David Bustos wrote: > Quoth Liane Praza on Tue, Sep 09, 2008 at 05:18:40PM -0700: >> David Bustos wrote: >>> Doesn't it indicate that the repository is corrupt? Sure, most clients >>> may choose to ignore that, but I don't think we should prevent those >>> capable of notifying the user. Unless there's some other function to >>> detect corruption. >> "Corrupt" isn't the word I'd use. It does has some vaguely odd property >> groups. > > Well maybe I don't understand.
I use "corrupt" to define a repository which has an underlying malformed database. "corrupt" has traditionally been a pretty severe condition which is caught by configd itself. This situation is not that case, but I doubt this semantic argument I made was a useful aside. > How can this happen? The only likely scenario I can come up with is that a program or administrator manipulated the repository and changed the type of the template property group. Alternatively (and less likely), a program or administrator created a property group with a template-ish name, but has no template data. I suppose there's also the possibility that the repository is nefariously eating itself and running around changing property group types, which would fit your definition of corruption. > Do supported > operations lead to this situation? If you consider directly manipulating restarter/state supported, then yes. But, realistically, no. We're protecting from an unlikely condition. > Does the state confuse any other > operations? No. You simply don't have a valid template definition. liane
