[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

RE: Re: 1.7.x returns invalid XML

From: Andreas Kirsch <kirsche40_at_gmx.de>
Date: Thu, 17 Jan 2013 07:04:49 -0800 (PST)

> SVN's goal is that every 1.x client can communicate with every 1.x
> server. This means that you can use 1.6 clients to have .svn dirs in
> every subdir while upgrading the server to 1.7.
This is what I expected. :-)

> When you mention network drives, there's an alarm ringing though
I think I did not explain the use case very well.

> Working copies are supposed to be stored and accessed locally only.
We are working at any time with local copies. Nobody works on the network drive because it is read-only for every user. The content of the network drive is updated by a new checkout each 15 minutes.

Only the trunc is copied to a network drive. A user can copy each directory of this read-only working copy from network drive locally - which makes it a local working copy. It is like checkout without locks from network drive instead of a checkout from server. I know this is stupid but there are some valid reasons to (mis-)use SVN and network drives it in this way.

> Another reason is that this reduces the chance
> that different clients access the same working copy, which causes
> conflicts when mixing e.g. 1.6 and 1.7.
This can never happen because each user has its own working copy by checkout over svn or by copy from network drive. If a user wants 1.6 metadata structure (each directory contains a .svn) he can use a svn 1.6 client or the network drive copy. If a user - like me - wants a 1.7 metadata structure (only top level directory contains a .svn) he has to use a 1.7.x client.

> In the past, I have managed to avoid these issues by simply not sharing
> working copies via network, so I don't really know if that explains your
> problem.
Like I said before it is not a sharing working copy but a shared read-only checkout without ability for manipulation.

> I also seem to remember reading that the changes in the
> metadata format (plain text files to sqlite DB) make such a setup even
> more brittle.
Hm, I have to investigate this one.

So far thanks for your comments. Today I become sick. Within the next days I am unable to check the results of the updated server task (mentioned earlier) in my department. But I will give feedback as soon as possible.



To unsubscribe from this discussion, e-mail: [users-unsubscribe_at_tortoisesvn.tigris.org].
Received on 2013-01-17 16:04:55 CET

This is an archived mail posted to the TortoiseSVN Users mailing list.