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

Re: still inconsistent, file in repository, but not in wc

From: Ben Collins-Sussman <sussman_at_collab.net>
Date: 2002-08-21 18:30:14 CEST

Philip Martin <philip@codematters.co.uk> writes:

> Ben Collins-Sussman <sussman@collab.net> writes:
> > OK, there is something messed up with our wc library in HEAD. I'm
> > able to reproduce these bugs using r3004, on an ra_local working copy
> > of our standard greek tree:
> >
> > $ cd A
> > $ mv D was_D
> > $ svn st
> > subversion/libsvn_wc/status.c:251: (apr_err=2, src_err=0)
> > svn: No such file or directory
> > svn: assemble_status: D: No such file or directory
> I thought the original bug was that the entries file didn't refer to
> the directory, even though the directory was present. You are
> demonstrating the opposite, the directory is missing but the entries
> file still refers to it.

That's correct, I'm not talking about the original bug anymore. I'm
talking about a new bug -- namely, one stemming from my proposed fix
to Herr Solo:

  If you simply use the operating system to remove a file or directory,
  we no longer see a "!" in status output, but a full-out error.

My proposal to fix the original bug was simply to blow away the
defective directory, and have 'svn up' recreate it. But now we've
been sidetracked by this other bug.

> The original bug, where the directory is missing from the entries file
> is harder to fix. In that case the client doesn't know that the
> directory is missing, so it won't report it. I suppose it might be
> possible to change the server to assume that anything not reported is
> missing?

Well, this is a much larger problem/design discussion. It's the whole
reason checkouts aren't restartable, and why Solo Turn is screwed when
his entries file is missing something. Our wc library is designed to
assume that an entries file is *always* complete. The plus side of
this is that it can describe tiny differences to the server when
sending an update report... (CVS has to explicitly describe the
existence of EVERY item to the server.) The downside of this is what
we're experiencing now -- if the entries file is incomplete, the wc
has no idea, and the server can't infer that either.

Issue 714 is similar to the restartable checkout issue; they both
revolve around this idea of changing our wc design.

But is there an issue filed already that describes the immediate

To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Wed Aug 21 18:32:01 2002

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

This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.