On Wed, Jan 10, 2001 at 02:30:39PM +1100, raf <[EMAIL PROTECTED]> wrote:
| > And supposing I want different commands per invoking user or something else?
| i don't get your question. that's precisely what this is for.
| if you mean local user, that's why each user has their own ~/.rsyncrc.
| if you mean remote user, that's why it takes "user" parameters.

And the "or something else" part? The point is that there may be an
arbitrary plethora of random things one might want to base such a
decision on. Obviously they can't all be encoded in your patch because
they aren't even defined yet. So while your patch might address your needs,
it will be inadequate in some weird portion of the world.

| > This seems like flexibility you don't need (or want) to put into rsync.
| i obviously do need the flexibility and rsync seems the best place for it :)
| to each their own.
| 
| > Just set RSYNC_RSH to the path of a shell script with this intelligence.
| > Works for me, and is arbitrarily flexible. Keeps rsync itself simple.
| 
| sounds like a slow waste of processes to me and i don't think a < 0.2%

How waste? Test whatever, assemble ssh command line, exec ssh.
And all the benfits of a fairly flexible programming language to
express the checks (sh).

I guess my two points are that yes you _can_ do it with $RSYNC_RSH and by
putting it inside rsync instead of outside in the RSYNC_RSH hook you restict
the flexibility to the user while adding bloat (albeit small bloat).
-- 
Cameron Simpson, DoD#743        [EMAIL PROTECTED]    http://www.zip.com.au/~cs/

A Master is someone who started before you did. - Gary Zukav

Reply via email to