On Mon, 14 Nov 2005, Daniel Berlin wrote:
> > No need to clear the entry first. That will just confuse readers.
>
> Actually, you do need to.
>
> It will live on from tmp_entry above.
>
So, you have a tmp_entry, you clear it and then assign to the only field
that will be used by the following call. How can that be necessary?
BTW, you can leave this svn_wc_merge_prop_diffs function alone for now.
I've left a TODO at the top (after you created this patch). We should look
into reimplementing this in terms of svn_wc__install_props.
> > How does this work with entries in the old format? (Note that we only
> > bump the WC format when we have write access, so we have to be able to
> > read the old format.)
>
> It doesn't.
> I told you it would fail with the old format, but i can make it work
> trivially.
>
Hmmm, maybe I missed that, but it needs to do this before going in.
Compatibility, you know... :-)
> > Isn't this error just the result of an internal inconsistency in this
> > file? Should be an assert in that case.
>
> > What is "the working copy" in this case? When we release this, we have
> > WCs that cache no properties and WCs that cache these three
> > properties.
>
> So you'd rather see "subversion" instead of "working copy"?
Nah, maybe not now when I rethink it. I was thinking about a function that
this isn't. If you document in the public struct that propcaching is wc
format dependent, then everything is clear IMO.
Also, please see my TODO item for using existence of has_props attribute
for optimizing svn_wc__has_props, which is actually necessary by the
status code.
Thanks,
//Peter
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Tue Nov 15 10:02:12 2005