Emacs 21.3.1 -nw over ssh, C-h not bringing up help

Emacs 21.3.1 -nw over ssh, C-h not bringing up help

[Todd H.]

Howdy, 

I've been an emacs user for over 10 years and have recently been
forced into an upgrade from 20.6.1 under redhat to 21.x under Debian.
I access the Linux box in question via SSH (SecureCRT on a Win2k box
if it matters).  I use a VT220 emulation in my ssh client and have
TERM set to VT220 in the Linux shell.

On the old Emacs 20.6/Redhat box I was on, I set SecureCRT to
"backspace sends delete" and life was good.  Backspace deleted stuff
left of the cursor, and Control-h invoked help.

The problem now is that Under "GNU Emacs 21.3.1 (i386-pc-linux-gnu, X
toolkit, Xaw3d scroll bars) of 2004-10-16 on raven, modified by
Debian" pressing C-h now deletes left of the cursor, and the only way
to bring up help is M-x help.  I've fiddled with various SEcureCRT
settings to no avail.  I at least get backspace to work right if I let
it send backspace for backspace. 

Anyone have the magic bullet for this?  I've searched for an answer,
but most of what i find is people withe reverse problem of backspace
bringing up help.  The other odd thing is if I try to bind help to a
function key, I need to hit shift and that function key to make it
happen. 
        (global-set-key [f5] 'help)     

Did something change in Emacs 21 on "help" key binding? 

Best Regards, 
-- 
Todd H.
http://www.toddh.net/

Re: Emacs 21.3.1 -nw over ssh, C-h not bringing up help

[Peter Dyballa]

Am 19.03.2005 um 21:27 schrieb Todd H.:

> pressing C-h now deletes left of the cursor

This is what I have in my xterm in Mac OS X too. The cure is to put in 
your shell's rc file something like

	stty erase '^?'

You can check the shell's settings with 'stty -a' and correct the 
setting of erase from ^H to ^?, which is delete. (You too could consult 
'normal-erase-is-backspace' in you Emacs ...)

--
Greetings

   Pete

Re: Emacs 21.3.1 -nw over ssh, C-h not bringing up help

[Eli Zaretskii]

> From: comphelp@toddh.net (Todd H.)
> Date: 19 Mar 2005 14:27:31 -0600
> 
> The problem now is that Under "GNU Emacs 21.3.1 (i386-pc-linux-gnu, X
> toolkit, Xaw3d scroll bars) of 2004-10-16 on raven, modified by
> Debian" pressing C-h now deletes left of the cursor, and the only way
> to bring up help is M-x help.  I've fiddled with various SEcureCRT
> settings to no avail.  I at least get backspace to work right if I let
> it send backspace for backspace. 
> 
> Anyone have the magic bullet for this?

Perhaps "M-x normal-erase-is-backspace-mode RET" is it?

Re: Emacs 21.3.1 -nw over ssh, C-h not bringing up help

[Olive]

Todd H. wrote:
> Howdy, 
> 
> I've been an emacs user for over 10 years and have recently been
> forced into an upgrade from 20.6.1 under redhat to 21.x under Debian.
> I access the Linux box in question via SSH (SecureCRT on a Win2k box
> if it matters).  I use a VT220 emulation in my ssh client and have
> TERM set to VT220 in the Linux shell.
> 
> On the old Emacs 20.6/Redhat box I was on, I set SecureCRT to
> "backspace sends delete" and life was good.  Backspace deleted stuff
> left of the cursor, and Control-h invoked help.
> 
> The problem now is that Under "GNU Emacs 21.3.1 (i386-pc-linux-gnu, X
> toolkit, Xaw3d scroll bars) of 2004-10-16 on raven, modified by
> Debian" pressing C-h now deletes left of the cursor, and the only way
> to bring up help is M-x help.  I've fiddled with various SEcureCRT
> settings to no avail.  I at least get backspace to work right if I let
> it send backspace for backspace. 
> 
> Anyone have the magic bullet for this?  I've searched for an answer,
> but most of what i find is people withe reverse problem of backspace
> bringing up help.  The other odd thing is if I try to bind help to a
> function key, I need to hit shift and that function key to make it
> happen. 
>         (global-set-key [f5] 'help)     
> 
> Did something change in Emacs 21 on "help" key binding? 
> 
> Best Regards, 

I do not know if it answer your question; but the default key sequence 
send by xterm is the same for the backspace key and C-h so emacs cannot 
make the difference; it is xterm that must be configured. Emacs works 
well after having configuring xterm on the following way. Use the 
following resources:

XTerm*ttyModes: erase ^?
XTerm*backarrowKey: false
XTerm*eightBitInput: false

env. variable term set to "xterm"

Re: Emacs 21.3.1 -nw over ssh, C-h not bringing up help

[Peter Dyballa]

Am 19.03.2005 um 23:10 schrieb Olive:

> XTerm*ttyModes: erase ^?
> XTerm*backarrowKey: false

Of course these settings work -- but only here, on localhost. I think 
Todd has the problems when GNU Emacs runs in a shell in a remote site. 
So re-programming the local X server won't change his problems, he has 
to fix it in the remote shell.

> XTerm*eightBitInput: false

I wouldn't do that! The European scripts are 8 bit.

--
Greetings

   Pete

Re: Emacs 21.3.1 -nw over ssh, C-h not bringing up help

[David Kastrup]

Peter Dyballa <Peter_Dyballa@Web.DE> writes:

> Am 19.03.2005 um 23:10 schrieb Olive:
>
>> XTerm*ttyModes: erase ^?
>> XTerm*backarrowKey: false
>
> Of course these settings work -- but only here, on localhost. I
> think Todd has the problems when GNU Emacs runs in a shell in a
> remote site. So re-programming the local X server won't change his
> problems, he has to fix it in the remote shell.

X resources are managed by the server, not the client.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: Emacs 21.3.1 -nw over ssh, C-h not bringing up help

[Stefan Monnier]

>>> XTerm*ttyModes: erase ^?
>>> XTerm*backarrowKey: false
>> 
>> Of course these settings work -- but only here, on localhost. I
>> think Todd has the problems when GNU Emacs runs in a shell in a
>> remote site. So re-programming the local X server won't change his
>> problems, he has to fix it in the remote shell.

> X resources are managed by the server, not the client.

Actually, IIRC they're managed at both ends: the ~/.Xdefaults file is read
directly by the client (thus from the client's file system) whereas the xrdb
database (traditionally filled at login time from ~/.Xresources) is on
the server.


        Stefan

Re: Emacs 21.3.1 -nw over ssh, C-h not bringing up help

[Olive]

Peter Dyballa wrote:
> 
> Am 19.03.2005 um 23:10 schrieb Olive:
> 
>> XTerm*ttyModes: erase ^?
>> XTerm*backarrowKey: false
> 
> 
> Of course these settings work -- but only here, on localhost. I think 
> Todd has the problems when GNU Emacs runs in a shell in a remote site. 
> So re-programming the local X server won't change his problems, he has 
> to fix it in the remote shell.
> 
>> XTerm*eightBitInput: false
> 
> 
> I wouldn't do that! The European scripts are 8 bit.
> 

This setting does not prevent xterm to handle 8 bit characters properly 
(I usually wrote French, ...). It tell emacs to send "meta characters" 
as control sequence (see man xterm). To handle 8 bit properly; you must 
also add the following in your .emacs (for language that use Latin1 
encoding).

(set-terminal-coding-system 'latin-1)
(set-keyboard-coding-system 'latin-1)
(xterm-mouse-mode 1)

Olive

Re: Emacs 21.3.1 -nw over ssh, C-h not bringing up help

[Olive]

David Kastrup wrote:
> Peter Dyballa <Peter_Dyballa@Web.DE> writes:
> 
> 
>>Am 19.03.2005 um 23:10 schrieb Olive:
>>
>>
>>>XTerm*ttyModes: erase ^?
>>>XTerm*backarrowKey: false
>>
>>Of course these settings work -- but only here, on localhost. I
>>think Todd has the problems when GNU Emacs runs in a shell in a
>>remote site. So re-programming the local X server won't change his
>>problems, he has to fix it in the remote shell.

This configure xterm; not the shell. Since xterm is run locally; this 
must be configured in the local X server.

Olive

Re: Emacs 21.3.1 -nw over ssh, C-h not bringing up help

[Todd H.]

"Eli Zaretskii" <eliz@gnu.org> writes:

> > From: comphelp@toddh.net (Todd H.)
> > Date: 19 Mar 2005 14:27:31 -0600
> > 
> > The problem now is that Under "GNU Emacs 21.3.1 (i386-pc-linux-gnu, X
> > toolkit, Xaw3d scroll bars) of 2004-10-16 on raven, modified by
> > Debian" pressing C-h now deletes left of the cursor, and the only way
> > to bring up help is M-x help.  I've fiddled with various SEcureCRT
> > settings to no avail.  I at least get backspace to work right if I let
> > it send backspace for backspace. 
> > 
> > Anyone have the magic bullet for this?
> 
> Perhaps "M-x normal-erase-is-backspace-mode RET" is it?

Woohoo!  We have a winner!    

Doing that, along with setting SecureCRT back to the way I had it with
"backspace sends delete"  gets me where I want to go.  

As others mentioned, the mentions of X resources aren't germane to
remote ssh use in terminal mode. 

Thanks all who replied though, and to Eli for the ultimate answer.
Perhaps the Debian elisp customizations that load on this version of
emacs here unset this while they were defaults on RedHat.

Best Regards, 
-- 
Todd H.
http://www.toddh.net/

Re: Emacs 21.3.1 -nw over ssh, C-h not bringing up help

[Todd H.]

Peter Dyballa <Peter_Dyballa@Web.DE> writes:
> Am 19.03.2005 um 21:27 schrieb Todd H.:
> 
> > pressing C-h now deletes left of the cursor
> 
> This is what I have in my xterm in Mac OS X too. The cure is to put in
> your shell's rc file something like
> 
> 	stty erase '^?'
> 
> You can check the shell's settings with 'stty -a' and correct the
> setting of erase from ^H to ^?, which is delete. (You too could
> consult 'normal-erase-is-backspace' in you Emacs ...)

Hi Pete, 

Thanks for the response.  I had tried mucking with the shell settings
for erase and it didn't seem to help unfortunately.  Seems there's
something new and odd in either Debian or Emacs 21.x that was
thwarting my efforts.

However, dropping 
        (normal-erase-is-backspace-mode 0)

into .emacs  (or running that function interactively with M-x
normal-erase-is-backspace  was the magic bullet for me (along with
setting my SecureCRT profile mapped keys for "backspace sends delete."
With those two steps, things are back to the way I was used to under
the RedHat box and Emacs 20.6.1.    

Best Regards,  
-- 
Todd H.
http://www.toddh.net/