On Wed, Feb 25, 2009 at 11:35 AM, Ingo Weinhold <ingo_weinhold@xxxxxx> wrote: > > This is not an obscure corner case. AFAIK all storage kit class setters are > destructive and render the object invalid when they fail (most have an > InitCheck() even). So, while I agree that in a new API it would be nicer to > have these BPath methods work differently, it's not even inconsistent as it > is now. Nor it is particularly annoying -- one can just use the API as > documented. I agree with Ingo, though I do understand the argument concerning the behavior shown. I think this behavior and a possible change/improvement should be documented somewhere and can maybe be put into consideration in Haiku R2. Regards, Ryan