Re: tramp trouble with 24.3

[Michael Albinus]

Neal Becker <ndbecker2@gmail.com> writes:

Hi Neal,

> Neal Becker wrote:
>
>> Michael Albinus wrote:
>> 
>>> Neal Becker <ndbecker2@gmail.com> writes:
>>> 
>>>> This is a linux box talking to a linux box, so if doesn't work for me, I'll
>>>> bet I'm not the only one who will/has seen this.
>>> 
>>> Your prompt doesn't look to be the default one. I guess it is
>>> customized, isn't it?
>>> 
>>> Best regards, Michael.
>> 
>> No it isn't.  Maybe it's cause color prompts are on by default?  That's not 
>> my
>> idea.  That's fedora 18.

Many people are using Fedora 18, and I haven't seen any other complaint yet.

> Also, shell-prompt-pattern is working in shell-mode.  In shell mode:
>
> echo $PS1
> \[\033[01;32m\]\u@\h\[\033[00m\] 
> \[\033[01;34m\]\W\[\033[00m\]\[\033[1;32m\]\$\[\033[m\]

And? Tramp does not use shell-mode. Anyway, this prompt is hard enough
to parse. The Tramp manual gives you some hints how to avoid escape
sequences, when scripts are accessing your account from remote.

> Also, don't forget, the problem is not seen in 24.2

Maybe there was a pattern change in Tramp. As I said, I don't remember, but ...

> I used emacs from fedora.  Used rpmbuild --rebuild to build the fedora
> 20 rpm on
> fedora 18.  I don't see anything added by the fedora rpm that would change 
> this 
> prompt regex (but haven't searched exhaustively)

Sorry, I don't use Fedore, and I don't understand what you have changed
via rpmbuild etc. Anyway, if I could see your complete prompt (including
escape sequence), I could try to adapt tramp-shell-prompt-pattern to
your needs. The example above is not sufficient, because it does not
expand user and host names, and alike.

Just run your test case with tramp-verbose 10. There will be traces like

--8<---------------cut here---------------start------------->8---
12:37:39.845362 tramp-accept-process-output (10) # 
#$
--8<---------------cut here---------------end--------------->8---

(#$ is my prompt). Send those lines as coded attachment, in order not to
loose escape sequences. I will analyze them.

You could send it directly to me, in order to avoid bit lossages.

Best regards, Michael.