> -----Original Message-----
> From: Kannan [mailto:[email protected]]
> Sent: maandag 18 januari 2010 5:46
> To: C. Michael Pilato
> Cc: Philip Martin; SVN Dev
> Subject: Re: [PATCH] Fix #3544 - svn update does not restore excluded files
>
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> C. Michael Pilato wrote:
> [..]
> > This patch doesn't apply cleanly anymore because I committed one of you
> > other ones. The conflict is pretty easy to resolve, though.
>
> Yes, as said in this thread before the change in this patch is the
> required one.
>
> > That said, I don't understand the changes to fold_entry(). The function's
> > purpose is to modify an entry_t by copying a subset of fields from another
> > entry_t. It seems an odd place to put code that tweaks that entry_t's
> > "depth" field without respect to either the source entry_t's value or to the
> > flagset that dictates which fields to modify.
> >
> >
>
> /me nods. A file has the default and only depth of `infinity', so it's
> depth can't be updated to the dir's(ie, source entry's), also
> fold_entry() is the method where individual entries are obtained and
> processed, so made the change there. (In the process of finding some
> other place to make the change) please CMIIW. Thank you for the
> comments.
For the WC-NG work we are trying to remove all calls to this fold function as a
step towards removing all handling of svn_wc_entry_t. As WC-NG can store more
state about nodes than can be expressed in svn_wc_entry_t, every call to
fold_entry() can and in many cases will loose data.
In the WC-NG store we can exclude files (which we couldn't do in 1.6) and when
you see such a node as an entry you will see that its depth is excluded. But
depth excluded is not stored as a depth, but as a node status.
As far as I know, we completed moving all depth handling from the entries level
and use wc_db apis instead.
Bert
