Re: atomic revprop commits at the RA level

On 8/10/06, C. Michael Pilato <cmpilato@collab.net> wrote:
> Suggest:
>
> svn_error_t *(*change_edit_prop)(void *edit_baton,
> const char *name,
> const svn_string_t *value,
> apr_pool_t *pool);
>
> The idea being we're modifying a property of the edit itself, not the tree
> that's being edited.

After trying out the other way (making this a new ra API that works
kind of in parallel with the editor), writing half a patch, and
putting it aside for a couple of days, I think I agree that going
through the editor is cleaner.

If we want to add a function to the svn_delta_editor_t struct, that
means we have to make a new svn_delta_editor2_t struct and bump the
~50 APIs that use it, right?

Would it be reasonable to, while we're at it, make svn_delta_editor2_t
an opaque type accessed through svn_delta_open_root(editor, ...)
functions and so on? This would allow future additions to the editor
API without needing to bump 50 APIs. On the other hand, maybe adding
methods to the delta editor API should be painful :)

--dave

-- 
David Glasser | glasser_at_mit.edu | http://www.davidglasser.net/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org