[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [Elist Home]
Subject: Re: [xacml] RE: Your input needed on Comment#33. Forwarded messag efrom Daniel Engovatov.
On Mon, 25 Nov 2002, Seth Proctor wrote: > > On Mon, Nov 25, 2002 at 10:29:03AM -0500, Polar Humenn wrote: > > What we have here is the generic application of the SAME FUNCTIONALITY, so > > why give the same functionality different names? > > Because this is how it is done with every other function in the spec. As a > result, you need this for the map function too if you want consistency. > Yes, I see the value of having a generic function that can map any type > to any type, but it breaks the type system that the rest of the spec has > defined. Not true. "integer-equal" is fundementally different that "string-equal", etc. Map is generic, or polymorphic over "bags". And, "integer-equal" specifies a function that determines equality over integers (i.e. as arguments). If you took your approach to naming the map functions, e.g. "integer-map" naming the resultant type only, then to be "consistent" you would really have to name the "integer-equal" function as "boolean-equal", which doesn't make sense. > Both Daniel and I are speaking from an implementors point of > view, and we both see the muddle that occurs when you have this > special-case function. The clean definitions in the spec completely > break when you allow the map function to not have its return type > explicitly defined. > > Implementors will already need to define things like -equal, or -one-and-only, > or -union for their new attribute types. Redefining the map function doesn't > change this. A good implementation will let a programmer do this with a > minimum of effort, but it is still necessary. To maintain consistency, and > to keep the type system intact, the map function must become type-map. > > > seth >
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [Elist Home]
Powered by eList eXpress LLC