emacs-orgmode
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [O] Invalid function: with-parsed-tramp-file-name with Perl


From: Eric Schulte
Subject: Re: [O] Invalid function: with-parsed-tramp-file-name with Perl
Date: Thu, 20 Sep 2012 12:29:50 -0600
User-agent: Gnus/5.13 (Gnus v5.13) Emacs/24.2.50 (gnu/linux)

Nick Dokos <address@hidden> writes:

> Nick Dokos <address@hidden> wrote:
>
>> I haven't chased it all the way down because the reverts are
>> making my head spin, but it may be that somehow the above commit
>> got lost somewhere - or it got fixed and then the big revert lost
>> the fix. Maybe Eric or Bastien remembers what happened.
>> 
>
> This doesn't make sense (I blame the head-spinning reverts :-) ), so let
> me try again: Eric's commit broke it between 7.8.03 and 7.8.04. It got
> fixed somehow, either because of a revert or because some fix was
> actually applied, I don't know which. It was working until 7.8.10 and
> then Bastien's commit reverted to (broken) 7.8.04, so from 7.8.11- on
> the breakage exists.
>
> So if there was a fix, we need to find it and reapply. If not, someone
> (presumably Eric) will need to take a look and possibly develop a fix.
>

I've spent some time looking at this.

A couple of things are happening.  Back in the time of 7.8.03 and
7.8.04, the org-babel-execute:sh call happened in such a context that
all external function calls would take place on the remote machine if
the :dir header specified a tramp path to a remote machine.  This is
true both before and after my 5cb80c7e5b9bca commit.

However this has changed between then and now, meaning that reverting
the 5cb80c7e5b9bca commit now will not solve this problem.  In the
current branch there is no change of context s.t. external calls made by
the org-babel-execute:* functions happen by default on the remote
machine.  I think this is the real problem.  Perhaps it would be
possible to git bisect with another language like perl?

Sorry I'm not more help here, the remote-execution magic was mainly
implemented by Dan Davison.

Best,

>
> Nick
>
>
>

-- 
Eric Schulte
http://cs.unm.edu/~eschulte



reply via email to

[Prev in Thread] Current Thread [Next in Thread]