On Thu, Dec 18, 2014 at 11:22:09AM +0000, Darren J Moffat wrote: > On 12/18/14 00:15, Jan Pechanec wrote: > > for example, metaslot on Solaris is always 0 so slot-id=0 > >would be reliable there to use to access the soft token. Jan. > > It is the zeroth slot in the list of slots not a slotid with a value > of 0 - the distinction is subtle. But it still works to speak of the nth slot. Perhaps the attrbite should be called 'slot', without the '-id', 'slot-list-offset' perhaps. > I don't think we should have slot-id, it isn't stable and I know > that some vendors use random values. But there's still a list with a zeroth element. The list could be returned in different order every time, but that would just make this attribute useless (and so unused) with such a provider, but not harmful. Perhaps we should say that this attribute should always be used in conjunction with the provider library attributes, to ensure that this attribute has the desired semantics. I.e., unless coupled to a library that gives it semantics, this attribute has none. Nico --