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

Re: solution to issue 704

From: Greg Stein <gstein_at_lyra.org>
Date: 2002-05-20 12:51:56 CEST

The refetch is expensive (right now, it is 3 requests per URL fetched; we
could drop that to two, and in some cases down to 1; with a bunch more work,
we could drop the "average" below 1). It is *way* better to return them in a
single operation.

Of course, the URLs are only needed at commit time. It is feasible that you
could 'svn update' a particular resource before needing to commit it, thus
you would be getting the new URL anyways. It is actually somewhat of a
gamble, with very imperfect information.

Cheers,
-g

On Mon, May 20, 2002 at 12:32:09PM +0300, Alon Ziv wrote:
> Just a dumb question from a lurker: can't the (relatively) new
> 'refetch-invalid-urls'
> mechanism solve this far more cleanly? We can just clear all URLs of the
> switched area,
> and they will be automatically re-fetched as needed, no?
>
> (Disclaimer: I haven't tried this...)
>
> -az
>
> ----- Original Message -----
> From: "Greg Stein" <gstein@lyra.org>
> To: "Ben Collins-Sussman" <sussman@collab.net>
> Cc: "SVN Dev List" <dev@subversion.tigris.org>
> Sent: Monday, May 20, 2002 11:53 AM
> Subject: Re: solution to issue 704
>
>
> > On Sun, May 19, 2002 at 12:43:57PM -0500, Ben Collins-Sussman wrote:
> > >...
> > > To generate this post-switch <resource-walk>, we're running dir_delta,
> > > comparing revision 0 against the new switched-subtree. So dir_delta
> > > is sending out an editor 'add' command for every single item in the
> > > switch-destination.
> >
> > Oh, geez. "Hey, I've got a hammer. That looks like a nail!"
> >
> > You're producing a tree. That is done by walking the tree using the
> > svn_fs_dir_entries() call. It is certainly not a delta operation.
> >
> > >...
> > > It's a shame, because I've been able to essentially re-use both
> > > mod_dav_svn's and ra_dav's editors to do the resource-walk, with
> > > almost no modifications. It's been very nice. But now it seems that
> >
> > "almost". But what you're doing is forcing an editor and a driver to do
> > double-duty across two entirely different semantic concepts. But with a
> > tweak. Not quite the same. IMO, it is a big recipe for confusion.
> >
> > On the ra_dav side, you're simply going to get a bunch of elements for the
> > walk. Don't try and drive an editor. Just start calling set_wc_prop. This
> is
> > NOT an editor. It is a simple walk of the tree. Don't try to pretend
> > otherwise.
> >
> > >...
> > > Instead, it will just call the 'set_wc_prop' callback sitting in the
> > > RA session baton directly. Does that sound reasonable?
> >
> > Absolutely.
> >
> > Screw dir_delta. It isn't a "walk" system. It is about deltas. On the
> > client, just map the incoming data to set_wc_prop. Clean and simple; and
> > even better: understandable.
> >
> > Cheers,
> > -g
> >
> > --
> > Greg Stein, http://www.lyra.org/
> >
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
> > For additional commands, e-mail: dev-help@subversion.tigris.org
> >
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
> For additional commands, e-mail: dev-help@subversion.tigris.org

-- 
Greg Stein, http://www.lyra.org/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Mon May 20 12:49:57 2002

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