Tino Schwarze wrote:
> On Wed, May 05, 2010 at 11:38:51AM -0500, Les Mikesell wrote:
>> Does anyone know of a way to backup/move/copy checked out working copies
>> such that you can skip all the parts that 'svn update' would
>> reconstruct? I'm about to copy the home directories of a machine used
>> mostly for checkout/build/test operations to a remote location and it
>> occurred to me that most of what I'll be copying is unnecessary but I
>> don't want to disrupt the directory structure or hooks to the
>> repositories.
>
> I'd rather not try that trick - you will spoil the "I did not update
> that working copy since I need it at exactly this stage" use case - I'd
> be upset as an developer if my files vanished because of some clever
> admin. (And I'm admin as well, so I understand your desire not to copy
> redundant files.)
Just to put things in perspective here, it took two full nights to copy the home
directories over a WAN to its new location when in fact everything except a few
scripts and the top level directories could have been checked back out and
didn't need to be copied at all. As a developer, would you be happy about
losing a couple of days of work for something that should have a better way to
do? I happen to use backuppc to back the system up so it at least de-dups the
multiple identical copies across the workspaces and their .svn counterparts but
it would be nicer if there were a way to explicitly avoid or even remove
anything that could be regenerated. Maybe this would involve creating a script
to update to the current revision of each file in the workspace to get back to
the identical state - and of course keeping any modified files.
--
Les Mikesell
lesmikesell_at_gmail.com
Received on 2010-05-11 15:04:11 CEST