Trent Nelson wrote on Tue, Jan 10, 2012 at 20:56:27 -0800: > Howdy folks, >
Hi! > The current wording of svn_delta_editor_t's absent_(file|directory) > callback functions is a bit misleading: > > "In the (file|directory) ... but cannot be conveyed to the > consumer (perhaps because of authorization restrictions)." > ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ > > Because of that wording, I just spent a chunk of time looking through > the code trying to find if these callbacks would be invoked for any > reason *other* than an authorization restriction. > > I've concluded that no, they wouldn't :-) > Sounds like this is a good thing to clarify... > +++ subversion/include/svn_delta.h (working copy) > @@ -916,8 +916,8 @@ > > /** In the directory represented by @a parent_baton, indicate that > * @a path is present as a subdirectory in the edit source, but > - * cannot be conveyed to the edit consumer (perhaps because of > - * authorization restrictions). > + * cannot be conveyed to the edit consumer because of authorization > + * restrictions. > * But I disagree with this fix: I think the original language here was inetntionally vague due to forwards compatibility considerations. How about saying that _currently_ (1.7) the only possible cause is authz restrictions? > * Any temporary allocations may be performed in @a scratch_pool. > */ > @@ -1040,8 +1040,8 @@ What about svn_editor_add_absent()? Is its docstring okay?
