Re: Time for Tramp 2.0?
> Before putting this out, a call for opinions on gnu-prog-disc and > emacs-pretesters would be appropriate. Hrm. Is tramp officially part of the GNU Emacs release at this point? No, it is not. Sorry, I was wrong.
check in by grossjoh: 'tramp/lisp tramp.el,1.454 ChangeLog,1.50'
Update of /services/emacs-rcp/cvsroot/tramp/lisp In directory lucy:/export/home/grossjoh/work/emacs-tramp/tramp/lisp Modified Files: tramp.el ChangeLog Log Message:
check in by grossjoh: ' .cvsignore,2.0'
Update of /services/emacs-rcp/cvsroot In directory lucy:/export/home/grossjoh/work/emacs-tramp Modified Files: Tag: 2.0 .cvsignore Log Message: Start a new release
check in by grossjoh: 'CVSROOT checkoutlist,2.0 commitinfo,2.0 config,2.0 cvswrappers,2.0 editinfo,2.0 loginfo,2.0 modules,2.0 notify,2.0 rcsinfo,2.0 readers,2.0 taginfo,2.0 verifymsg,2.0'
Update of /services/emacs-rcp/cvsroot/CVSROOT In directory lucy:/export/home/grossjoh/work/emacs-tramp/CVSROOT Modified Files: Tag: 2.0 checkoutlist commitinfo config cvswrappers editinfo loginfo modules notify rcsinfo readers taginfo verifymsg Log Message: Start a new release
check in by grossjoh: 'tramp .cvsignore,2.0 CONTRIBUTORS,2.0 Makefile,2.0'
Update of /services/emacs-rcp/cvsroot/tramp In directory lucy:/export/home/grossjoh/tmp/junk/xx/tramp Modified Files: Tag: 2.0 .cvsignore CONTRIBUTORS Makefile Log Message: Going stable?
check in by grossjoh: 'tramp/texi .cvsignore,2.0 ChangeLog,2.0 Makefile,2.0 tramp.texi,2.0 tramp_ja.texi,2.0'
Update of /services/emacs-rcp/cvsroot/tramp/texi In directory lucy:/export/home/grossjoh/tmp/junk/xx/tramp/texi Modified Files: Tag: 2.0 .cvsignore ChangeLog Makefile tramp.texi tramp_ja.texi Log Message: Going stable?
check in by grossjoh: 'tramp/lisp ChangeLog,2.0 Makefile,2.0 base64.el,2.0 format-spec.el,2.0 tramp-vc.el,2.0 tramp.el,2.0 trampcache.el,2.0'
Update of /services/emacs-rcp/cvsroot/tramp/lisp In directory lucy:/export/home/grossjoh/tmp/junk/xx/tramp/lisp Modified Files: Tag: 2.0 ChangeLog Makefile base64.el format-spec.el tramp-vc.el tramp.el trampcache.el Log Message: Going stable?
Re: Time for Tramp 2.0?
Kai> Maybe `::' could be the end of the connect list? Maybe nobody uses VMS and DECNET anymore, but isn't "::" a separator between host and path components in a DECNET address? (It's been a *long* while since I've seen a VMS system...) It's also a separator for Mac file names.
Re: Time for Tramp 2.0?
On 28 Feb 2001, Daniel Pittman wrote: > Well... I bow to Kai on this. My personal feeling is that having a > branch for the new development and retaining the older tramp for > bug-fixing 'til the new one works is good. I prefer to have the development be on the head. I _think_ I have now created a revision `rel-2-0' which checks out the current status, and a branch `stable' which is currently equal to rel-2-0 but can be used to put bug fixes in. The head will get new features, and I think I'll put the stable branch on the ftp server every now and then. So. Daniel, as long as there isn't an closing parenthesis missing or something obvious like this, feel free to commit to the head. kai -- Be indiscrete. Do it continuously.
check in by daniel: 'tramp/test - New directory'
Update of /services/emacs-rcp/cvsroot/tramp/test In directory bonny:/tmp/cvs-serv14614/test Log Message: Directory /services/emacs-rcp/cvsroot/tramp/test added to the repository
check in by daniel: 'tramp/lisp tramp2-ops.el,2.1 tramp2.el,2.1 tramp.el,2.1'
Update of /services/emacs-rcp/cvsroot/tramp/lisp In directory bonny:/tmp/cvs-serv14859/lisp Modified Files: tramp.el Added Files: tramp2-ops.el tramp2.el Log Message: Initial checkin of the *NOT WORKING* tramp2 shell. Even if you just updated to this version, though, tramp still works because the old code isn't gone... yet.
check in by daniel: 'tramp/test test-harness.el,1.1 tramp-expr.el,1.1 tramp-path.el,1.1'
Update of /services/emacs-rcp/cvsroot/tramp/test In directory bonny:/tmp/cvs-serv14859/test Added Files: test-harness.el tramp-expr.el tramp-path.el Log Message: Initial checkin of the *NOT WORKING* tramp2 shell. Even if you just updated to this version, though, tramp still works because the old code isn't gone... yet.
Re: Time for Tramp 2.0?
On Wed, 28 Feb 2001, Francesco Potorti` wrote: >Kai> Maybe `::' could be the end of the connect list? > >Maybe nobody uses VMS and DECNET anymore, but isn't "::" a >separator between host and path components in a DECNET address? >(It's been a *long* while since I've seen a VMS system...) > > It's also a separator for Mac file names. The current path parser assumes that the *first* '::' in a TRAMP2 path (and only in a TRAMP2 path) signifies that this is the end of the connection set. This should allow DECNET and Mac remote hosts to work fine, assuming that their login shell uses that style of path (and the Unix tools work with them. :) Unless the '::' is actually significant in a host name or a user name, I don't think there will be problems. So, even the DECNET ones seem safe from what I have been told. Is this so? Daniel -- Women who seek to be equal with men lack ambition. -- Timothy Leary
Re: Time for Tramp 2.0?
On 28 Feb 2001, Kai Großjohann wrote: > On 28 Feb 2001, Daniel Pittman wrote: > >> Well... I bow to Kai on this. My personal feeling is that having a >> branch for the new development and retaining the older tramp for >> bug-fixing 'til the new one works is good. > > I prefer to have the development be on the head. > > I _think_ I have now created a revision `rel-2-0' which checks out the > current status, and a branch `stable' which is currently equal to > rel-2-0 but can be used to put bug fixes in. > > The head will get new features, and I think I'll put the stable branch > on the ftp server every now and then. > > So. Daniel, as long as there isn't an closing parenthesis missing or > something obvious like this, feel free to commit to the head. Right. The development work has gone in. It does not /do/ anything yet and, well, it's kind of in a state of flux. I think that there are a couple of logic errors in my design, because it seems a touch klunky in places. There may be changes in the remote command stuff in the not too distant future. Anyway, the good stuff is in 'tramp2.el' and 'tramp2-ops.el'. They are the current shell of the code. If you want to look at `def-tramp-handler' and the path/connect infrastructure, that would be most useful. I intend to modify the way `def-tramp-handler' works in the near future, though, because the magic worked on the `file' argument should be more generic... It's worth looking in test/ as well - some automated tests for the path parser and the like. Those have helped quite a bit already in the development cycle. :) Daniel -- No matter how hard you push and no matter what the priority, you can't increase the speed of light. -- RFC 1925
Re: Time for Tramp 2.0?
On 28 Feb 2001, Daniel Pittman wrote: > Unless the '::' is actually significant in a host name or a user > name, I don't think there will be problems. Right. So, is there a host name or a user name in any kind of naming scheme which contains "::"? And this is a variable anyway, so we can still change the "::" part later on. kai -- Be indiscrete. Do it continuously.
Putty compatibility
The right command for putty is plink -ssh username@hostname But the current tramp-open-connection-rsh code wants to do plink -ssh hostname -l username Clearly, this is bad. So my idea is to change tramp-methods so that the rsh-args list contains "%u" and %h" escapes for the user name and the host name, like done in other places inside Tramp. Daniel, do you think it makes sense to add this before your Great Connection Method Rewrite, or should we wait? Tom, you can make a copy of the tramp-open-connection-rsh function and change the place where the rsh-program is invoked. I think you can see where it adds "hostname -l username", and you can just replace that part. Give the new function a different name, and then add an entry to tramp-methods which refers to the new function. This would be a workaround until Daniel is finshed, if Daniel thinks it's not worthwhile changing the old version. kai -- Be indiscrete. Do it continuously.
doc bug in tramp-open-connection-rsh?
The docstring for tramp-open-connection-rsh for tramp-1.449 says (in part) > "Open a connection using an rsh METHOD. This starts the command > `rsh HOST -l USER'[*], then waits for a remote password or shell > prompt. If a password prompt is seen, the user is queried for a > password, this function sends the password to the remote host and > waits for a shell prompt. > If USER is nil, start the command `rsh HOST'[*] instead > Recognition of the remote shell prompt is based on the variable > `shell-prompt-pattern' which must be set up correctly. > Please note that it is NOT possible to use this connection method > with an inline transfer method if this function asks the user for a > password! You must use an inline transfer method in this case. Am I missing something, or is that last paragraph in error? FWIW, [EMAIL PROTECTED]
check in by grossjoh: 'tramp/lisp tramp.el,2.2 ChangeLog,2.1'
Update of /services/emacs-rcp/cvsroot/tramp/lisp In directory lucy:/export/home/grossjoh/work/kai/tramp-devel/tramp/lisp Modified Files: tramp.el ChangeLog Log Message: (tramp-open-connection-rsh): Doc bug. Reported by Tom Roche.
check in by grossjoh: 'tramp/lisp tramp.el,2.0.2.1 ChangeLog,2.0.2.1'
Update of /services/emacs-rcp/cvsroot/tramp/lisp In directory lucy:/export/home/grossjoh/work/kai/tramp-stable/tramp/lisp Modified Files: Tag: stable tramp.el ChangeLog Log Message: (tramp-open-connection-rsh): Doc bug. Reported by Tom Roche.
Re: Putty compatibility
[EMAIL PROTECTED] 28 Feb 2001 15:18:41 +0100
> The right command for putty is
> plink -ssh username@hostname
> But the current tramp-open-connection-rsh code wants to do
> plink -ssh hostname -l username
> Clearly, this is bad.
At least, it's not general. (To assume standards-compliance can be
a bit naive, esp in the M$ world :-)
> So my idea is to change tramp-methods so that the rsh-args list
> contains "%u" and %h" escapes for the user name and the host name,
> like done in other places inside Tramp.
> you can make a copy of the tramp-open-connection-rsh function and
> change the place where the rsh-program is invoked. I think you can
> see where it adds "hostname -l username", and you can just replace
> that part.
Perhaps :-) I did (whitespace hacked for mail)
HACKED-tramp-open-connection-rsh
> (let ((process-environment (copy-sequence process-environment)))
> (setenv "TERM" tramp-terminal-type)
> (let* ((default-directory (tramp-temporary-file-directory))
>(coding-system-for-read (unless
> (and (not (featurep 'xemacs))
>(> emacs-major-version 20))
> tramp-dos-coding-system))
>(p (if user
> (apply #'start-process
> (tramp-buffer-name multi-method method user host)
> (tramp-get-buffer multi-method method user host)
> (tramp-get-rsh-program multi-method method)
> ; TLR: hacked for PuTTY compatibility 28 Feb 01
> ;host "-l" user
> user "@" host
> (tramp-get-rsh-args multi-method method))
> (apply #'start-process
> (tramp-buffer-name multi-method method user host)
> (tramp-get-buffer multi-method method user host)
> (tramp-get-rsh-program multi-method method)
> host
> (tramp-get-rsh-args multi-method method
>(found nil))
I also changed
_emacs
> (setq tramp-methods
> (append
> (list
> '("t"
> ; (tramp-connection-function tramp-open-connection-rsh)
> (tramp-connection-function HACKED-tramp-open-connection-rsh)
> (tramp-rsh-program "plink")
> (tramp-rcp-program "pscp")
> (tramp-remote-sh"/bin/sh")
> (tramp-rsh-args
> ("-v" "-ssh"))
> (tramp-rcp-keep-date-arg"-p")
> (tramp-su-program nil)
> (tramp-su-args nil)
> (tramp-encoding-command nil)
> (tramp-decoding-command nil)
> (tramp-encoding-functionnil)
> (tramp-decoding-functionnil)
> (tramp-telnet-program nil))
> )
> tramp-methods)
> ) ; end (setq tramp-methods
However, M-x find-file /r@t:[EMAIL PROTECTED]:.emacs gets
*tramp/t [EMAIL PROTECTED]*
> Unable to open connection:
> gethostbyname: unknown error
> Process *tramp/t [EMAIL PROTECTED]* exited abnormally with code 1
(the generic error message for all the PuTTY clients) with
*Backtrace*
> Signaling: (error "Couldn't find remote shell or passwd prompt")
> signal(error ("Couldn't find remote shell or passwd prompt"))
> error("Couldn't find remote shell or passwd prompt")
> HACKED-tramp-open-connection-rsh(nil "t" "tlroche" "ssh.ncsu.edu")
> tramp-maybe-open-connection(nil "t" "tlroche" "ssh.ncsu.edu")
> tramp-send-command(nil "t" "tlroche" "ssh.ncsu.edu" "cd ~; pwd")
> tramp-handle-expand-file-name(
> "/r@t:[EMAIL PROTECTED]:.emacs" nil)
> apply(tramp-handle-expand-file-name
> ("/r@t:[EMAIL PROTECTED]:.emacs" nil))
> tramp-file-name-handler(expand-file-name
> "/r@t:[EMAIL PROTECTED]:.emacs" nil)
> expand-file-name("/r@t:[EMAIL PROTECTED]:.emacs")
> find-file-noselect("/r@t:[EMAIL PROTECTED]:.emacs" nil nil 1)
> find-file("/r@t:[EMAIL PROTECTED]:.emacs" 1)
> call-interactively(find-file)
> execute-extended-command(nil)
> call-interactively(execute-extended-command)
However, in cmd (though not in a *shell* with cmdproxy--more later) I
can do
> D:\temp>plink -v -ssh [EMAIL PROTECTED]
> Server version: SSH-1.5-1.2.27
> We claim version: SSH-1.5-PuTTY
> Using SSH protocol version 1
> Received public keys
> Host key fingerprint is:
> 1024 65:5d:00:07:9c:87:55:89:99:58:e0:71:33:38:1d:59
> Encrypted session key
> Using 3DES encryption
> Trying to enable encryption...
> Successfully started encryption
> Sent username "tlroche"
> [EMAIL PROTECTED]'s password:
> Sent password
> Remote: Kerberos TGT could not be verified.
> Authentication successful
> Allocated pty
> Started session
> You have 151 messages (1030775 bytes) on uni00map.unity.ncsu.edu
> Volume Name Quota Used %Used Partition
> users.t.tlroche 5 33913 68% 58%
> ?7?[r?[999;999H?[6nresize: unknown character, exiting.
> Running user customized standard startup activit
Re: [putty]Re: putty, tramp, NtEmacs...
[EMAIL PROTECTED] 27 Feb 2001 09:41:08 +0100 >>> But these days, it seems that most people are using putty. So it >>> would be very useful if you could make it work with putty. >>> Open a DOS window, then type a command which will open an ssh >>> connection to the Unix host (without opening a new window!). [EMAIL PROTECTED] Tue, 27 Feb 2001 15:42:41 -0500 >> For a purely cmdline interface (only tested with w2k cmd.exe), one >> can use 'plink'. I haven't used it much: I only downloaded it for >> the purposes of this discussion, but I will try to use it more. For >> its cmdline help, just run 'plink' with no options. (Why can't >> 'putty' do that?). More usefully, one can do >> plink -ssh [EMAIL PROTECTED] >> and get an SSH session in the current window. Owen Dunn <[EMAIL PROTECTED]> Wed, 28 Feb 2001 09:28:06 > As the documentation says: > | Plink (PuTTY Link), is a command-line connection tool similar to > | UNIX `ssh'. It is probably not what you want if you want to run an > | interactive session in a console window. > (It's mostly intended for use like Unix ssh with a remote command, > which makes it ideal for use as a transport for e.g. Windows CVS.) Actually, I believe this _is_ my intended use. The above was mentioned purely for exposition; my application domain is Emacs TRAMP http://ls6-www.informatik.uni-dortmund.de/~grossjoh/emacs/tramp.html http://www.mail-archive.com/emacs-rcp@ls6.cs.uni-dortmund.de/ >> In short: the Putty tools allow me to connect to the servers I >> need, but only as long as I type a password. I have never been able >> to get anything working with the Kerberos servers with which I must >> interact that didn't require typing a password. If you have any >> suggestions regarding this, please lemme know. > PuTTY does not support Kerberos. Nevertheless, putty, plink, and pscp all work !-) from cmd, as long as I give them passwords. Any suggestions for making them work inside Emacs and TRAMP would be appreciated. TIA, [EMAIL PROTECTED]
debugging output, was: Putty compatibility
On Wed, 28 Feb 2001, Tom Roche wrote:
>> * "feature" suggestion: I have
>> (setq rcp-verbose 10)
>> (setq rcp-debug-buffer t)
>> With verbose debugging thus enabled, IMHO it would be nice to
>> have tramp print to the buffer exactly what command it's
>> uttering.
[EMAIL PROTECTED] 28 Feb 2001 21:17:38 +0100
> With `tramp' rather than `rcp',
Doh!
> this should work. No?
No: output is still (broken for mail)
*Backtrace*
> Signaling: (error "Couldn't find remote shell or passwd prompt")
> signal(error ("Couldn't find remote shell or passwd prompt"))
> error("Couldn't find remote shell or passwd prompt")
> HACKED-tramp-open-connection-rsh(nil "t" "tlroche" "ssh.ncsu.edu")
> tramp-maybe-open-connection(nil "t" "tlroche" "ssh.ncsu.edu")
> tramp-send-command(nil "t" "tlroche" "ssh.ncsu.edu" "cd ~; pwd")
> tramp-handle-expand-file-name("/r@t:[EMAIL PROTECTED]:.emacs"
> nil)
> apply(tramp-handle-expand-file-name
> ("/r@t:[EMAIL PROTECTED]:.emacs" nil))
> tramp-file-name-handler(expand-file-name
> "/r@t:[EMAIL PROTECTED]:.emacs" nil)
> expand-file-name("/r@t:[EMAIL PROTECTED]:.emacs")
> find-file-noselect("/r@t:[EMAIL PROTECTED]:.emacs" nil nil 1)
> find-file("/r@t:[EMAIL PROTECTED]:.emacs" 1)
> call-interactively(find-file)
> execute-extended-command(nil)
> call-interactively(execute-extended-command)
I'm sure the command is implicit to you all, but I'd still like to see
it explicitly :-)
FWIW, [EMAIL PROTECTED]
Re: Putty compatibility
On 28 Feb 2001, Kai Großjohann wrote: > The right command for putty is > > plink -ssh username@hostname > > But the current tramp-open-connection-rsh code wants to do > > plink -ssh hostname -l username > > Clearly, this is bad. So my idea is to change tramp-methods so that > the rsh-args list contains "%u" and %h" escapes for the user name and > the host name, like done in other places inside Tramp. > > Daniel, > > do you think it makes sense to add this before your Great Connection > Method Rewrite, or should we wait? I think that bug-fixing on the current release should go ahead for a while. Wait 'til my code actually establishes a connection and runs a command on the remote host first. :) Daniel -- Every old idea will be proposed again with a different name and a different presentation, regardless of whether it works. -- RFC 1925
Re: Putty compatibility
On Wed, 28 Feb 2001, Tom Roche wrote: > * "feature" suggestion: I have > > (setq rcp-verbose 10) > (setq rcp-debug-buffer t) > > With verbose debugging thus enabled, IMHO it would be nice to have > tramp print to the buffer exactly what command it's uttering. With `tramp' rather than `rcp', this should work. No? kai -- Be indiscrete. Do it continuously.
parameterize wait time?
Dunno if this is already implemented, but it's apparently not done in tramp-1.449 (I see a lotta 60s in tramp.el): IMHO it would be good to parameterize the wait time (e.g. setq tramp-wait-seconds), so as to expose that to the user. I suppose if one's testing is going well, this is not a problem, but I get annoyed waiting a minute for my connection to fail :-( FWIW, [EMAIL PROTECTED]
Re: check in by grossjoh: 'tramp/lisp tramp.el,2.0.2.1 ChangeLog,2.0.2.1'
[Kai Grossjohann] > Update of /services/emacs-rcp/cvsroot/tramp/lisp > In directory lucy:/export/home/grossjoh/work/kai/tramp-stable/tramp/lisp > > Modified Files: >Tag: stable > tramp.el ChangeLog being a chicken, I guess I prefer staying with the stable version of Tramp. since I know very little about cvs, I've let Emacs/pcl-cvs handle updating from the CVS repository. does pcl-cvs have any command for updating to a specific version, or a specific branch (and is the 2.0.x.x branch the one I should pursue?) (a plain, regular cvs command would also work great :) -- Henrik Q: Why did the chicken enter the Klein Bottle?
Re: parameterize wait time?
On Wed, 28 Feb 2001, Tom Roche wrote: > Dunno if this is already implemented, but it's apparently not done in > tramp-1.449 (I see a lotta 60s in tramp.el): > > IMHO it would be good to parameterize the wait time (e.g. setq > tramp-wait-seconds), so as to expose that to the user. I suppose if > one's testing is going well, this is not a problem, but I get annoyed > waiting a minute for my connection to fail :-( If you can wait for the new code to be finished, you get the feature. Until then, if you make sure your patch to do that is up to date (and Kai does not object), I think it could be applied... Daniel -- When we cannot bear to be alone, it means we do not properly value the only companion we will have from birth to death-ourselves. -- Eda LeShan
tramp with 'plink -pw'
Bill Pringlemeir <[EMAIL PROTECTED]> 28 Feb 2001 13:43:25
>> However, you can use the "-pw secret" option to allow plink to
>> connect. This worked on my NTEmacs 20.7.3.
[EMAIL PROTECTED] Wed, 28 Feb 2001 16:30:10 -0500
> And for me on GNU Emacs 20.7.1 (i386-*-nt5.0.2195).
I.e. in GNU Emacs 20.7.1 on win2k with (I believe) cmdproxy, I can do
M-x shell and enter
> plink -ssh -v -pw password [EMAIL PROTECTED]
and connect, and interact. However, with tramp-1.449 and
> (setq tramp-methods
> (append
> (list
> '("t"
> ; (tramp-connection-function tramp-open-connection-rsh)
> (tramp-connection-function HACKED-tramp-open-connection-rsh)
> (tramp-rsh-program "plink")
> (tramp-rcp-program "pscp")
> (tramp-remote-sh"/bin/sh")
> (tramp-rsh-args
> ("-v"
>"-pw password"
>"-ssh"))
> (tramp-rcp-keep-date-arg"-p")
> (tramp-su-program nil)
> (tramp-su-args nil)
> (tramp-encoding-command nil)
> (tramp-decoding-command nil)
> (tramp-encoding-functionnil)
> (tramp-decoding-functionnil)
> (tramp-telnet-program nil))
> )
> tramp-methods)
> ) ; end (setq tramp-methods
in my _emacs, when I do
M-x find-file /r@t:[EMAIL PROTECTED]:.emacs
it still fails to connect, with the same *tramp* and *Backtrace*
buffer contents. Any suggestions on how to fix this?
One thing I'm wondering: since the error is
*Backtrace*
> Signaling: (error "Couldn't find remote shell or passwd prompt")
is there a problem with my shell-prompt-pattern? It's currently set to
> (setq shell-prompt-pattern "^[^#$%>\n]*[#$>] *")
The servers to which I'm trying to connect default to tcsh, and return
stuff like (copied from *shell*, broken for mail at '..')
> +-- .. -+
> | By accessing this machine, it .. med that you have read and agree |
> | with the dialup usage policy .. |
> | http://www.ncsu.edu/polic .. .html [Last Updated 11/10/00] |
> | If you have not read the abov .. se do so immediately.|
> |- SIGNIFICANT CHANGES IN POLIC .. WEB BROWSERS RUN FROM DIALUPS |
> | HAVE BEEN MADE. ..--|
> +-- .. -+
> | For the latest Eos/Unity news .. on reporting system problems and |
> | getting help with the system, .. s'. For the Eos/Unity usage |
> | policy, type 'policy'... |
> +-- .. -+
> | Be sure to change your passwo .. ce a semester. Choose a password |
> | with a mix of upper and lower .. s, numbers, or other special |
> | characters. Password changes .. effect for up to 30 minutes |
> | after requesting the change. .. |
> +-- .. -+
> You have 109 messages (590062 bytes) on uni00map.unity.ncsu.edu
> Volume Name Quota Used %Used Partition
> users.t.tlroche 5 33939 68% 58%
> 7[r[999;999H[6nresize: Time out occurred
Note that this times out after about 10 seconds.
> Running user customized standard startup activities ...
> End of login--close console.
> [1m/ncsu/tlroche> [m
Also: could someone tell me where/with what variable the shell program
is set? I don't see anything in apropos. I'm assuming mine is
cmdproxy, because when I do M-x shell, I see
> Microsoft Windows 2000 [Version 5.00.2195]
> (C) Copyright 1985-2000 Microsoft Corp.
> h:\>
TIA, [EMAIL PROTECTED]
Re: parameterize wait time?
On Wed, 28 Feb 2001, Tom Roche wrote: >> IMHO it would be good to parameterize the wait time (e.g. setq >> tramp-wait-seconds), so as to expose that to the user. I suppose if >> one's testing is going well, this is not a problem, but I get >> annoyed waiting a minute for my connection to fail :-( Daniel Pittman <[EMAIL PROTECTED]> 01 Mar 2001 13:30:49 +1100 > If you can wait for the new code to be finished, you get the > feature. No rush: my life has bigger problems :-) > When we cannot bear to be alone, it means we do not properly value > the only companion we will have from birth to death-ourselves. > -- Eda LeShan Not a bad .sig quote, either. TIA, [EMAIL PROTECTED]
Re: tramp with 'plink -pw'
On Wed, 28 Feb 2001, Tom Roche wrote: > Bill Pringlemeir <[EMAIL PROTECTED]> 28 Feb 2001 13:43:25 >>> However, you can use the "-pw secret" option to allow plink to >>> connect. This worked on my NTEmacs 20.7.3. > > [EMAIL PROTECTED] Wed, 28 Feb 2001 16:30:10 -0500 >> And for me on GNU Emacs 20.7.1 (i386-*-nt5.0.2195). > > I.e. in GNU Emacs 20.7.1 on win2k with (I believe) cmdproxy, I can do > M-x shell and enter > >> plink -ssh -v -pw password [EMAIL PROTECTED] > > and connect, and interact. However, with tramp-1.449 and > [...] > > in my _emacs, when I do > > M-x find-file /r@t:[EMAIL PROTECTED]:.emacs > > it still fails to connect, with the same *tramp* and *Backtrace* > buffer contents. Any suggestions on how to fix this? Hack your profile on the remote machine to *not* spit out ANSI/VT100 when TERM is DUMB. :) > One thing I'm wondering: since the error is > > *Backtrace* >> Signaling: (error "Couldn't find remote shell or passwd prompt") > > is there a problem with my shell-prompt-pattern? It's currently set to > >> (setq shell-prompt-pattern "^[^#$%>\n]*[#$>] *") Yeah, that fails to match: [...] >> Running user customized standard startup activities ... >> End of login--close console. >> [1m/ncsu/tlroche> [m Because of the trailing escape sequence "[m". The right fix, as I said above, is to have the silly tcsh on the remote machine *not* automatically try to underline your command prompt. Probably by setting PS1 in the login script. The wrong fix, but one that makes life for you easier, is to hack the regexp in `shell-prompt-pattern' to match the escape sequences as well: (setq shell-prompt-pattern "^\\(\e[1m\\)?[^#$%>\n]*[#$>] *\\(\e[m\\)?") You can even make the escape sequence matching more generic if you want. :) Daniel -- A large number of installed systems work by fiat. That is, they work by being declared to work. -- Anatol Holt
cygwin and nt-emacs
A while back I said I was going to try to get tramp working on ntemacs. I would like to use the ssh that comes with the cygwin toolset, though, instead of putty. I spent a little bit on it, but I'd like to get it done. The problem is that ssh under ntemacs complains that it doesn't have a tty even if I give it the '-t'. I think I have comint configured correctly. I can use 'M-x shell' and (I think) 'M-x ansi-term' with cygwin's bash, but I still get the tty problem for ssh. Would any of you have a clue on this? Mark. -- So what are you going to do with more money? Eat six meals a day? -- http://glennf.weblogs.com/2001/02/10
Re: tramp with 'plink -pw'
On Wed, 28 Feb 2001, Tom Roche wrote: >> One thing I'm wondering: since the error is >> *Backtrace* >>> Signaling: (error "Couldn't find remote shell or passwd prompt") >> is there a problem with my shell-prompt-pattern? It's currently set >> to >> (setq shell-prompt-pattern "^[^#$%>\n]*[#$>] *") Daniel Pittman <[EMAIL PROTECTED]> 01 Mar 2001 13:48:16 +1100 > Yeah, that fails to match: >>> Running user customized standard startup activities ... >>> End of login--close console. >>> [1m/ncsu/tlroche> [m > Because of the trailing escape sequence "[m". > The right fix, as I said above, is to have the silly tcsh on the remote > machine *not* automatically try to underline your command prompt. > Probably by setting PS1 in the login script. OK, I'll get back to you about that. Meanwhile, something easier is to take the 'set prompt' out of .mycshrc, which changes the last line to > unity% which might be easier to match. I tried changing to (setq shell-prompt-pattern "^.*unity%\s-*") which still doesn't work. Any suggestions? TIA, [EMAIL PROTECTED]
Re: tramp with 'plink -pw'
On Wed, 28 Feb 2001, Tom Roche wrote: > On Wed, 28 Feb 2001, Tom Roche wrote: >>> One thing I'm wondering: since the error is > >>> *Backtrace* Signaling: (error "Couldn't find remote shell or passwd prompt") > >>> is there a problem with my shell-prompt-pattern? It's currently set >>> to > >>> (setq shell-prompt-pattern "^[^#$%>\n]*[#$>] *") > > Daniel Pittman <[EMAIL PROTECTED]> 01 Mar 2001 13:48:16 +1100 >> Yeah, that fails to match: > Running user customized standard startup activities ... End of login--close console. [1m/ncsu/tlroche> [m > >> Because of the trailing escape sequence "[m". > >> The right fix, as I said above, is to have the silly tcsh on the >> remote machine *not* automatically try to underline your command >> prompt. Probably by setting PS1 in the login script. > > OK, I'll get back to you about that. Meanwhile, something easier is > to take the 'set prompt' out of .mycshrc, which changes the last > line to That was your work? Oh. Our default tcsh here does it for free so I assumed... :) >> unity% > > which might be easier to match. I tried changing to > > (setq shell-prompt-pattern "^.*unity%\s-*") You want "^.*unity%", which should work as well. I don't think that you need to match punctuation characters at the end at all, really... Oh, and you wanted "\\s-*" - the double-backslash is essential. :) The default `shell-prompt-pattern' should work fine, though, for that prompt. Can you send a full trace thingy (elide the banner if you wish) with the new shell prompt? Daniel -- Open the tombs and see the bones there mixed in mockery! Which dust was servant, and which was lord's?--open the tomb and see! -- Shaikh Saadi, _Gulistan_, (c. 1265)
Re: doc bug in tramp-open-connection-rsh?
On Wed, 28 Feb 2001, Tom Roche wrote: > The docstring for tramp-open-connection-rsh for tramp-1.449 says > (in part) > >> Please note that it is NOT possible to use this connection method >> with an inline transfer method if this function asks the user for a >> password! You must use an inline transfer method in this case. > > Am I missing something, or is that last paragraph in error? The first `inline' should be `out-of-band'. Thanks. Fixed in head and stable. kai -- Be indiscrete. Do it continuously.
