nice in this package ?

[...]

> If the text cursor is at the end of the *shell* buffer and is pushed  
> away by fast effervescent output, faster than you can read, it's one  
> command to return to the prompt. It's easier to send the command and  
> make the cursor go up a line.

no.

BTW, some tools which being designed this 3 last couple years do send
their output to a pager by default.

> GNU Emacs is no terminal emulation.

But it offers the possibility. It's well described in the info pages
and the info pages never say that it's a deprecated usage.

> It does not need a pager. These  
> programmes were developed for hardware terminals from 20 or more  
> years ago. There is no sense in keeping them today, in a time of  
> feature full terminal emulations. They can be abolished together with  
> atom reactors and bombs.

So make it clear in the info page and see how people react.

> And your splendid idea.

I never claim to have a splendid idea. I just found weird that this is
not offered by M-x shell, whereas it is by M-x term...

Re: page-at-a-time output for M-x shell

[Francis Moreau]

On Jul 12, 7:54 pm, Teemu Likonen <tliko...@iki.fi> wrote:
> On 2009-07-12 10:40 (-0700), Francis Moreau wrote:
>
> > Some other Emacs commands already do that, M-x shell get the output of
> > the command and put it in the buffer, so I would guess that it's not
> > that hard to tell Emacs to display one page of output at time but it's
> > just a guess...
>
> Please understand that "less" needs a _terminal_. Emacs's shell-mode is
> not a terminal so "less" doesn't work. If the programs you use needs
> terminal features then you must use a terminal, such as "M-x term".

I did understand that !

If you read my post carefully, I'm asking why 'page-at-a-time' feature
(available in M-x term) is not implemented in M-x shell where it would
seem to be more needed.

Re: page-at-a-time output for M-x shell

[Peter Dyballa]

Am 13.07.2009 um 07:17 schrieb tomas@tuxteam.de:

> How are you supposed to page around a hundreds-of-megabytes heavy log
> file across a ssh connection: by first pulling everything before  
> the pos
> you want to jump to?

Well, I'd use dired and tail. To give a real recommendation I first  
need some real experience. I've did some remote log file searching  
and similiar things, configuring and serving Apache, and I did not  
run into a situation I would have needed  a pager, using TRAMP with  
SSH or PuTTY. Log rotation is a nice thing. At least, I don't feel  
great when I have to work with GB log files.

--
Greetings

   Pete

"We need a president who's fluent in at least one language."
				– Buck Henry

Re: page-at-a-time output for M-x shell

[Teemu Likonen]

On 2009-07-13 01:07 (-0700), Francis Moreau wrote:

> On Jul 12, 7:54 pm, Teemu Likonen <tliko...@iki.fi> wrote:
>> Please understand that "less" needs a _terminal_. Emacs's shell-mode
>> is not a terminal so "less" doesn't work. If the programs you use
>> needs terminal features then you must use a terminal, such as "M-x
>> term".
>
> I did understand that !
>
> If you read my post carefully, I'm asking why 'page-at-a-time' feature
> (available in M-x term) is not implemented in M-x shell where it would
> seem to be more needed.

OK, then it's probably me who fails to understand why do you keep
talking about "M-x shell" when you need a terminal. This
"page-at-a-time" feature comes with terminal emulation. There's no need
to make "M-x shell" a real terminal because there already is one: "M-x
term". Is there something wrong with "M-x term"?

Re: page-at-a-time output for M-x shell

[Peter Dyballa]

Am 13.07.2009 um 10:03 schrieb Francis Moreau:

> I never claim to have a splendid idea. I just found weird that this is
> not offered by M-x shell, whereas it is by M-x term...


The one offers an environment to run a shell interpreter  
interactively in it, the other a terminal emulation. The one is  
restricted, the other not that much.

--
Greetings

   Pete

Math illiteracy affects 7 out of every 5 Americans.

Re: page-at-a-time output for M-x shell

[Francis Moreau]

On Jul 13, 10:21 am, Teemu Likonen <tliko...@iki.fi> wrote:
>
> OK, then it's probably me who fails to understand why do you keep
> talking about "M-x shell" when you need a terminal. This
> "page-at-a-time" feature comes with terminal emulation. There's no need
> to make "M-x shell" a real terminal because there already is one: "M-x
> term".

Ok then it's probably me who fails to understand the point of "M-x
shell"...

Just because it's not a real terminal emulation, does that mean it
should be boring to be used with commands with a lot of output (hence
lot of processing) ?

If "M-x shell" implements this 'page-at-a-time' feature, does that
mean it becomes a real term emulation ?

Is it really that much hard to make "M-x shell" mimics the one page at
a time ?

Maybe this shows one of the Emacs' drawback: there're so many packages
out there which implement the same functionnality but differently and
with different limitations and that make the user confused and spend a
lot of time to wonder/try each of all of these packages. And
eventually once the use makes a choice: "ok this package is good
enough but I miss this X functionnality". So now he's asking why X
functionnality is not implemented, and he got some answers like:
"there's an 10th package you should look at" or "just don't be silly
and use package 'A'"....

> Is there something wrong with "M-x term"?

Not really, but I found it not very well suited to be used inside
Emacs: the key bindings are not very friendly with the rest of Emacs
and sometimes the output is screwed, I dunno why. So it's not very
integrated to Emacs, and I just prefer start another term outside
emacs if I want a real term emulation.

Re: page-at-a-time output for M-x shell

[Francis Moreau]

On Jul 13, 11:04 am, Peter Dyballa <Peter_Dyba...@Web.DE> wrote:
> Am 13.07.2009 um 10:03 schrieb Francis Moreau:
>
> > I never claim to have a splendid idea. I just found weird that this is
> > not offered by M-x shell, whereas it is by M-x term...
>
> The one offers an environment to run a shell interpreter  
> interactively in it,

OK, so what am I supposed to run in this shell interpreter ?

I can understand that I can't start mutt or any other appli that needs
all term environment in M-x shell, but the feature I'm asking is not
specific to term emulation, it's just a goodies to make life easier
when reading the _text_ output of any commands I start from this shell
interpreter.

One other thing: why 'M-x term' proposes the 'page-at-a-time' output
feature whereas it supports the " | less" thing ? maybe because it's
just convenient enough to make it the default of all commands...

Re: page-at-a-time output for M-x shell

[Teemu Likonen]

On 2009-07-13 02:16 (-0700), Francis Moreau wrote:

> On Jul 13, 10:21 am, Teemu Likonen <tliko...@iki.fi> wrote:
>>
>> OK, then it's probably me who fails to understand why do you keep
>> talking about "M-x shell" when you need a terminal. This
>> "page-at-a-time" feature comes with terminal emulation. There's no
>> need to make "M-x shell" a real terminal because there already is
>> one: "M-x term".
>
> Ok then it's probably me who fails to understand the point of "M-x
> shell"...

The documentation says "Major mode for interacting with an inferior
shell" (C-h f shell-mode). It's like repeated M-! but more convenient.

> If "M-x shell" implements this 'page-at-a-time' feature, does that
> mean it becomes a real term emulation ?

To make "less" work it needs some terminal features so it would heading
towards that direction, that is, towards term-mode which already exists.
Perhaps someone could write some custom pager system in shell-mode but
it still wouldn't make "less" work.

> Is it really that much hard to make "M-x shell" mimics the one page at
> a time ?

You can write to emacs-devel mailing list and ask. I don't know.

>> Is there something wrong with "M-x term"?
>
> Not really, but I found it not very well suited to be used inside
> Emacs: the key bindings are not very friendly with the rest of Emacs
> and sometimes the output is screwed, I dunno why. So it's not very
> integrated to Emacs, and I just prefer start another term outside
> emacs if I want a real term emulation.

If you can present the problems in a clear way you could report them as
bugs, with "M-x report-emacs-bug" command. I don't use term-mode (nor
shell-mode for that matter) much but I have successfully tested Mutt
email client, Vim and Emacs itself inside term-mode, with no problems.
What comes to key bindings I have changed the escape char to C-x:

    (add-hook 'term-mode-hook 'my-term-mode-hook)
    (defun my-term-mode-hook ()
      (term-set-escape-char ?\C-x))

Re: page-at-a-time output for M-x shell

[Richard Riley]

Peter Dyballa <Peter_Dyballa@Web.DE> writes:

> Am 13.07.2009 um 10:03 schrieb Francis Moreau:
>
>> I never claim to have a splendid idea. I just found weird that this is
>> not offered by M-x shell, whereas it is by M-x term...
>
>
> The one offers an environment to run a shell interpreter  
> interactively in it, the other a terminal emulation. The one is  
> restricted, the other not that much.

The point is that you pretty much derided the idea that "Gnu Emacs" needing a pager whereas you
probably meant shell as opposed to a term inside emacs.

Re: page-at-a-time output for M-x shell

[Peter Dyballa]

Am 13.07.2009 um 11:38 schrieb Francis Moreau:

> OK, so what am I supposed to run in this shell interpreter ?

Everything but pagers and other programmes or utilities that control  
the cursor. Some ANSI compliance is achieved, to support gls'  
colourful output or bash colourising the prompt.

>
> I can understand that I can't start mutt or any other appli that needs
> all term environment in M-x shell, but the feature I'm asking is not
> specific to term emulation,

You're wrong: pagers take full cursor control over the terminal.  
They're no simple filters (source code is available), because then  
they would have been implemented as simple shell scripts. Like, for  
example, which.

> it's just a goodies to make life easier when reading the _text_  
> output of any commands I start from this shell
> interpreter.


Write some simple Lua, Ruby, Python, Perl, Shell, or whatever script  
which counts lines and echoes a (fixed or not) number of them to  
stdout. Or, much simpler, change your mind and bethink yourself of  
the task or the aim you want to reach. Is it using a pager? Then take  
*term* buffer or xterm, or aterm, or…


Probably the *very* best is to take one week holidays. At least.

--
Greetings

   Pete

Sending unsolicited commercial eMail to this account incurs a fee of  
€ 500 per message and acknowledges the legality of this contract.

Re: page-at-a-time output for M-x shell

[Francis Moreau]

On Jul 13, 11:51 am, Teemu Likonen <tliko...@iki.fi> wrote:
> On 2009-07-13 02:16 (-0700), Francis Moreau wrote:
>
> > On Jul 13, 10:21 am, Teemu Likonen <tliko...@iki.fi> wrote:
>
> >> OK, then it's probably me who fails to understand why do you keep
> >> talking about "M-x shell" when you need a terminal. This
> >> "page-at-a-time" feature comes with terminal emulation. There's no
> >> need to make "M-x shell" a real terminal because there already is
> >> one: "M-x term".
>
> > Ok then it's probably me who fails to understand the point of "M-x
> > shell"...
>
> The documentation says "Major mode for interacting with an inferior
> shell" (C-h f shell-mode). It's like repeated M-! but more convenient.
>
> > If "M-x shell" implements this 'page-at-a-time' feature, does that
> > mean it becomes a real term emulation ?
>
> To make "less" work it needs some terminal features so it would heading
> towards that direction, that is, towards term-mode which already exists.
> Perhaps someone could write some custom pager system in shell-mode but
> it still wouldn't make "less" work.
>

Damn, I repeat one more time: I don't want "| less" or less(1) work in
"M-x shell" (hope this is clear now)

IOW, I'd like 'M-x shell' to output in its buffer one page at time,
and this could be enabled/disabled by a variable like "M-x term" does.

Re: page-at-a-time output for M-x shell

[Francis Moreau]

On Jul 13, 12:43 pm, Peter Dyballa <Peter_Dyba...@Web.DE> wrote:
> Am 13.07.2009 um 11:38 schrieb Francis Moreau:
>
> > OK, so what am I supposed to run in this shell interpreter ?
>
> Everything but pagers and other programmes or utilities that control  
> the cursor. Some ANSI compliance is achieved, to support gls'  
> colourful output or bash colourising the prompt.
>

[...]

>
> You're wrong: pagers take full cursor control over the terminal.  
> They're no simple filters (source code is available), because then  
> they would have been implemented as simple shell scripts. Like, for  
> example, which.
>

Ok, I give up.

>
> Probably the *very* best is to take one week holidays. At least.
>

yeah, I probably need it specially after such discussion ;)

thanks anyway

Re: page-at-a-time output for M-x shell

[Sébastien Vauban]

Francis Moreau wrote:
> On Jul 13, 11:51 am, Teemu Likonen <tliko...-X3B1VOXEql0@public.gmane.org> wrote:
>> On 2009-07-13 02:16 (-0700), Francis Moreau wrote:
>>> On Jul 13, 10:21 am, Teemu Likonen <tliko...-X3B1VOXEql0@public.gmane.org> wrote:
>>>> OK, then it's probably me who fails to understand why do you keep talking
>>>> about "M-x shell" when you need a terminal. This "page-at-a-time" feature
>>>> comes with terminal emulation. There's no need to make "M-x shell" a real
>>>> terminal because there already is one: "M-x term".
>
> Damn, I repeat one more time: I don't want "| less" or less(1) work in "M-x
> shell" (hope this is clear now)

Just one (real) question: what's the advantage of `M-x shell'? I understand
`M-x term' allows for full-blown terminal emulation, but can we use it every
time we were going to use `M-x shell'?

Or are there some things `M-x term' can't do while `M-x shell' can?

Thanks,
  Seb

-- 
Sébastien Vauban

Re: page-at-a-time output for M-x shell

[Anselm Helbig]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 1580 bytes --]

At Mon, 13 Jul 2009 15:12:25 +0200,
Sébastien Vauban <wxhgmqzgwmuf@spammotel.com> wrote:
> 
> Francis Moreau wrote:
> > On Jul 13, 11:51 am, Teemu Likonen <tliko...@iki.fi> wrote:
> >> On 2009-07-13 02:16 (-0700), Francis Moreau wrote:
> >>> On Jul 13, 10:21 am, Teemu Likonen <tliko...@iki.fi> wrote:
> >>>> OK, then it's probably me who fails to understand why do you keep talking
> >>>> about "M-x shell" when you need a terminal. This "page-at-a-time" feature
> >>>> comes with terminal emulation. There's no need to make "M-x shell" a real
> >>>> terminal because there already is one: "M-x term".
> >
> > Damn, I repeat one more time: I don't want "| less" or less(1) work in "M-x
> > shell" (hope this is clear now)
> 
> Just one (real) question: what's the advantage of `M-x shell'? I understand
> `M-x term' allows for full-blown terminal emulation, but can we use it every
> time we were going to use `M-x shell'?
> 
> Or are there some things `M-x term' can't do while `M-x shell' can?

M-x term is a terminal emulation, so things work more or less like
they do in an xterm, gnome-terminal or putty. With M-x shell you're
using emacs to construct the command to pass to the shell. This is an
advantage if you shell doesn't keep a history or offers no completion
mechanism. Also, you can use isearch to search through the command
output, copy stuff into the kill ring or use the rectangle
functions. Or you might just prefer the emacs keybindings over the
one's your shell offers. 8-)

HTH, 

Anselm


-- 
Anselm Helbig 
mailto:anselm.helbig+news2009@googlemail.com

Re: page-at-a-time output for M-x shell

[Teemu Likonen]

On 2009-07-13 16:48 (+0200), Anselm Helbig wrote:

> With M-x shell you're using emacs to construct the command to pass to
> the shell. This is an advantage if you shell doesn't keep a history or
> offers no completion mechanism. Also, you can use isearch to search
> through the command output, copy stuff into the kill ring or use the
> rectangle functions. Or you might just prefer the emacs keybindings
> over the one's your shell offers. 8-)

I'd like to point out that term-mode has also the so called "line mode"
(C-c C-j) in which user can wander around the buffer pretty much like
anywhere else in Emacs. Kill-ring commands, isearch etc. work. Then
there is "char mode" (C-c C-k), the default, which is like your normal
terminal emulator except the escape key.

Re: page-at-a-time output for M-x shell

[Samuel Wales]

I read the whole thread, and I am surprised at how thoroughly and
repeatedly the OP was misunderstood.

So much so that:

On Mon, Jul 13, 2009 at 04:29, Francis Moreau<francis.moro@gmail.com> wrote:
> Ok, I give up.

Many respondents took time out of their busy lives to explain the
difference between term and shell -- but I think he knew those
differences.

Many graciously told him that he can't run less in shell -- but I
think he knew that.

I think what he wants is shell mode (not term) but to have some sort
of less-like solution (not less itself) for paging output.  Page up
and prompt search and term in line mode are all good ideas, but he
probably tried them.

It doesn't seem like he's so wrong for merely asking how to do that or
whether it is possible.

Perhaps a simple "I don't know of a solution for this" would suffice.

-- 
Myalgic encephalomyelitis causes death and severe suffering.
You can get it any time and never recover.  Conflicts of
interest are destroying research.  Do science and justice
matter to you?  http://www.meactionuk.org.uk/What_Is_ME_What_Is_CFS.htm

Re: page-at-a-time output for M-x shell

[Pascal J. Bourguignon]

Francis Moreau <francis.moro@gmail.com> writes:

> Hello,
>
> M-x term has a nice feature 'page-at-a-time' which I'd like to have in
> M-x shell specially because commands like
>
>    $ cmd-with-a-lof-of-output | less
>    WARNING: terminal is not fully functional
>     -  (press RETURN)
>
> doesn't work properly.
>
> Does anybody know some trick here ?

So, it seems that you don't like PgUp and PgDn, and you don't like M-x
term RET.   Then if you insist om M-x shell, you will have to write
your own pager that doesn't require a smart terminal (which shell is
not):


[pjb@galatea :0.0 ~]$ cat ~/bin/epager
#!/bin/bash
HEIGHT=${HEIGHT:=25}
eof=0
while [ $eof -eq 0 ] ; do
    i=0
    while [ $i -lt $HEIGHT -a $eof -eq 0 ] ; do
        line=''
        read line 
        if [ $? -eq 0 ] ; then
            printf "%s\n" "$line" 
        else
            eof=1
        fi
        i=$(( $i + 1 ))
    done
    read -p 'Press RET to continue, q RET to abort:' rep < /dev/tty
    if [ "$rep" = 'q' ] ; then
        eof=1
    fi
done
[pjb@galatea :0.0 ~]$ cat /home/pjb/tmp/misc/wang | HEIGHT=10 epager
Hao Wang, logicien americain.

L'algorithme en  question  a  ete  publie  en  1960  dans l'IBM Journal,
article intitule "Toward  Mechanical Mathematics", avec des variantes et
une  extension au calcul  des  predicats.  Il  s'agit  ici  du  "premier
programme" de Wang, systeme "P".

L'article a ete ecrit en 1958, et les experiences effectuees sur IBM 704
- machine a lampes, 32 k  mots  de 36 bits, celle-la meme qui vit naitre
LISP a la meme epoque. Le programme  a  ete ecrit en assembleur (Fortran
Press RET to continue, q RET to abort:

existait, mais il ne s'etait pas encore impose)  et  l'auteur estime que
"there is very little in the program that is not straightforward".

Il observe que les preuves engendrees sont "essentiellement des arbres",
et  annonce  que  la  machine  a  demontre 220 theoremes du  calcul  des
propositions  (tautologies)  en  3  minutes. Il en tire argument pour la
superiorite  d'une  approche  algorithmique  par  rapport a une approche
heuristique comme celle du "Logic Theorist" de Newell, Shaw et  Simon (a
partir de 1956 sur la machine JOHNNIAC de la Rand Corporation): un debat
qui dure encore...
Press RET to continue, q RET to abort:


Cet  algorithme  a  ete popularise par J. McCarthy, comme exemple-fanion
d'application  de LISP. Il figure dans le manuel de la premiere  version
de  LISP  (LISP  1,  sur IBM 704 justement, le manuel est date  de  Mars
1960), et il a ete repris dans le celebre "LISP 1.5 Programmer's Manual"
publie en 1962 par MIT Press, un des maitres-livres de l'Informatique.



Press RET to continue, q RET to abort:

[pjb@galatea :0.0 ~]$ cat /home/pjb/tmp/misc/wang | HEIGHT=10 epager
Hao Wang, logicien americain.

L'algorithme en  question  a  ete  publie  en  1960  dans l'IBM Journal,
article intitule "Toward  Mechanical Mathematics", avec des variantes et
une  extension au calcul  des  predicats.  Il  s'agit  ici  du  "premier
programme" de Wang, systeme "P".

L'article a ete ecrit en 1958, et les experiences effectuees sur IBM 704
- machine a lampes, 32 k  mots  de 36 bits, celle-la meme qui vit naitre
LISP a la meme epoque. Le programme  a  ete ecrit en assembleur (Fortran
Press RET to continue, q RET to abort:q
[pjb@galatea :0.0 ~]$ 


-- 
__Pascal Bourguignon__

Re: page-at-a-time output for M-x shell

[Teemu Likonen]

On 2009-07-13 13:42 (-0700), Samuel Wales wrote:

> I read the whole thread, and I am surprised at how thoroughly and
> repeatedly the OP was misunderstood.

Then I'm certainly one of those. Sorry about that, Francis.

> I think what he wants is shell mode (not term) but to have some sort
> of less-like solution (not less itself) for paging output. Page up and
> prompt search and term in line mode are all good ideas, but he
> probably tried them.

Hmm, speaking speculatively, would it be some code in shell.el (or much
likely comint.el) which stores shell commands' output in an intermediate
hidden buffer or some Lisp data type and then pages it to *shell* buffer
for user? I'm pretty sure that that would be possible to implement.

From quick glance it looks like comint-output-filter is the function
which handles the output that's coming from shell commands. The function
is called automatically when Emacs sees data coming from an external
process. I can see no pager features there but that's the place to start
for someone who's interested. Split the functionality: have the output
filter function write directly to hidden buffer and implement another
output function which moves content from hidden buffer to *shell*
buffer, possibly page by page.

> It doesn't seem like he's so wrong for merely asking how to do that or
> whether it is possible.

I agree with that. Assuming I'm finally understanding what is being
asked. :-)

Re: page-at-a-time output for M-x shell

[Barry Margolin]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 1319 bytes --]

In article 
<0eab476b-cd5f-444d-b2af-d82ac076e73b@r34g2000vba.googlegroups.com>,
 Francis Moreau <francis.moro@gmail.com> wrote:

> On Jul 13, 11:04 am, Peter Dyballa <Peter_Dyba...@Web.DE> wrote:
> > Am 13.07.2009 um 10:03 schrieb Francis Moreau:
> >
> > > I never claim to have a splendid idea. I just found weird that this is
> > > not offered by M-x shell, whereas it is by M-x term...
> >
> > The one offers an environment to run a shell interpreter  
> > interactively in it,
> 
> OK, so what am I supposed to run in this shell interpreter ?
> 
> I can understand that I can't start mutt or any other appli that needs
> all term environment in M-x shell, but the feature I'm asking is not
> specific to term emulation, it's just a goodies to make life easier
> when reading the _text_ output of any commands I start from this shell
> interpreter.

Why do you need a pager, when the Emacs shell buffer allows you to page 
back and forward through the output, search it, etc.?  What can you do 
in a pager that you can't do with Emacs's own features?  And if there is 
something Emacs is missing, it should be easy to add it.

-- 
Barry Margolin, barmar@alum.mit.edu
Arlington, MA
*** PLEASE post questions in newsgroups, not directly to me ***
*** PLEASE don't copy me on replies, I'll read them in the group ***

Re: page-at-a-time output for M-x shell

[Miles Bader]

Barry Margolin <barmar@alum.mit.edu> writes:
> Why do you need a pager, when the Emacs shell buffer allows you to page 
> back and forward through the output, search it, etc.?  What can you do 
> in a pager that you can't do with Emacs's own features?  And if there is 
> something Emacs is missing, it should be easy to add it.

BTW, note that if a command outputs lots of text, it's easy to go back
to the previous prompt using "C-c C-p" and then page through the output
with "C-v".

-Miles

-- 
Success, n. The one unpardonable sin against one's fellows.

Re: page-at-a-time output for M-x shell

[Pascal J. Bourguignon]

Teemu Likonen <tlikonen@iki.fi> writes:

> On 2009-07-13 13:42 (-0700), Samuel Wales wrote:
>
>> I read the whole thread, and I am surprised at how thoroughly and
>> repeatedly the OP was misunderstood.
>
> Then I'm certainly one of those. Sorry about that, Francis.
>
>> I think what he wants is shell mode (not term) but to have some sort
>> of less-like solution (not less itself) for paging output. Page up and
>> prompt search and term in line mode are all good ideas, but he
>> probably tried them.
>
> Hmm, speaking speculatively, would it be some code in shell.el (or much
> likely comint.el) which stores shell commands' output in an intermediate
> hidden buffer or some Lisp data type and then pages it to *shell* buffer
> for user? I'm pretty sure that that would be possible to implement.

Of course.  It is ALREADY implemented.  That data structure is called
BUFFER, and the pages are called WINDOWS.  There is a user interface
to let the user scan the pages, they're the PgUp and PgDn keys (PAGE
UP, PAGE DOWN).

How much silliness must we endure each day???


But note however, that M-x term implements an automatic pager in
emacs, so you don't even have to remember to type |less : as soon as
the output is more than a window-long, the term emacs pager jumps in.  

-- 
__Pascal Bourguignon__

Re: page-at-a-time output for M-x shell

[Francis Moreau]

On 13 juil, 15:12, Sébastien Vauban <wxhgmqzgw...@spammotel.com>
wrote:
> Francis Moreau wrote:
> > On Jul 13, 11:51 am, Teemu Likonen <tliko...@iki.fi> wrote:
> >> On 2009-07-13 02:16 (-0700), Francis Moreau wrote:
> >>> On Jul 13, 10:21 am, Teemu Likonen <tliko...@iki.fi> wrote:
> >>>> OK, then it's probably me who fails to understand why do you keep talking
> >>>> about "M-x shell" when you need a terminal. This "page-at-a-time" feature
> >>>> comes with terminal emulation. There's no need to make "M-x shell" a real
> >>>> terminal because there already is one: "M-x term".
>
> > Damn, I repeat one more time: I don't want "| less" or less(1) work in "M-x
> > shell" (hope this is clear now)
>
> Just one (real) question: what's the advantage of `M-x shell'? I understand
> `M-x term' allows for full-blown terminal emulation, but can we use it every
> time we were going to use `M-x shell'?

This one has already been asked before just look at:

<76036640-c499-482c-81c0-9e0cb325e96a@k26g2000vbp.googlegroups.com>

>
> Or are there some things `M-x term' can't do while `M-x shell' can?
>

Actually you can read this in the emacs info pages: "In line mode,
Term basically acts like Shell mode".

The problem here is the word 'basically' which probably means
'almost'. So they differ in the details.

In my case, I started to use 'M-x shell', and like it since I almost
never have to use applications that need a real term emulation. But if
you look at them (M-x term and M-x shell) carefully, then 'M-x term'
has some weird behaviours when using the shell mode key bindings; try
for instance: 'C-c C-e', 'C-c C-a', 'C-c C-o'... but as I said they're
details.

But the main problem I think, which is a different issue, is why the
hell Emacs has 2 functionalities that do 'basically' the same thing ?!

Why can't it offer one and only one package that give the best of the
2 worlds ?

Re: page-at-a-time output for M-x shell

[Francis Moreau]

On 14 juil, 12:55, p...@informatimago.com (Pascal J. Bourguignon)
wrote:
> Teemu Likonen <tliko...@iki.fi> writes:
> > On 2009-07-13 13:42 (-0700), Samuel Wales wrote:
>
> >> I read the whole thread, and I am surprised at how thoroughly and
> >> repeatedly the OP was misunderstood.
>
> > Then I'm certainly one of those. Sorry about that, Francis.
>
> >> I think what he wants is shell mode (not term) but to have some sort
> >> of less-like solution (not less itself) for paging output. Page up and
> >> prompt search and term in line mode are all good ideas, but he
> >> probably tried them.
>
> > Hmm, speaking speculatively, would it be some code in shell.el (or much
> > likely comint.el) which stores shell commands' output in an intermediate
> > hidden buffer or some Lisp data type and then pages it to *shell* buffer
> > for user? I'm pretty sure that that would be possible to implement.
>
> Of course.  It is ALREADY implemented.  That data structure is called
> BUFFER, and the pages are called WINDOWS.  There is a user interface
> to let the user scan the pages, they're the PgUp and PgDn keys (PAGE
> UP, PAGE DOWN).

No it's not.

What you propose with PgUp/PgDn is a _workaround_.

>
> How much silliness must we endure each day???
>
> But note however, that M-x term implements an automatic pager in
> emacs, so you don't even have to remember to type |less : as soon as
> the output is more than a window-long, the term emacs pager jumps in.  
>

So why is it silly to implement the 'automatic pager emulation' in
shell mode and not in 'M-x term' ?

Re: page-at-a-time output for M-x shell

[Francis Moreau]

On 14 juil, 05:12, Barry Margolin <bar...@alum.mit.edu> wrote:
> In article
> <0eab476b-cd5f-444d-b2af-d82ac076e...@r34g2000vba.googlegroups.com>,
>  Francis Moreau <francis.m...@gmail.com> wrote:
>
>
>
> > On Jul 13, 11:04 am, Peter Dyballa <Peter_Dyba...@Web.DE> wrote:
> > > Am 13.07.2009 um 10:03 schrieb Francis Moreau:
>
> > > > I never claim to have a splendid idea. I just found weird that this is
> > > > not offered by M-x shell, whereas it is by M-x term...
>
> > > The one offers an environment to run a shell interpreter  
> > > interactively in it,
>
> > OK, so what am I supposed to run in this shell interpreter ?
>
> > I can understand that I can't start mutt or any other appli that needs
> > all term environment in M-x shell, but the feature I'm asking is not
> > specific to term emulation, it's just a goodies to make life easier
> > when reading the _text_ output of any commands I start from this shell
> > interpreter.
>
> Why do you need a pager, when the Emacs shell buffer allows you to page
> back and forward through the output, search it, etc.?

Ok, let me ask you one question: why do anybody use ('| less') in all
term emulation whereas they all have the PgUp/PgDw functionality ?

because it's obviously not convenient.

It's not convenient because a) you get the wanted result without any
user interventions (search, key press...) b) the command which outputs
the data is stopped as soon as the reader stops to read the produced
data.

Note: just to be sure I'm understood, I'm not asking for a pager in
shell mode.

Re: page-at-a-time output for M-x shell

[Pascal J. Bourguignon]

Francis Moreau <francis.moro@gmail.com> writes:
> But the main problem I think, which is a different issue, is why the
> hell Emacs has 2 functionalities that do 'basically' the same thing ?!

Two? I know at least three!  M-x terminal is yet another.

As for mail UA, there are at least four (gnus, rmail, vm, mew), and
probably many more.


> Why can't it offer one and only one package that give the best of the
> 2 worlds ?

Because this is not a scarce world.  There's nobody to strangle the
offer and make you pay dearly for a few features.  There's an
aboundance of coders, and an aboundance of code produced, so you can
choose amongst several options.  If you want empty shops, URSS^W North
Korea is that way.

-- 
__Pascal Bourguignon__

Re: page-at-a-time output for M-x shell

[Pascal J. Bourguignon]

Francis Moreau <francis.moro@gmail.com> writes:

> On 14 juil, 12:55, p...@informatimago.com (Pascal J. Bourguignon)
> wrote:
>> Teemu Likonen <tliko...@iki.fi> writes:
>> > On 2009-07-13 13:42 (-0700), Samuel Wales wrote:
>>
>> >> I read the whole thread, and I am surprised at how thoroughly and
>> >> repeatedly the OP was misunderstood.
>>
>> > Then I'm certainly one of those. Sorry about that, Francis.
>>
>> >> I think what he wants is shell mode (not term) but to have some sort
>> >> of less-like solution (not less itself) for paging output. Page up and
>> >> prompt search and term in line mode are all good ideas, but he
>> >> probably tried them.
>>
>> > Hmm, speaking speculatively, would it be some code in shell.el (or much
>> > likely comint.el) which stores shell commands' output in an intermediate
>> > hidden buffer or some Lisp data type and then pages it to *shell* buffer
>> > for user? I'm pretty sure that that would be possible to implement.
>>
>> Of course.  It is ALREADY implemented.  That data structure is called
>> BUFFER, and the pages are called WINDOWS.  There is a user interface
>> to let the user scan the pages, they're the PgUp and PgDn keys (PAGE
>> UP, PAGE DOWN).
>
> No it's not.
>
> What you propose with PgUp/PgDn is a _workaround_.
>
>>
>> How much silliness must we endure each day???
>>
>> But note however, that M-x term implements an automatic pager in
>> emacs, so you don't even have to remember to type |less : as soon as
>> the output is more than a window-long, the term emacs pager jumps in.  
>>
>
> So why is it silly to implement the 'automatic pager emulation' in
> shell mode and not in 'M-x term' ?

It is clear from the first paragraph of my previous answer, but since
you don't understand (to the point of making this thread sound
trollish), I'll repeat: it is silly to implement the 'automatic pager
emulation' in shell mode, because the  feature is ALREADY IMPLEMENTED
IN THE GUISE of a BUFFER, a WINDOW and PgUp and PgDn to browse the
buffer.  What more do you want?  more or less implement exactly the
same features, they load the file in a buffer, and display a
window-long at a time, and let you advance and (for less) retroced on
the press of a key.  If you don't like the keys PgUp or PgDn, there's
already substitute keys implemented in emacs shell, namely C-v and
M-v, and if you don't like them either, it's trivial to bind other
keys.


-- 
__Pascal Bourguignon__

Re: page-at-a-time output for M-x shell

[Pascal J. Bourguignon]

Francis Moreau <francis.moro@gmail.com> writes:

> On 14 juil, 05:12, Barry Margolin <bar...@alum.mit.edu> wrote:
>> In article
>> <0eab476b-cd5f-444d-b2af-d82ac076e...@r34g2000vba.googlegroups.com>,
>>  Francis Moreau <francis.m...@gmail.com> wrote:
>>
>>
>>
>> > On Jul 13, 11:04 am, Peter Dyballa <Peter_Dyba...@Web.DE> wrote:
>> > > Am 13.07.2009 um 10:03 schrieb Francis Moreau:
>>
>> > > > I never claim to have a splendid idea. I just found weird that this is
>> > > > not offered by M-x shell, whereas it is by M-x term...
>>
>> > > The one offers an environment to run a shell interpreter  
>> > > interactively in it,
>>
>> > OK, so what am I supposed to run in this shell interpreter ?
>>
>> > I can understand that I can't start mutt or any other appli that needs
>> > all term environment in M-x shell, but the feature I'm asking is not
>> > specific to term emulation, it's just a goodies to make life easier
>> > when reading the _text_ output of any commands I start from this shell
>> > interpreter.
>>
>> Why do you need a pager, when the Emacs shell buffer allows you to page
>> back and forward through the output, search it, etc.?
>
> Ok, let me ask you one question: why do anybody use ('| less') in all
> term emulation whereas they all have the PgUp/PgDw functionality ?
>
> because it's obviously not convenient.

You are wrong, and mixing concepts.

1- Not all terminal emulator have an history that you can navigate
   (with suposedly PgUp and PgDn).  The job of a terminal emulator is
   to send bytes to the application for any key pressed including PgUp
   and PgDn, so normally they wouldn't preempt them to scroll the
   history, if they have one.

2- In the case of M-x shell, nobody use |less, since we can scroll
   back painlessly.


Actually, the invention of more or less comes directly from this
characteristic feature of a glass terminal, that it shows only a
limited number of lines, and you cannot scroll back.

When we still worked with teletypes, there was no need for more or
less, since you could always check what was printed in the roll of
paper (and it would have been totally ludicrous to print again a page
that had been already printed).

Then glass tty were introduced, that behaved like teletypes, but they
showed only 24 or 25 lines.  This was a big backset, but it presented
some advantages: we didn't need to use so many trees anymore, and we
could now directly address and update any characters in the window.
Anyway with these glass tty, since anything beyond the 24th/25th line
was lost, a workaround was needed, and more or less (the pager
concept) was invented.  Notice that 25*80 = 2000 bytes which is about
the size of central memory that was allocated to each user on the
computers used at that time, so it would have been impossible to keep
the whole output in a RAM buffer too (data was dumped in temporary
spool files).


But then bitmap screens and X servers and emacs were invented, on
machines with more memory than you could fill with your trolls, so the
previous workaround could be lost, and we can now implement the normal
feature, dump the whole output in the in RAM buffer, and let the user
see whatever part he wants with normal buffer navigation commands.



> It's not convenient because a) you get the wanted result without any
> user interventions (search, key press...) 

So you're saying that user intervention is "convenient"?  Computer are
made to aleviate human work, not to give more work to human slaves.


> b) the command which outputs the data is stopped as soon as the
> reader stops to read the produced data.

You'll have to explain in what aspect it is convenient to stop a
process, and leave resources busy for an indeterminate amount of time,
instead of releasing them for the other processes to use.


> Note: just to be sure I'm understood, I'm not asking for a pager in
> shell mode.

How strange, it seemed to everybody that was exactly what you were
asking in your second post:

    I'd want to use a pager with 'M-x shell' but ' | $PAGER' doesn't work
    nicely in emacs shell. So if I loose the '| $PAGER' part, I loose the
    functionnality I want to have as well...

and we've spend an inordinate amount of time explaining that you
didn't loose the functionality, since shell implements it inherently,
in addition to the epager script I provided if you didn't want to
loose the '| $PAGER' part.


You should consider yourself satisfied and quit it.

-- 
__Pascal Bourguignon__

Re: page-at-a-time output for M-x shell

[tomas]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Tue, Jul 14, 2009 at 07:21:02PM +0200, Pascal J. Bourguignon wrote:
> Francis Moreau <francis.moro@gmail.com> writes:
> 
> > On 14 juil, 12:55, p...@informatimago.com (Pascal J. Bourguignon)
> > wrote:

[...]

> >> Of course.  It is ALREADY implemented [...]

> > No it's not.
> >
> > What you propose with PgUp/PgDn is a _workaround_.

[...]

> It is clear from the first paragraph of my previous answer, but since
> you don't understand (to the point of making this thread sound
> trollish), I'll repeat: [...]

To be fair, a buffer with PgUp and PgDn doesn't completely emulate a
pager at least in two respects:

  * performance (as was pointed out upthread)
  * stopping the output stream of the process at each page.

They are just much better to cope with huge/fast streams. So the OP's
query isn't silly at all.

Regards
- -- tomás
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)

iD8DBQFKXNq6Bcgs9XrR2kYRAuJiAJ0ek30XyPhh2ZHWxgA6SwmFfnm1GACfW6N8
AI57k0jnjgJIYVGstwfeT0M=
=KS8V
-----END PGP SIGNATURE-----

Re: page-at-a-time output for M-x shell

[Francis Moreau]

On 14 juil, 21:21, to...@tuxteam.de wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> On Tue, Jul 14, 2009 at 07:21:02PM +0200, Pascal J. Bourguignon wrote:
> > Francis Moreau <francis.m...@gmail.com> writes:
>
> > > On 14 juil, 12:55, p...@informatimago.com (Pascal J. Bourguignon)
> > > wrote:
>
> [...]
>
> > >> Of course.  It is ALREADY implemented [...]
> > > No it's not.
>
> > > What you propose with PgUp/PgDn is a _workaround_.
>
> [...]
>
> > It is clear from the first paragraph of my previous answer, but since
> > you don't understand (to the point of making this thread sound
> > trollish), I'll repeat: [...]
>
> To be fair, a buffer with PgUp and PgDn doesn't completely emulate a
> pager at least in two respects:
>
>   * performance (as was pointed out upthread)
>   * stopping the output stream of the process at each page.

I actually mentioned the second point which was more important than
performance when using Emacs, I guess.

> They are just much better to cope with huge/fast streams. So the OP's
> query isn't silly at all.

Yeah but some people here are cleverer than I am, specially Monsieur
Bourguignon with his nice PgUp/PgDw solution and his nice followups.

So yes Monsieur Bourguignon, you're right, it's enough for this empty
discussion.

Re: page-at-a-time output for M-x shell

[Francis Moreau]

On 14 juil, 00:05, Teemu Likonen <tlikonen@iki.fi> wrote:
> On 2009-07-13 13:42 (-0700), Samuel Wales wrote:
>
> > I read the whole thread, and I am surprised at how thoroughly and
> > repeatedly the OP was misunderstood.
>
> Then I'm certainly one of those. Sorry about that, Francis.
>

Well, no problem and thanks anyway to spend some time of your busy
life to answer my question.

> > I think what he wants is shell mode (not term) but to have some sort
> > of less-like solution (not less itself) for paging output. Page up and
> > prompt search and term in line mode are all good ideas, but he
> > probably tried them.
>
> Hmm, speaking speculatively, would it be some code in shell.el (or much
> likely comint.el) which stores shell commands' output in an intermediate
> hidden buffer or some Lisp data type and then pages it to *shell* buffer
> for user? I'm pretty sure that that would be possible to implement.
>
> From quick glance it looks like comint-output-filter is the function
> which handles the output that's coming from shell commands. The function
> is called automatically when Emacs sees data coming from an external
> process. I can see no pager features there but that's the place to start
> for someone who's interested. Split the functionality: have the output
> filter function write directly to hidden buffer and implement another
> output function which moves content from hidden buffer to *shell*
> buffer, possibly page by page.

Ok, I'll may take a look at this just because that might be a good
exercise for learning elisp.

>
> > It doesn't seem like he's so wrong for merely asking how to do that or
> > whether it is possible.
>
> I agree with that. Assuming I'm finally understanding what is being
> asked. :-)

Take a look at the subject ;-)

Re: page-at-a-time output for M-x shell

[Francis Moreau]

On 14 juil, 19:16, pjb@informatimago.com (Pascal J. Bourguignon)
wrote:
> Francis Moreau <francis.moro@gmail.com> writes:
> > But the main problem I think, which is a different issue, is why the
> > hell Emacs has 2 functionalities that do 'basically' the same thing ?!
>
> Two? I know at least three!  M-x terminal is yet another.
>
> As for mail UA, there are at least four (gnus, rmail, vm, mew), and
> probably many more.
>
> > Why can't it offer one and only one package that give the best of the
> > 2 worlds ?
>
> Because this is not a scarce world.  There's nobody to strangle the
> offer and make you pay dearly for a few features.  There's an
> aboundance of coders, and an aboundance of code produced, so you can
> choose amongst several options.  If you want empty shops, URSS^W North
> Korea is that way.
>

Damn, Monsieur, you're crazy :-)

Re: page-at-a-time output for M-x shell

[Pascal J. Bourguignon]

Francis Moreau <francis.moro@gmail.com> writes:
> Yeah but some people here are cleverer than I am, specially Monsieur
> Bourguignon with his nice PgUp/PgDw solution and his nice followups.
>
> So yes Monsieur Bourguignon, you're right, it's enough for this empty
> discussion.

Don't be so up tight, I provided you with three solutions, including a
pager that works in M-x shell!

-- 
__Pascal Bourguignon__

Re: page-at-a-time output for M-x shell

[Barry Margolin]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 2584 bytes --]

In article 
<19dfd9e3-c7ad-4680-bfb7-2d25f79fca91@o36g2000vbl.googlegroups.com>,
 Francis Moreau <francis.moro@gmail.com> wrote:

> On 14 juil, 05:12, Barry Margolin <bar...@alum.mit.edu> wrote:
> > In article
> > <0eab476b-cd5f-444d-b2af-d82ac076e...@r34g2000vba.googlegroups.com>,
> >  Francis Moreau <francis.m...@gmail.com> wrote:
> >
> >
> >
> > > On Jul 13, 11:04 am, Peter Dyballa <Peter_Dyba...@Web.DE> wrote:
> > > > Am 13.07.2009 um 10:03 schrieb Francis Moreau:
> >
> > > > > I never claim to have a splendid idea. I just found weird that this is
> > > > > not offered by M-x shell, whereas it is by M-x term...
> >
> > > > The one offers an environment to run a shell interpreter  
> > > > interactively in it,
> >
> > > OK, so what am I supposed to run in this shell interpreter ?
> >
> > > I can understand that I can't start mutt or any other appli that needs
> > > all term environment in M-x shell, but the feature I'm asking is not
> > > specific to term emulation, it's just a goodies to make life easier
> > > when reading the _text_ output of any commands I start from this shell
> > > interpreter.
> >
> > Why do you need a pager, when the Emacs shell buffer allows you to page
> > back and forward through the output, search it, etc.?
> 
> Ok, let me ask you one question: why do anybody use ('| less') in all
> term emulation whereas they all have the PgUp/PgDw functionality ?

Terminal emulators often have a scrollback limit.  Also, you can search 
for things with "less", you can't easily search in the emulator window.  
Or maybe you can -- I've been using more and less since the days of real 
terminals, so that's where my habits got formed.

> 
> because it's obviously not convenient.
> 
> It's not convenient because a) you get the wanted result without any
> user interventions (search, key press...) b) the command which outputs
> the data is stopped as soon as the reader stops to read the produced
> data.

I agree with these points, too.  Scrollback in some terminal emulators 
doesn't work well if the output is still going -- sometimes they spring 
you back to the end, or the position moves as the end moves, etc.  

Pagers give more control, but you get all the same control in Emacs.

Actually, I generally try to avoid running programs that produce lots of 
output in my shell buffer.  I'll usually run them with M-! so they go to 
their own buffer.

-- 
Barry Margolin, barmar@alum.mit.edu
Arlington, MA
*** PLEASE post questions in newsgroups, not directly to me ***
*** PLEASE don't copy me on replies, I'll read them in the group ***

Re: page-at-a-time output for M-x shell

[Peter Dyballa]

Am 14.07.2009 um 22:02 schrieb Francis Moreau:

> Damn, Monsieur, you're crazy :-)


I'm in the same situation! I can't understand how my family switches  
through thirty-something TV programmes!

Some years ago our old TV set broke. The new one had no buttons to  
switch programmes. And the cable delivered thirty-something! I did  
not care, the important programmes like arte, 3sat, BR, WDR I could  
easily reach. And for the others I had my folks. One summer I had to  
stay at home while the family was on holidays (I was offered to come  
later in my car). Now I had a problem! From my car, a big truck, I  
was used to twenty-something gear programmes, en route on route  
national 7 or German autobahns. No problem to switch! But that damned  
TV set, it had no gear box! Being a born mechanic (and Elisp  
programmer) I could easily dismantle my truck's gear and take it up  
home (second floor). Now, stupid TV set, I'll teach how to switch  
programmes! No dice…

OK, I did not solve the programme, uhm, problem last millennium, I  
could get away into holidays, so that I still don't know whether I  
have to bring the whole TV set into the truck or the complete truck  
into the living room…

It actually still puzzles me! (Dancing my onze danses pour combattre  
ma migraine.)

--
Greetings

   Pete

Imbecility, n.:
	A kind of divine inspiration, or sacred fire affecting
	censorious critics of this dictionary.
			– Ambrose Bierce: _The Devil's Dictionary_