# The following was supposedly scribed by
# Roger Keays
# on Thursday 24 June 2004 07:56 pm:
>I'm interested in running subversion in a distributed fashion
me two:)
>and found many
>interesting articles about this on the web.
Any more references?
>I liked the sound of svk
>(http://svk.elixus.org), except after reading the code and documentation I
>found it didn't seem to match up to svn's largely well documented, well
>commented C code. No offence intended to the author of course - I just get
>a little bit intimidated by long lists of perl code with no whitespace or
>comments.
Well, svk is only at revision 0.15, so you've got to allow some bloody edges,
but what I would like to know (and it doesn't seem to be in the svk
documentation) is why are there 4500+ code lines of perl needed to make
subversion distributable? (the fact that there are more white lines than
comments is also disturbing.)
The way I see it, this could be done (albeit in-efficiently) with the svn
tool, some hook scripts, and a little glue. Here's a scenario:
My workstation is the 'spoke' and the master repository is the 'hub'. I start
the process with a standard checkout from the hub, and run an export from
there to the 'hop' on the spoke machine. I then import the hop into a clean
repository on the spoke, and checkout to a working copy, from which I work.
When I'm done working, (presumably after several local checkins) I export
back to the hop and import (--force) back to the original working copy on the
spoke, from which I checkin to the hub.
Yes, this is a bit clunky, but it doesn't seem a far cry for svn to have some
kind of --local switch to facilitate the whole thing (preferably without the
hop directory.)
--Eric
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Fri Jun 25 04:35:34 2004