> -----Original Message-----
> From: Bert Huijben [mailto:bert_at_qqmail.nl]
> Sent: woensdag 14 oktober 2009 10:27
> To: 'Greg Stein'; 'dev_at_subversion.tigris.org'
> Subject: RE: svn commit: r39964 - trunk/subversion/libsvn_wc
>
>
>
> > -----Original Message-----
> > From: Greg Stein [mailto:gstein_at_gmail.com]
> > Sent: dinsdag 13 oktober 2009 18:48
> > To: dev_at_subversion.tigris.org
> > Subject: Re: svn commit: r39964 - trunk/subversion/libsvn_wc
> >
> > On Mon, Oct 12, 2009 at 15:56, Bert Huijben <rhuijben_at_sharpsvn.net>
> > wrote:
> > >...
> > > +++ trunk/subversion/libsvn_wc/update_editor.c Mon Oct 12 12:56:27
> > 2009 (r39964)
> > >...
> > > + if (kind == svn_wc__db_kind_unknown)
> > > + break;
> > >
> > > - ancestor_abspath = APR_ARRAY_IDX(ancestors, i, const char
> *);
> > > + SVN_ERR(svn_wc__db_node_hidden(&hidden, db,
> ancestor_abspath,
> > iterpool));
> > > +
> > > + if (hidden)
> > > + break;
> >
> > An ancestor better not be missing or hidden. These checks are better
> > as an assertion.
>
> That would raise an assertion if somebody added a new external
> (/separate working copy) in place of a base not_present, excluded or
> absent node. (The checkout phase would reach this part of the code)
>
> I don't think we want to raise exceptions on this kind of (user?)
> errors. A normal error could be valid, but this code path used to work
> in this case. (Because the entry api would just return NULL for hidden
> nodes)
Note that this is mostly temporary code until svn_wc__check_wc_root () returns some saner answers on unversioned nodes. (Its public wrappers document an error on not versioned nodes, but the code just returns false since the move to WC-NG).
I'm working on a patch that fixes this function in a fully WC-NG way, but it currently breaks the merge code.
Bert
------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=462&dsMessageId=2407448
Received on 2009-10-14 10:30:23 CEST