Re: Tramp and NT-Emacs with plink

[Kai Großjohann]

Douglas Gray Stephens <address@hidden> writes:

> Sorry, more things to look at ....
>
> I'm currently on tramp.el version 2.0.23.
>
> I still believe that tramp is filling my ~/.bash_history file up, but
> am not sure how to help debug this issue, as it probably affects other
> users. (It appears to only happen on some machines, so there must be a
> machine/shell/??? issue, but I'm using bash as my default shell on all
> the machines.)

Tramp tries to turn off the ksh history already, it is easy to add
commands to also turn off the bash history.

How does one turn off the bash history?  (I'm too lazy to read the
documentation...)

> I have also started started to look at the version control ability,
> and so stretch plink a bit further.
>
> I can check a file into/out of SCCS,

Good.  I'm surprised.  Nobody else has used Tramp with SCCS, as far
as I know.

> however some of the other commands are failing vc-print-log
> vc-version-other-window
>
> It appears that the vc-print-log details are in the debug buffer
>  $ prs /export/home/gray/work/dumps/SCCS/s.ldap-extract-summaries.pl; 
> tramp_old_status=$?
>  /export/home/gray/work/dumps/SCCS/s.ldap-extract-summaries.pl:
>
>  D 1.2 02/10/01 08:53:35 gray 2 1        00059/00029/00197
>  MRs:
>  COMMENTS:
>  Update messages in log (stop using Date::Manip).
>
>
>  D 1.1 02/09/29 21:08:10 gray 1 0        00226/00000/00000
>  MRs:
>  COMMENTS:
>  date and time created 02/09/29 21:08:10 by gray
>
>  $ cd
> but the *vc* buffer comes up with
>  'prs' is not recognized as an internal or external command,
>  operable program or batch file.

Hm.  This is strange.  Is it running prs twice?  Once on the remote
host and once on the local host?

It's not clear to me how to change this.

> while vc-version-other 
>
> For vc-version-other-window, again the file is in the debug buffer
> (with no base64 encoding) following the command
>  $ get -s -p -r1.1 
> /export/home/gray/work/dumps/SCCS/s.ldap-extract-summaries.pl; 
> tramp_old_status=$?
> and the buffer comes up with
>  'get' is not recognized as an internal or external command,
>  operable program or batch file.
> I've attached the debug buffer (with verbose set to 10).
>
> Let me what additional data I need to provide to help resolve the
> issue. (I've explicitly added /usr/ccs/bin to my tramp-remote-path to
> ensure there are no path issues.)

That's a good idea (even though it still doesn't help).

I wonder if there is any way for you to debug this further?  Does
Emacs print an error message?  If so, M-x toggle-debug-on-error RET
might turn up a useful backtrace.

Otherwise, you might have to start edebugging code (go into the
middle of a defun and type M-x edebug-defun RET) and use single-step
mode (with SPC) to see what's happening.  Another useful thing to do
is to open more frames and to display pertinent buffers there, such
as the internal buffer used by VC for its commands.

Another thing to try is to check whether shell-command works.  The VC
support in Tramp uses shell-command, so if that fails, then it's
clear that VC will fail.

Open a remote file and try something like C-u M-! date RET.  It should
insert the output of the command (ie, the current date/time) into the
current buffer at point.  If you use uname -a instead of date, you can
verify that the command is actually run on the remote host.

kai
-- 
~/.signature is: umop ap!sdn    (Frank Nobis)