On 15 Jan 2004, [EMAIL PROTECTED] wrote: > On Thu, 15 Jan 2004, Michael Albinus wrote: > >> Adrian Lanz <[EMAIL PROTECTED]> writes: >> >>> My question: is there some tramp built-in functionality to >>> "execute" remote files on the local machine? After step 2 (tramp >>> and dired have brought up the remote directry listing) I would >>> like to do something like dired-do-shell-command (pressing "!"), >>> which in this context would mean: download the remote file to some >>> (configurable) temporary place on the local file system and then >>> use the local application I provide in the mini-buffer on that >>> temporary file. What is sent back to the remote system after the >>> local command has been run would be configurable. After latex'ing >>> a remote file on the local machine, I probably would like tramp to >>> save the dvi file on the remote system; after xpdf'ing a remote >>> file on the local machine, nothing at all would have to be sent >>> back to the remote machine. > > The suggestion here doesn't really make sense (to me). If you have a > remote latex file, and you want to operate on that, then either > operate on it remotely, or copy it locally, operate on it locally, > and store the source file and all intermediate and final results > locally. (I can see one case where the described action could be > useful - where the local node is faster than the remote node, and so > you want to process locally. But this still doesn't make sense - why > invoke the remote node at *all*?) > > The problem is how do you tell tramp what files are needed from the > remote machine in order to be able to operate on it locally? In the > LaTeX example you give - when I am writing, I typically have 5 > separate files which are `included'. How to tell tramp you have to > copy them all over locally before latex can possibly be run? There > already on the remote system, why not just run latex there?
I agree, that things get complicated when compiling the downloaded file depends on further remotely stored files. Let's start with simple cases ;-) Concerning the remote system. In my case these systems are Windows NT or whatever Windows machines, whereas my local machine is a Unix Desktop (Solaris) or notebook (Linux). My collegues store thier contribution in one of MS-Office formats or hopefully in some generally readable output format (pdf, html, ...). I would like to quickly read (and possibly change) these files with my locally installed OpenOffice or xpdf or whatever (I can't get a Windows display on my X window system, do I?), and put my contributions - fully documented - as source (tex) and in some output formats (pdf, ps, dvi, ...) on that repository. By downloading and uploading the files I can do both now (thanks to tramp!), and I just wondered if I could do that without the intermediate downloading/uploading step. >> No, it isn't implemented (yet). But it is at least on my private >> TODO list. I even would like to have the opportunity to execute >> commands on the remote host, which is a little bit different to >> your scenario. By this, commands like find-grep-dired could be >> extended to work on remote hosts, too. > > However, *this* kind of functionality would be very nice. The common > scenario is that I am editing source files remotely, and want them > compiled remotely. Press f4 to invoke make, and it is made. But > unfortunately, tramp doesn't yet have this functionality :) > > If it can hook into the debugging mode or whatever, then doubly > nice. Well, this functionality is not on *my* wish list (yet), and what I would like to see implemented should be much easier to write, isn't it? Just a simple dired-mode hook? ;-) Thanks, Adrian. _______________________________________________ Tramp-devel mailing list [EMAIL PROTECTED] http://mail.nongnu.org/mailman/listinfo/tramp-devel
