[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: Re: [xacml-comment] Permissable function arguments for the any-of-any function
Hi Erik, On 16/02/2012 3:10 AM, Erik Rissanen wrote:
Hi Steven, Yes, it's just a typo.
Thanks. I assumed so. It's just that I had a developer challenge me to justify the assumption. > It's clear from the description of the any-of-any function that the argument function
receives atomic values only since there are no bags of bags in XACML.
I was actually thinking that if the omission were deliberate (and you've explained that it was not), then a component of the tuple could be either a primitive value or a bag containing just the primitive value, to suit the needs of the argument function. The description of the tuple and the procedure of taking the cross-product is sufficiently vague to allow such an interpretation, however unlikely. For the record, there is no value in supporting such complexity since a boolean function taking a bag argument with only one member can generally be replaced by a type-equal function operating on the primitive value directly. That line of thinking reinforced my belief that the omission was not deliberate, but it's hard to settle an argument with a belief. Regards, Steven
Best regards, Erik On 2012-01-29 23:35, Steven Legg wrote:The description of the any-of-any function in Appendix A.3.12 of the Committee Specification 1 version of the XACMLv3 core specification describes the first argument as naming "a Boolean function that takes n arguments" in contrast to the other higher-order bag functions where the first argument names "a Boolean function that takes n arguments of primitive types". I assume that the omission of the "of primitive types" in the description of the any-of-any function is just an editorial oversight. If not, then there needs to be a better description of how named functions that take bag arguments are to be applied by the any-of-any function. Regards, Steven