What I'm missing when using M-x shell

What I'm missing when using M-x shell

[Francis Moreau]

Hello,

I'm still trying to do all my jobs inside emacs but I'm missing a couple
things.

The first one is what I can't do anymore when working with a shell/terminal
inside emacs.

  1/ Most applications based on ncurse behave strangely when using
      M-x term

  2/ Although M-x shell has history reference completions, it doesn't complete
      for '!#:<n>'. But '!!:<n>' works fine though.

  3/ There's no readline "reverse-search-history" function (C-r) which does
     an _incremental_  search;

  4/ There's no readline "yank-last-arg" function which inserts the last
      argument to the previous commands

  5/ emacs doesn't expand shell variable when hitting <TAB>


Am I missing something ?

Thanks
--
Francis

Re: What I'm missing when using M-x shell

[Dan Espen]

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

> Hello,
>
> I'm still trying to do all my jobs inside emacs but I'm missing a couple
> things.
>
> The first one is what I can't do anymore when working with a shell/terminal
> inside emacs.
...
> Am I missing something ?

Maybe.

I do all my work inside emacs except for browsing
and 3270 emulation.

I've used M-x shell in the past but not for many years.

Compiles are M-x compile, greps are M-x grep,
ls is dired, email is MH-E.

Maybe most emacs users aren't that committed to M-x shell and
it's friends.

Re: What I'm missing when using M-x shell

[Oleksandr Gavenko]

Francis Moreau wrote:
>   4/ There's no readline "yank-last-arg" function which inserts the last
>       argument to the previous commands

Is the feature useful?
In C-x shell you can move cursor to select any previous typing and 
pressing M-> to back to writing command.

Re: What I'm missing when using M-x shell

[rustom]

On Sep 12, 9:20 pm, "Francis Moreau" <francis.m...@gmail.com> wrote:
> Hello,
>
> I'm still trying to do all my jobs inside emacs but I'm missing a couple
> things.
>
> The first one is what I can't do anymore when working with a shell/terminal
> inside emacs.
>
>   1/ Most applications based on ncurse behave strangely when using
>       M-x term
>
>   2/ Although M-x shell has history reference completions, it doesn't complete
>       for '!#:<n>'. But '!!:<n>' works fine though.
>
>   3/ There's no readline "reverse-search-history" function (C-r) which does
>      an _incremental_  search;
>
>   4/ There's no readline "yank-last-arg" function which inserts the last
>       argument to the previous commands
>
>   5/ emacs doesn't expand shell variable when hitting <TAB>
>
> Am I missing something ?
>
> Thanks
> --
> Francis

shell's inadequacies in emacs are well known.
eshell often works better

Re: What I'm missing when using M-x shell

[Livin Stephen]

On Sep 12, 9:20 pm, "Francis Moreau" <francis.m...@gmail.com> wrote:
> Hello,
>
> I'm still trying to do all my jobs inside emacs but I'm missing a couple
> things.
>
> The first one is what I can't do anymore when working with a shell/terminal
> inside emacs.
>
>   1/ Most applications based on ncurse behave strangely when using
>       M-x term
>
>   2/ Although M-x shell has history reference completions, it doesn't complete
>       for '!#:<n>'. But '!!:<n>' works fine though.
>
>   3/ There's no readline "reverse-search-history" function (C-r) which does
>      an _incremental_  search;
>
>   4/ There's no readline "yank-last-arg" function which inserts the last
>       argument to the previous commands
>
>   5/ emacs doesn't expand shell variable when hitting <TAB>
>
> Am I missing something ?
>
> Thanks
> --
> Francis


Hi, FM.
I too found M-x term to initially fall terribly short - but not
anymore :)

YMMV, but I had to "export TERM=xterm" in my ~/.bashrc  to make M-x
term work as per expectations...
- i.e. to be usable for everyday use: on my Mac I'm perfectly happy
with it's behaviour now!


 (previous value in M-x term term used to show $TERM to be eterm-color
or something like that)

After this change my man-pages, etc don't look like their full of odd
characters (especiallly when trying to print single-quotes), nor is
man-page text underlined randomly .

This TERM change definitely took care of at least issues 3/, and 5/
which you mention
 - I have both "C-r" history-search and "TAB"-completion of shell-vars
AND file-names working.

Cheers
 --livin.stephen

Re: What I'm missing when using M-x shell

[Francis Moreau]

[ please CC me when replying to me ]

On Fri, Sep 12, 2008 at 7:34 PM, Dan Espen
<daneNO@more.mk.spamtelcordia.com> wrote:
> Compiles are M-x compile, greps are M-x grep,
> ls is dired, email is MH-E.

But what would you suggest as replacement for:

$ make && gdb || mail -s "Compilation failed" home

More generally sh scripts allow automatisation processes that you
can't simply do with emacs commands, it similar to using a mouse
actually. I don't think you can get rid of them.

Futhermore there are a lot of tools out there that are not integrated
to emacs if you do so, you don't have enough keys on the keyboard to
create bindings ;)

So I don't think you can realistically get rid of a shell

> Maybe most emacs users aren't that committed to M-x shell and
> it's friends.

Probably.

So for now I don't think I can forget Gnu Screen: one screen to handle
a terminal with a shell where all my ncurse based applications can
work properly and another screen where emacs lives.

Thanks
-- 
Francis

Re: What I'm missing when using M-x shell

[Francis Moreau]

Hello,

On Fri, Sep 12, 2008 at 7:59 PM, Oleksandr Gavenko <gavenkoa@gmail.com> wrote:
> Francis Moreau wrote:
>>
>>  4/ There's no readline "yank-last-arg" function which inserts the last
>>      argument to the previous commands
>
> Is the feature useful?
> In C-x shell you can move cursor to select any previous typing and pressing
> M-> to back to writing command.
>

Do you realize how many key pressings you're doing to do what you described ?

In my use case I only press one time.

-- 
Francis

Re: What I'm missing when using M-x shell

[Francis Moreau]

Hello,

On Sat, Sep 13, 2008 at 8:10 AM, rustom <rustompmody@gmail.com> wrote:
>
> shell's inadequacies in emacs are well known.

Do you have any pointers that describes them ?

> eshell often works better

I just try history reference completion, and it's even worse...

thanks
-- 
Francis

Re: What I'm missing when using M-x shell

[Francis Moreau]

Hello,

On Sat, Sep 13, 2008 at 10:08 AM, Livin Stephen <livin.stephen@gmail.com> wrote:
>
> YMMV, but I had to "export TERM=xterm" in my ~/.bashrc  to make M-x
> term work as per expectations...

hmm I'm probably missing something here but why exporting TERM=xterm
would change anything since it is already if you're using a window
manager.

> - i.e. to be usable for everyday use: on my Mac I'm perfectly happy
> with it's behaviour now!

ah ok you're using a Mac, so that explains probably why you need to export TERM

>  (previous value in M-x term term used to show $TERM to be eterm-color
> or something like that)
>

yes and that should be again the case. Let see:

  $ TERM=xterm emacs
  < M-x term>
  $ echo $TERM
  eterm-color

I don't understand how you did that.

> After this change my man-pages, etc don't look like their full of odd
> characters (especiallly when trying to print single-quotes), nor is
> man-page text underlined randomly .

did you try to start mutt for example ?

> This TERM change definitely took care of at least issues 3/, and 5/
> which you mention
>  - I have both "C-r" history-search and "TAB"-completion of shell-vars
> AND file-names working.

I don't think so, I'm talking about expansion, not completion for issue 5)

Thanks
-- 
Francis

Re: What I'm missing when using M-x shell

[Livin Stephen Sharma]

[-- Attachment #1: Type: text/plain, Size: 1790 bytes --]

2008/9/13 Francis Moreau <francis.moro@gmail.com>

> Hello,
>
> On Sat, Sep 13, 2008 at 10:08 AM, Livin Stephen <livin.stephen@gmail.com>
> wrote:
> >
> > YMMV, but I had to "export TERM=xterm"...
>
> hmm I'm probably missing something here but why exporting TERM=xterm
> would change anything since it is already if you're using a window
> manager.
>
> > - i.e. to be usable for everyday use...
>
> ah ok you're using a Mac, so that explains probably why you need to export
> TERM


Actually, without exporting TERM, I saw emacs' term to show it to be
"eterm-color":
 the need to export it was to get it to be "xterm" instead.


>
> >  (previous value in...



>
>
> yes and that should be again the case. Let see:
>
>  $ TERM=xterm emacs
>  < M-x term>
>  $ echo $TERM
>  eterm-color
>
> I don't understand how you did that.



I set TERM=xterm ***in my /home/<user>/.bashrc***
 (**not** on the command-line prior to invoking emacs).

 My only guess is that When "M-x term" starts up a shell (/bin/bash for me)
it would 'load' the .bashrc file from my home directory. Maybe that way
you'd wouldn't get "eterm-color" within M-x term. Worth a shot ?



>
> > After this change my man-pages, etc don't look like their full of odd
> > characters (especiallly when trying to print single-quotes), nor is
> > man-page text underlined randomly .
>
> did you try to start mutt for example ?


No, I don't use mutt, so I can't comment on that :) .


>
> > This TERM change definitely took care of at least issues 3/, and 5/
> > which you mention
> >  - I have both "C-r" history-search and "TAB"-completion of shell-vars
> > AND file-names working.
>
> I don't think so, I'm talking about expansion, not completion for issue 5)



Yes, I misunderstood that.

-- LSS



>
> Thanks
> --
> Francis
>

[-- Attachment #2: Type: text/html, Size: 3340 bytes --]

Re: What I'm missing when using M-x shell

[David Hansen]

On Sat, 13 Sep 2008 01:08:19 -0700 (PDT) Livin Stephen wrote:

> I too found M-x term to initially fall terribly short - but not
> anymore :)
>
> YMMV, but I had to "export TERM=xterm" in my ~/.bashrc  to make M-x
> term work as per expectations...
> - i.e. to be usable for everyday use: on my Mac I'm perfectly happy
> with it's behaviour now!

You should install the terminfo stuff from /etc/e.  The emacs term for
sure is different from an xterm.

David

Re: What I'm missing when using M-x shell

[Francis Moreau]

On Sat, Sep 13, 2008 at 1:21 PM, Livin Stephen Sharma
<livin.stephen@gmail.com> wrote:
>
> I set TERM=xterm ***in my /home/<user>/.bashrc***
>  (**not** on the command-line prior to invoking emacs).
>  My only guess is that When "M-x term" starts up a shell (/bin/bash for me)
> it would 'load' the .bashrc file from my home directory. Maybe that way
> you'd wouldn't get "eterm-color" within M-x term. Worth a shot ?

Ah ok.

But you're claiming your terminal has xterm capabilites whereas M-x
term has probably not. I don't know why it improves thing for you
though.

And no, I tried this and it doesn't improve anything.

-- 
Francis

Re: What I'm missing when using M-x shell

[Dan Espen]

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

> [ please CC me when replying to me ]
>
> On Fri, Sep 12, 2008 at 7:34 PM, Dan Espen
> <daneNO@more.mk.spamtelcordia.com> wrote:
>> Compiles are M-x compile, greps are M-x grep,
>> ls is dired, email is MH-E.
>
> But what would you suggest as replacement for:
>
> $ make && gdb || mail -s "Compilation failed" home

First, the make is always run as M-x compile.

In my case, the most likely outcome from M-x compile is
that I'll be correcting syntax errors.  Since I've run
M-x compile, M-x next-error becomes really useful.

So, no make from the command line for me.

I'm got a big fan of running under gdb.
If I did want to run under gdb a lot, I'd probably figure out
Emacs gdb.

Typically my make target is not just a compile but a compile
and test and check results.  No need to send myself
an email since the M-x compile clearly shows whether the
compile/test worked.

Re: What I'm missing when using M-x shell

[Francis Moreau]

On Sat, Sep 13, 2008 at 5:42 PM, Dan Espen
<daneNO@more.mk.spamtelcordia.com> wrote:
> "Francis Moreau" <francis.moro@gmail.com> writes:
>
>> [ please CC me when replying to me ]
>>
>> On Fri, Sep 12, 2008 at 7:34 PM, Dan Espen
>> <daneNO@more.mk.spamtelcordia.com> wrote:
>>> Compiles are M-x compile, greps are M-x grep,
>>> ls is dired, email is MH-E.
>>
>> But what would you suggest as replacement for:
>>
>> $ make && gdb || mail -s "Compilation failed" home
>
> First, the make is always run as M-x compile.
>
> In my case, the most likely outcome from M-x compile is
> that I'll be correcting syntax errors.  Since I've run
> M-x compile, M-x next-error becomes really useful.
>
> So, no make from the command line for me.
>
> I'm got a big fan of running under gdb.
> If I did want to run under gdb a lot, I'd probably figure out
> Emacs gdb.
>
> Typically my make target is not just a compile but a compile
> and test and check results.  No need to send myself
> an email since the M-x compile clearly shows whether the
> compile/test worked.
>

That was just a (silly) example where scripting can be powerful and
can't be replace by some M-x <cmd> or whatever...

-- 
Francis

Re: What I'm missing when using M-x shell

[Thierry Volpiatto]

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

> On Sat, Sep 13, 2008 at 5:42 PM, Dan Espen
> <daneNO@more.mk.spamtelcordia.com> wrote:
>> "Francis Moreau" <francis.moro@gmail.com> writes:
>>
>>> [ please CC me when replying to me ]
>>>
>>> On Fri, Sep 12, 2008 at 7:34 PM, Dan Espen
>>> <daneNO@more.mk.spamtelcordia.com> wrote:
>>>> Compiles are M-x compile, greps are M-x grep,
>>>> ls is dired, email is MH-E.
>>>
>>> But what would you suggest as replacement for:
>>>
>>> $ make && gdb || mail -s "Compilation failed" home
>>
>> First, the make is always run as M-x compile.
>>
>> In my case, the most likely outcome from M-x compile is
>> that I'll be correcting syntax errors.  Since I've run
>> M-x compile, M-x next-error becomes really useful.
>>
>> So, no make from the command line for me.
>>
>> I'm got a big fan of running under gdb.
>> If I did want to run under gdb a lot, I'd probably figure out
>> Emacs gdb.
>>
>> Typically my make target is not just a compile but a compile
>> and test and check results.  No need to send myself
>> an email since the M-x compile clearly shows whether the
>> compile/test worked.
>>
>
> That was just a (silly) example where scripting can be powerful and
> can't be replace by some M-x <cmd> or whatever...

Did you try using a window manager that work correctly with emacs+shell
like stumpwm or rat-poison.
I am using stumpwm and i am really happy with it.
I can switch to a shell at anytime just like in emacs.

Have also a look at eev:(you may like that if you are a fan of shell)

http://angg.twu.net/eev-article.html

and try to eval from anywhere in emacs (C-x C-e) for example:

(find-sh "ps -u $USER | grep emacs | awk '{print $1}'") 

Try also to use M-x ansi-term that is much better than term.
You can run many shell in emacs and more.
 
-- 
A + Thierry Volpiatto
Location: Saint-Cyr-Sur-Mer - France

Re: What I'm missing when using M-x shell

[Francis Moreau]

Hello,

On Sat, Sep 13, 2008 at 7:32 PM, Thierry Volpiatto
<thierry.volpiatto@gmail.com> wrote:
> Did you try using a window manager that work correctly with emacs+shell
> like stumpwm or rat-poison.
> I am using stumpwm and i am really happy with it.
> I can switch to a shell at anytime just like in emacs.
>

No I haven't and I don't think I will have some spare time to try them.

For now I'm using Screen with emacs and it works ok.

> Have also a look at eev:(you may like that if you are a fan of shell)
>
> http://angg.twu.net/eev-article.html
>
> and try to eval from anywhere in emacs (C-x C-e) for example:
>
> (find-sh "ps -u $USER | grep emacs | awk '{print $1}'")
>

Interesting thanks.

> Try also to use M-x ansi-term that is much better than term.

I haven't seen any differences. Maybe because they both use an
'eterm-color' terminal.

> You can run many shell in emacs and more.

That's the problem I think.

Too many way to achieve the main goal.
Why not doing one implementation and doing it well ?

-- 
Francis

Re: What I'm missing when using M-x shell

[Thierry Volpiatto]

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

> Hello,
>
> On Sat, Sep 13, 2008 at 7:32 PM, Thierry Volpiatto
> <thierry.volpiatto@gmail.com> wrote:
>> Did you try using a window manager that work correctly with emacs+shell
>> like stumpwm or rat-poison.
>> I am using stumpwm and i am really happy with it.
>> I can switch to a shell at anytime just like in emacs.
>>
>
> No I haven't and I don't think I will have some spare time to try them.
>
> For now I'm using Screen with emacs and it works ok.
>
>> Have also a look at eev:(you may like that if you are a fan of shell)
>>
>> http://angg.twu.net/eev-article.html
>>
>> and try to eval from anywhere in emacs (C-x C-e) for example:
>>
>> (find-sh "ps -u $USER | grep emacs | awk '{print $1}'")
>>
>
> Interesting thanks.
>
>> Try also to use M-x ansi-term that is much better than term.
>
> I haven't seen any differences. Maybe because they both use an
> 'eterm-color' terminal.
>
>> You can run many shell in emacs and more.
>
> That's the problem I think.
>
> Too many way to achieve the main goal.
> Why not doing one implementation and doing it well ?
No, what i mean is that with M-x term, you can run only one shell.
with ansi-term you can start more than one.
*ansi-term*<2>
*ansi-term*<3>
etc...
-- 
A + Thierry Volpiatto
Location: Saint-Cyr-Sur-Mer - France

Re: What I'm missing when using M-x shell

[Chris F.A. Johnson]

On 2008-09-12, Francis Moreau wrote:
...
>   4/ There's no readline "yank-last-arg" function which inserts the last
>       argument to the previous commands

   C-up

-- 
   Chris F.A. Johnson, webmaster         <http://Woodbine-Gerrard.com>
   ===================================================================
   Author:
   Shell Scripting Recipes: A Problem-Solution Approach (2005, Apress)

Re: What I'm missing when using M-x shell

[Tim X]

Livin Stephen <livin.stephen@gmail.com> writes:

> On Sep 12, 9:20 pm, "Francis Moreau" <francis.m...@gmail.com> wrote:
>> Hello,
>>
>> I'm still trying to do all my jobs inside emacs but I'm missing a couple
>> things.
>>
>> The first one is what I can't do anymore when working with a shell/terminal
>> inside emacs.
>>
>>   1/ Most applications based on ncurse behave strangely when using
>>       M-x term
>>
>>   2/ Although M-x shell has history reference completions, it doesn't complete
>>       for '!#:<n>'. But '!!:<n>' works fine though.
>>
>>   3/ There's no readline "reverse-search-history" function (C-r) which does
>>      an _incremental_  search;
>>
>>   4/ There's no readline "yank-last-arg" function which inserts the last
>>       argument to the previous commands
>>
>>   5/ emacs doesn't expand shell variable when hitting <TAB>
>>
>> Am I missing something ?
>>
>> Thanks
>> --
>> Francis
>
>
> Hi, FM.
> I too found M-x term to initially fall terribly short - but not
> anymore :)
>
> YMMV, but I had to "export TERM=xterm" in my ~/.bashrc  to make M-x
> term work as per expectations...
> - i.e. to be usable for everyday use: on my Mac I'm perfectly happy
> with it's behaviour now!
>
>
>  (previous value in M-x term term used to show $TERM to be eterm-color
> or something like that)
>
> After this change my man-pages, etc don't look like their full of odd
> characters (especiallly when trying to print single-quotes), nor is
> man-page text underlined randomly .
>
> This TERM change definitely took care of at least issues 3/, and 5/
> which you mention
>  - I have both "C-r" history-search and "TAB"-completion of shell-vars
> AND file-names working.
>

I think the reason you got better results setting the term to xterm
rather than eterm-color is to do with terminfo. 

There are many programs that when executed check the terminfo database
to find out what the capabilities of the terminal are. Often, a program
will drop back to a 'dumb' terminal setting if the terminal returned by
$TERM is not in the terminfo database. 

IIRC, eterm-color is not set in the terminfo database on most systems,
so when you run a program, it cannot determine what the
capabilities/characteristics of the terminal your running under are. 

Setting the TERM variable to xterm probably gives resonable results
because the capabilities are similar. However, you would probably get
even better results by adding eterm-color to your terminfo database. I
think you will find the relevant information in the Emacs etc/e
directory. 

I'm surprised setting TERM=xterm and exporting it prior to running M-x
term works at all. The TERM setting is (I think) set as part of the
startup (see term.el). I'm assuming what you are doing is issuing an
export TERM=xterm or something similar once you have started M-x term?

On my system, Debian, I've put the eterm-color terminfo file in
/etc/terminfo/e/ and I don't seem to have many of the problems being
reported under M-x term. For example, if I enter echo $T and hit tab, it
is completed to echo $TERM. 

HTH

Tim


-- 
tcross (at) rapttech dot com dot au

Re: What I'm missing when using M-x shell

[Tim X]

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

> [ please CC me when replying to me ]
>
> On Fri, Sep 12, 2008 at 7:34 PM, Dan Espen
> <daneNO@more.mk.spamtelcordia.com> wrote:
>> Compiles are M-x compile, greps are M-x grep,
>> ls is dired, email is MH-E.
>
> But what would you suggest as replacement for:
>
> $ make && gdb || mail -s "Compilation failed" home
>

Surely you don't use such a command when doing things interactively? A
command like that is something I would use in an automated build process
that isn't run interactively. For example, I've used something similar
in a envrionment where we wanted to rebuild each night from SVN. 

This to me is a different question to working interactively. I don't see
you cold ever eliminate the shell completely as many non-interactive
processes use it. However, I do believe you can work totally through
emacs. In fact, I know you can because that is how I *have* to work. 

I'm a blind programmer and work almost exclusively in the Unix world. My
main interface to the system is through emacs and using a package called
emacspeak. In general, if there is an emacs specific package to do a
job, I use that (e.g. dired, tramp, VM, w3m, view-process (now using
proced) ediff, PCL-CVS, psvn, man/woman, etc) and then revert to M-x shell, M-x
term and M-x eshell when I need access to the shell or need to run a
program directly in a shell etc. 

Though I've not used them, you also have elisp clones of things like
screen and even basic window management packages. At one time I even
used emacs as my window manager under X e.g. last part in my
.xsession file was to try and run emacs and if that failed, run an
exterm that executed a program that told me emacs had failed and started
a very basic screen reader program in an xterm. 


> More generally sh scripts allow automatisation processes that you
> can't simply do with emacs commands, it similar to using a mouse
> actually. I don't think you can get rid of them.
>

I use lots of shell scripts to do various things, but I run them from
within emacs and in some cases, have them bound to specific key
bindings. 

> Futhermore there are a lot of tools out there that are not integrated
> to emacs if you do so, you don't have enough keys on the keyboard to
> create bindings ;)
>
yes, there are a lot of things that don't have 'native' emacs
modes. However, I disagree there are not enough available key bindings -
there are more than anyone is ever going to remember. What a lot of
people don't do or possibly don't even realise is that in addition to M-
and C- for meta and control bindings, you also have hyper and
super. These can be bound to keys like the windows menu key and some of
those other often unused keys on your keyboard. Combine those with shift
and you have a lot of combinations. Admittedly, some of them can be
uncomfortable to use, but thats a different issue and depends on things
liike the type of keyborad, your style of typeing and even things like
the size of your hands, length of your fingers, degree of flexibility
you have in your hands, table height and chair. 



> So I don't think you can realistically get rid of a shell
>

You can't get rid of the shell - its fundamental to how the system
works. However, you can create an environment wehre you only interact
with the shell through emacs without losing any functionality or
flexibility. It may take some work to get it configured, but thats a
different issue. 


> So for now I don't think I can forget Gnu Screen: one screen to handle
> a terminal with a shell where all my ncurse based applications can
> work properly and another screen where emacs lives.
>

Your ncurses applications should behave fine under M-x term provided you
have the correct terminfo information setup. Things can be a little more
complex if you want to un ncurses programs over ssh on another host, but
this can be fixed by adding the necessary terminfo data to the remote
servers. 

Tim


-- 
tcross (at) rapttech dot com dot au

Re: What I'm missing when using M-x shell

[Francis Moreau]

On Sun, Sep 14, 2008 at 12:43 AM, Chris F.A. Johnson
<cfajohnson@gmail.com> wrote:
> On 2008-09-12, Francis Moreau wrote:
> ...
>>   4/ There's no readline "yank-last-arg" function which inserts the last
>>       argument to the previous commands
>
>   C-up
>

really ?

could you tell me on which function this is binded ?

thanks
-- 
Francis

Re: What I'm missing when using M-x shell

[Chris F.A. Johnson]

On 2008-09-14, Francis Moreau wrote:
> On Sun, Sep 14, 2008 at 12:43 AM, Chris F.A. Johnson
><cfajohnson@gmail.com> wrote:
>> On 2008-09-12, Francis Moreau wrote:
>> ...
>>>   4/ There's no readline "yank-last-arg" function which inserts the last
>>>       argument to the previous commands
>>
>>   C-up
>>
>
> really ?
>
> could you tell me on which function this is binded ?

        <C-up> runs the command comint-previous-input

-- 
   Chris F.A. Johnson, webmaster         <http://Woodbine-Gerrard.com>
   ===================================================================
   Author:
   Shell Scripting Recipes: A Problem-Solution Approach (2005, Apress)

Re: What I'm missing when using M-x shell

[Thierry Volpiatto]

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

> On Sun, Sep 14, 2008 at 12:43 AM, Chris F.A. Johnson
> <cfajohnson@gmail.com> wrote:
>> On 2008-09-12, Francis Moreau wrote:
>> ...
>>>   4/ There's no readline "yank-last-arg" function which inserts the last
>>>       argument to the previous commands
>>
>>   C-up
>>
>
> really ?
>
> could you tell me on which function this is binded ?

<C-up> runs the command comint-previous-input

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

RE: What I'm missing when using M-x shell

[Drew Adams]

> >>>   4/ There's no readline "yank-last-arg" function which 
> >>>      inserts the last argument to the previous commands
> >>
> >>   C-up
> >
> > really ?
> >
> > could you tell me on which function this is binded ?
> 
>         <C-up> runs the command comint-previous-input

FWIW, if you use Icicles, then `C-c TAB' in any comint (aka shell) mode lets you
cycle past inputs or complete against them (with regexp/substring completion
etc.). And `C-c `' also does these things (but differently).

http://www.emacswiki.org/cgi-bin/wiki/Icicles_-_Other_Search_Commands#IciclesShe
ll

Re: What I'm missing when using M-x shell

[Barry Margolin]

In article <mailman.19222.1221417839.18990.help-gnu-emacs@gnu.org>,
 Thierry Volpiatto <thierry.volpiatto@gmail.com> wrote:

> "Francis Moreau" <francis.moro@gmail.com> writes:
> 
> > On Sun, Sep 14, 2008 at 12:43 AM, Chris F.A. Johnson
> > <cfajohnson@gmail.com> wrote:
> >> On 2008-09-12, Francis Moreau wrote:
> >> ...
> >>>   4/ There's no readline "yank-last-arg" function which inserts the last
> >>>       argument to the previous commands
> >>
> >>   C-up
> >>
> >
> > really ?
> >
> > could you tell me on which function this is binded ?
> 
> <C-up> runs the command comint-previous-input

But that yanks the whole command line, not just the last argument.

I still the shell's history features for !$ and !*.  The nice thing is 
that Emacs recognizes them and substitutes them into its history list, 
so that if I use commands like C-up it will yank them with the 
appropriate values.

-- 
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: What I'm missing when using M-x shell

[Francis Moreau]

On Sun, Sep 14, 2008 at 8:33 PM, Chris F.A. Johnson
<cfajohnson@gmail.com> wrote:
> On 2008-09-14, Francis Moreau wrote:
>> On Sun, Sep 14, 2008 at 12:43 AM, Chris F.A. Johnson
>><cfajohnson@gmail.com> wrote:
>>> On 2008-09-12, Francis Moreau wrote:
>>> ...
>>>>   4/ There's no readline "yank-last-arg" function which inserts the last
>>>>       argument to the previous commands
>>>
>>>   C-up
>>>
>>
>> really ?
>>
>> could you tell me on which function this is binded ?
>
>        <C-up> runs the command comint-previous-input
>

no this paste the whole command line, I'm asking for the last _argument_

-- 
Francis

Re: What I'm missing when using M-x shell

[Chris F.A. Johnson]

On 2008-09-14, Francis Moreau wrote:
> On Sun, Sep 14, 2008 at 8:33 PM, Chris F.A. Johnson
><cfajohnson@gmail.com> wrote:
>> On 2008-09-14, Francis Moreau wrote:
>>> On Sun, Sep 14, 2008 at 12:43 AM, Chris F.A. Johnson
>>><cfajohnson@gmail.com> wrote:
>>>> On 2008-09-12, Francis Moreau wrote:
>>>> ...
>>>>>   4/ There's no readline "yank-last-arg" function which inserts the last
>>>>>       argument to the previous commands
>>>>
>>>>   C-up
>>>>
>>>
>>> really ?
>>>
>>> could you tell me on which function this is binded ?
>>
>>        <C-up> runs the command comint-previous-input
>>
>
> no this paste the whole command line, I'm asking for the last _argument_

(defun last-arg () "" (interactive)
  (comint-previous-input 1)
  (search-backward " ")
  (kill-line)
  (comint-next-input 2)
  (yank)
)


-- 
   Chris F.A. Johnson, webmaster         <http://Woodbine-Gerrard.com>
   ===================================================================
   Author:
   Shell Scripting Recipes: A Problem-Solution Approach (2005, Apress)

Re: What I'm missing when using M-x shell

[Francis Moreau]

On Sun, Sep 14, 2008 at 8:55 PM, Barry Margolin <barmar@alum.mit.edu> wrote:
> In article <mailman.19222.1221417839.18990.help-gnu-emacs@gnu.org>,
>  Thierry Volpiatto <thierry.volpiatto@gmail.com> wrote:
>>
>> <C-up> runs the command comint-previous-input
>
> But that yanks the whole command line, not just the last argument.
>

Actually thinking about it, it should be quite easy to achieve this:
just create a key binding to evaluate "!$<TAB>" sequence.

-- 
Francis

Re: What I'm missing when using M-x shell

[Francis Moreau]

On Sun, Sep 14, 2008 at 10:28 PM, Francis Moreau <francis.moro@gmail.com> wrote:
> On Sun, Sep 14, 2008 at 8:55 PM, Barry Margolin <barmar@alum.mit.edu> wrote:
>> In article <mailman.19222.1221417839.18990.help-gnu-emacs@gnu.org>,
>>  Thierry Volpiatto <thierry.volpiatto@gmail.com> wrote:
>>>
>>> <C-up> runs the command comint-previous-input
>>
>> But that yanks the whole command line, not just the last argument.
>>
>
> Actually thinking about it, it should be quite easy to achieve this:
> just create a key binding to evaluate "!$<TAB>" sequence.
>

well to be more accurate we should create 'something' that evaluate to

!-1:$

when call for the first time.

If a second call is done without any <RETURN> key press then evaluate to

!-2:$

etc...

Though I dunno how to write that in elisp..
-- 
Francis

Re: What I'm missing when using M-x shell

[Francis Moreau]

On Sun, Sep 14, 2008 at 10:27 PM, Chris F.A. Johnson
<cfajohnson@gmail.com> wrote:
> On 2008-09-14, Francis Moreau wrote:
>> no this paste the whole command line, I'm asking for the last _argument_
>
> (defun last-arg () "" (interactive)
>  (comint-previous-input 1)
>  (search-backward " ")
>  (kill-line)
>  (comint-next-input 2)
>  (yank)
> )
>

Even though I don't speak (yet) elisp, it does seem that if I call
last-arg() several times then this yanks always the same arg...

Try in an xterm to use M-. to see what I'm talkng about.

Thanks
-- 
Francis

Re: What I'm missing when using M-x shell

[Jonathan Groll]

On Sat, Sep 13, 2008 at 10:27:47AM +0200, Francis Moreau wrote:
>[ please CC me when replying to me ]
>
>On Fri, Sep 12, 2008 at 7:34 PM, Dan Espen
><daneNO@more.mk.spamtelcordia.com> wrote:
>> Compiles are M-x compile, greps are M-x grep,
>> ls is dired, email is MH-E.
>
>But what would you suggest as replacement for:
>
>$ make && gdb || mail -s "Compilation failed" home
>
>More generally sh scripts allow automatisation processes that you
>can't simply do with emacs commands, it similar to using a mouse
>actually. I don't think you can get rid of them.
>
>Futhermore there are a lot of tools out there that are not integrated
>to emacs if you do so, you don't have enough keys on the keyboard to
>create bindings ;)
>
>So I don't think you can realistically get rid of a shell
>
>> Maybe most emacs users aren't that committed to M-x shell and
>> it's friends.
>
>Probably.
>
>So for now I don't think I can forget Gnu Screen: one screen to handle
>a terminal with a shell where all my ncurse based applications can
>work properly and another screen where emacs lives.
>

On the subject of GNU screen and emacs: Bill Clementson wrote a good
entry on using emacsclient and screen -
http://bc.tech.coop/blog/071001.html

Regarding mutt with emacs, there is a page on the emacswiki you may
want to read:
http://www.emacswiki.org/cgi-bin/emacs-en/MuttInEmacs

It may also possibly be necessary to adjust the colour scheme used in
.muttrc:
http://www.mail-archive.com/mutt-users@mutt.org/msg21810.html

Personally, I have been using gnu-screen with an emacs server running
in one window and mutt in another window, based on what Bill
Clementson wrote.

Regards,
Jonathan

Re: What I'm missing when using M-x shell

[Livin Stephen]

On Sep 14, 7:20 am, Tim X <t...@nospam.dev.null> wrote:
> .....
>
> I think the reason you got better results setting the term to xterm
> rather than eterm-color is to do with terminfo.
>
> ...
>
> Tim
>
> --
> tcross (at) rapttech dot com dot au



Tim,
 Thanks for a very informative post!
 This is something I'll check out myself.


Tim, Francis,
 ... I know some keyboard shortcuts (like M-t, C-k, C-r) that work in,
say bash, but I always thought of them as emacs-ish shortcuts
conveniently available there. Then seeing the man-page for 'bash' i
came across 'Readline' which seems to actually provide these shortcuts
and functions.

 So my question is:
How do you invoke commands like (yank-last-arg) in any shell [ not
just emacs' term ] if they don't have a keybinding ?
For example,
invoke 'kill-word' without using the shortcut key 'M-d'. Is this
possible ?

[ Posting here so as to not start another term/shell thread ]

Re: What I'm missing when using M-x shell

[Francis Moreau]

Hello.

On Mon, Sep 15, 2008 at 9:32 AM, Jonathan Groll <lists@groll.co.za> wrote:
>
> On the subject of GNU screen and emacs: Bill Clementson wrote a good
> entry on using emacsclient and screen -
> http://bc.tech.coop/blog/071001.html

Thanks !

That was interesting.

I actually had a similar concern but come up with a different solution.
I use desktop in my main projects directory and in my home directory.
When starting emacs in one of those directories, I start an emacs server
whose name is exported through a environment variable. This variable
has a different value depending on the desktop I used.

After that in my bashrc I do:

export EDITOR="emacsclient -s $EMACS_SERVER_NAME"

and that's it. Each time I start a term in emacs and start an application
that uses EDITOR, it start an emacsclient in the ritght emacs desktop.

But I agree this solution works only inside an emacs term unlike the
solution you pointed out. But that's my first step in "do it all in emacs !"

;)

And as the author said with emacs tty-muti-support, this is now useless.

>
> Regarding mutt with emacs, there is a page on the emacswiki you may
> want to read:
> http://www.emacswiki.org/cgi-bin/emacs-en/MuttInEmacs
>
> It may also possibly be necessary to adjust the colour scheme used in
> .muttrc:
> http://www.mail-archive.com/mutt-users@mutt.org/msg21810.html
>
> Personally, I have been using gnu-screen with an emacs server running
> in one window and mutt in another window, based on what Bill
> Clementson wrote.

Thanks again but I'm planning to get rid of mutt and to use either gnus or
mh-e. I need to test both of them first to get an opinion.

-- 
Francis

Re: What I'm missing when using M-x shell

[Francis Moreau]

Hello,

On Mon, Sep 15, 2008 at 9:56 AM, Livin Stephen <livin.stephen@gmail.com> wrote:
>  ... I know some keyboard shortcuts (like M-t, C-k, C-r) that work in,
> say bash, but I always thought of them as emacs-ish shortcuts
> conveniently available there. Then seeing the man-page for 'bash' i
> came across 'Readline' which seems to actually provide these shortcuts
> and functions.
>
>  So my question is:
> How do you invoke commands like (yank-last-arg) in any shell [ not
> just emacs' term ] if they don't have a keybinding ?
> For example,
> invoke 'kill-word' without using the shortcut key 'M-d'. Is this
> possible ?

I don't think you can but I'm really not sure.

The only thing I know for sure is that you can use ~/.inputrc to map some
binding with realdine functions. But this is well described in bash man page.


-- 
Francis

Re: What I'm missing when using M-x shell

[Francis Moreau]

On Mon, Sep 15, 2008 at 10:42 AM, Francis Moreau <francis.moro@gmail.com> wrote:
> But I agree this solution works only inside an emacs term unlike the
> solution you pointed out. But that's my first step in "do it all in emacs !"
>

OK I tried the solution you pointed out for all emacsclient started outside
an emacs session and it works great !

Thanks.
-- 
Francis

Re: What I'm missing when using M-x shell

[Bernardo Bacic]

>>>
>>> could you tell me on which function this is binded ?
>>        <C-up> runs the command comint-previous-input
>>
> 
> no this paste the whole command line, I'm asking for the last _argument_
> 

FWIW (probably bash specific)

M-x shell
$ ls -l notes.txt
-rw-rw---- 1 user user 196307 2008-09-14 21:29 notes.txt
$ ls $_
notes.txt
$ echo $_
notes.txt

Re: What I'm missing when using M-x shell

[Francis Moreau]

On Mon, Sep 15, 2008 at 8:32 AM, Francis Moreau <francis.moro@gmail.com> wrote:
> On Sun, Sep 14, 2008 at 10:27 PM, Chris F.A. Johnson
> <cfajohnson@gmail.com> wrote:
>> On 2008-09-14, Francis Moreau wrote:
>>> no this paste the whole command line, I'm asking for the last _argument_
>>
>> (defun last-arg () "" (interactive)
>>  (comint-previous-input 1)
>>  (search-backward " ")
>>  (kill-line)
>>  (comint-next-input 2)
>>  (yank)
>> )
>>
>
> Even though I don't speak (yet) elisp, it does seem that if I call
> last-arg() several times then this yanks always the same arg...
>
> Try in an xterm to use M-. to see what I'm talkng about.
>

Actually I'm wondering why not doing in M-x shell a char mode present
in M-x term shell. I found it very convenient.

-- 
Francis

Re: What I'm missing when using M-x shell

[Thierry Volpiatto]

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

> On Sun, Sep 14, 2008 at 10:27 PM, Chris F.A. Johnson
> <cfajohnson@gmail.com> wrote:
>> On 2008-09-14, Francis Moreau wrote:
>>> no this paste the whole command line, I'm asking for the last _argument_
>>
>> (defun last-arg () "" (interactive)
>>  (comint-previous-input 1)
>>  (search-backward " ")
>>  (kill-line)
>>  (comint-next-input 2)
>>  (yank)
>> )
>>
>
> Even though I don't speak (yet) elisp, it does seem that if I call
> last-arg() several times then this yanks always the same arg...
>
> Try in an xterm to use M-. to see what I'm talkng about.

May be what you are looking for...

in M-x shell:
,----
| C-c .  ==>  comint-insert-previous-argument
`----

and with ansi-term ==> the command M-. work
it have the same behavior than in my urxvt

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

Re: What I'm missing when using M-x shell

[Francis Moreau]

On Mon, Sep 15, 2008 at 1:38 PM, Thierry Volpiatto
<thierry.volpiatto@gmail.com> wrote:
> "Francis Moreau" <francis.moro@gmail.com> writes:
>
>> On Sun, Sep 14, 2008 at 10:27 PM, Chris F.A. Johnson
>> <cfajohnson@gmail.com> wrote:
>>> On 2008-09-14, Francis Moreau wrote:
>>>> no this paste the whole command line, I'm asking for the last _argument_
>>>
>>> (defun last-arg () "" (interactive)
>>>  (comint-previous-input 1)
>>>  (search-backward " ")
>>>  (kill-line)
>>>  (comint-next-input 2)
>>>  (yank)
>>> )
>>>
>>
>> Even though I don't speak (yet) elisp, it does seem that if I call
>> last-arg() several times then this yanks always the same arg...
>>
>> Try in an xterm to use M-. to see what I'm talkng about.
>
> May be what you are looking for...
>
> in M-x shell:
> ,----
> | C-c .  ==>  comint-insert-previous-argument
> `----
>

Damn yes it is !

While at it, do you have any idea if it's possible to switch between char/line
mode in M-x shell ?

Thanks
-- 
Francis

Re: What I'm missing when using M-x shell

[Thierry Volpiatto]

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

> On Mon, Sep 15, 2008 at 1:38 PM, Thierry Volpiatto
> <thierry.volpiatto@gmail.com> wrote:
>> "Francis Moreau" <francis.moro@gmail.com> writes:
>>
>>> On Sun, Sep 14, 2008 at 10:27 PM, Chris F.A. Johnson
>>> <cfajohnson@gmail.com> wrote:
>>>> On 2008-09-14, Francis Moreau wrote:
>>>>> no this paste the whole command line, I'm asking for the last _argument_
>>>>
>>>> (defun last-arg () "" (interactive)
>>>>  (comint-previous-input 1)
>>>>  (search-backward " ")
>>>>  (kill-line)
>>>>  (comint-next-input 2)
>>>>  (yank)
>>>> )
>>>>
>>>
>>> Even though I don't speak (yet) elisp, it does seem that if I call
>>> last-arg() several times then this yanks always the same arg...
>>>
>>> Try in an xterm to use M-. to see what I'm talkng about.
>>
>> May be what you are looking for...
>>
>> in M-x shell:
>> ,----
>> | C-c .  ==>  comint-insert-previous-argument
>> `----
>>
>
> Damn yes it is !
>
> While at it, do you have any idea if it's possible to switch between char/line
> mode in M-x shell ?

I don't know that, what is the equivalent in an xterm, and what is that
for ?

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

Re: What I'm missing when using M-x shell

[Francis Moreau]

On Mon, Sep 15, 2008 at 3:00 PM, Thierry Volpiatto
<thierry.volpiatto@gmail.com> wrote:
> "Francis Moreau" <francis.moro@gmail.com> writes:
>> While at it, do you have any idea if it's possible to switch between char/line
>> mode in M-x shell ?
>
> I don't know that, what is the equivalent in an xterm, and what is that
> for ?
>

no it's part of M-x term.

Basically all you type in char mode is sent to the terminal (except C-c).

-- 
Francis

Re: What I'm missing when using M-x shell

[Oleksandr Gavenko]

Francis Moreau пишет:
> Hello,
> 
> On Fri, Sep 12, 2008 at 7:59 PM, Oleksandr Gavenko <gavenkoa@gmail.com> wrote:
>> Francis Moreau wrote:
>>>  4/ There's no readline "yank-last-arg" function which inserts the last
>>>      argument to the previous commands
>> Is the feature useful?
>> In C-x shell you can move cursor to select any previous typing and pressing
>> M-> to back to writing command.
>>
> 
> Do you realize how many key pressings you're doing to do what you described ?
> 
> In my use case I only press one time.
> 
Using key typo depend of your task. I write sh, C, Java related type
of files, often on Windows (at work) and so I do not need remember
specials key binding for a such task, I construct it from that I know.

If you need more than 3 time for a 10 second period do such task
you have reason.