Michael,

I found that the patch doesn't work for sql code blocks. Here is an example

* this works
#+BEGIN_SRC sh :results output :dir /grid:
  ls
#+END_SRC

* this doesn't work
#+BEGIN_SRC sql :engine postgresql :dir /grid: :results output
  select 1+2 as three;
#+END_SRC

and here is what appears in *Messages*:

executing Sh code block...
Tramp: Encoding region using function `base64-encode-region'...done
Tramp: Decoding region into remote file
/plinkx:grid:/tmp/ob-input-7928Z4Q...done
Tramp: Encoding region using function `base64-encode-region'...done
Tramp: Decoding region into remote file
/plinkx:grid:/tmp/ob-error-7928zMd...done
Tramp: Encoding region using function `base64-encode-region'...done
Tramp: Decoding region into remote file
/plinkx:grid:/tmp/ob-input-7928Z4Q...done
Wrote /plinkx:grid:/tmp/ob-input-7928Z4Q
Code block evaluation complete.
executing Sql code block...
Tramp: Encoding region using function `base64-encode-region'...done
Tramp: Decoding region into remote file
/plinkx:grid:/tmp/sql-in-7928arv...done
Tramp: Encoding region using function `base64-encode-region'...done
Tramp: Decoding region into remote file
/plinkx:grid:/tmp/sql-out-7928Z_E...done
Tramp: Encoding region using function `base64-encode-region'...done
Tramp: Decoding region into remote file
/plinkx:grid:/tmp/sql-in-7928arv...done
psql -A  -F " "  -f "c:/tmp/sql-in-7928arv" -o "c:/tmp/sql-out-7928Z_E"
-f c:/tmp/sql-in-7928arv -o c:/tmp/sql-out-7928Z_E
No such file or directory 'c:/tmp/sql-in-7928arv'
Tramp: Inserting `/plinkx:grid:/tmp/sql-out-7928Z_E'...
Tramp: Encoding remote file /plinkx:grid:/tmp/sql-out-7928Z_E...done
Tramp: Decoding remote file /plinkx:grid:/tmp/sql-out-7928Z_E with function
base64-decode-region...
Wrote c:/Users/avorobi/AppData/Local/Temp/tramp.7928AeX
Tramp: Decoding remote file /plinkx:grid:/tmp/sql-out-7928Z_E with function
base64-decode-region...done
Tramp: Inserting `/plinkx:grid:/tmp/sql-out-7928Z_E'...done

So shell commands work but the arguments passed to psql have "c:" at the
beginning.

Thanks
Alex



On Wed, Sep 25, 2013 at 2:03 PM, Michael Albinus <michael.albinus@gmx.de>wrote:

> Eric Schulte <schulte.eric@gmail.com> writes:
>
> > I'd rather not hard-code the value of "/tmp/".  Perhaps you could rework
> > the patch so that it introduces a new customizable variable (including a
> > documentation string) so that users can set the value for their system.
>
> Will do, tomorrow. I would even prefer a more general solution, a new
> function which returns a ("the") temp directory on a remote host. This
> would be useful also outside org-mode. But this would require changes in
> Emacs/Tramp, which won't be applicable for org immediately (backwards
> compatibility, and alike).
>
> > If you can keep the patch under 15LOC we can include it w/o requiring
> > FSF copyright assignment, otherwise see [1] for contribution details.
>
> No problem. As Tramp maintainer, I have signed the FSF papers for Emacs
> 10+ years ago. This shall be valid also for org patches.
>
> > Thanks,
>
> Best regards, Michael.
>