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

Re: [PATCH] issue #3641: svnsync fails to mirror certain dir replaces

From: Tino Schwarze <subversion.lists_at_tisc.de>
Date: Thu, 23 Sep 2010 14:06:44 +0200

Hi Daniel et al,

On Mon, Sep 20, 2010 at 05:21:41PM +0200, Daniel Shahaf wrote:

> > > > Has anybody reviewed the patch?
> > > >
> > > > Too late for 1.6.13, I suppose?
> > >
> > > No. I've just added it to STATUS. That's the easiest way to ask for
> > > reviews here :-)
> >
> > Thanks, but the Justification is not just "Could lead to sync'd
> > repositories being different from the master." It's more severe: svnsync
> > fails to copy certain operations. It's actually a "your sync totally
> > stops working and you can't do anything" kind of failure - see
> > http://subversion.tigris.org/issues/show_bug.cgi?id=3641
> >
> > If you happen to run into this:
> > "svnsync: File not found: revision 5, path '/H/B/C'"
> > you won't be able to sync the repository anymore after that. Which I
> > consider rather disastrous given that we're going to use svnsync to
> > provide partial copies of our repository to our customers.
> >
>
> Scenario #1. As you describe: svnsync bombs out and refuses to
> continue.
>
> Scenario #2. Per my comment on the issue, svnsync looks for /H/B/C ---
> which DOES happen to exist --- and copies that to the target repository.
>
> #1 is noisy failure. #2 is silent corruption (the source and target
> repositories differ).
>
> So, yes, both of them are possible outcomes; #2 is more severe and #1 is
> more likely to occur in practice. :-)

Agreed. #1 is more shocking in the short therm (which I currently cared
since it would have been a showstopper for me). #2 is probably more
shocking to discover later.

Thanks,

Tino.

-- 
"What we nourish flourishes." - "Was wir nähren erblüht."
www.tisc.de
Received on 2010-09-23 14:07:24 CEST

This is an archived mail posted to the Subversion Dev mailing list.