Re: svn commit: rev 1985 - trunk/subversion/libsvn_client trunk/subversion/mod_dav_svn trunk/subversion/libsvn_ra_dav

On Mon, May 20, 2002 at 09:16:58PM -0500, Ben Collins-Sussman wrote:
> Greg Stein <gstein@lyra.org> writes:
>...
> > > + CHKERR( rb->ras->callbacks->set_wc_prop(rb->ras->callback_baton,
> > > + rb->current_wcprop_path,
> > > + rb->vuh.name->data,
> >
> > The rb->vuh.name->data should be replaced with: SVN_RA_DAV__LP_VSN_URL
> >
> > (the need to keep 'vuh' around is an artifact of the old editor interface;
> > as we migrate to the new interface, we'll be tossing it. set_wc_prop
> > already uses a sane interface, so 'vuh.name' isn't needed there)
>
> Huh? rb->vuh is a structure that is being reused over and over, with
> a fixed unchanging 'name' value of SVN_RA_DAV__LP_VSN_URL. Why would
> you want to toss it? Isn't it more efficient this way?

The "old" property interfaces required an svn_stringbuf_t. And since the
property name was a simple constant, I introduced 'vuh' to avoid creating
the stupid stringbuf every time that I wanted to set the property.

Now that we've moved to 'const char *' for the prop name, we can just pass
in the symbolic constant. No need to wrap a stringbuf around that.

set_wc_prop has the new char* interface, so it can be changed. Eventually,
fetch.c will drive 'new' editors, and we can toss 'vuh' and just pass the
symbolic constant to the (new-style) property setters functions.

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