RE: [xacml] functions specifications

[Polar Humenn <polar@syr.edu>]

On Fri, 6 Sep 2002, Daniel Engovatov wrote:

> Some of this is recommended, with appropriate standards cited in this
> document: http://www.w3.org/TR/xmlschema-2/

Can you point out which ones?

Thanks,
-Polar

>
>
>
> -----Original Message-----
> From: Polar Humenn [mailto:polar@syr.edu]
> Sent: Friday, September 06, 2002 8:45 AM
> To: XACML
> Subject: [xacml] functions specifications
>
>
>
> We need explicit specifications or pointers to specifications for the
> certain functions. A specification should include the exact type (i.e.
> xs:string, xs:dateTime, etc) of its arguments and the exact type of its
> return value. Also, we need explicit, if not formal semantics for the
> function's behavior.
>
> Note, their type signature is not sufficient. If we cannot find
> specifications for the following, I'd just assume we leave them out.
> However, that might not be good for some of the functions that we think
> are necessary.
>
> I'm reluctant to issue defintions for these things, because, I'm not up on
> internationalization of strings, and I'm definiately not familiar with the
> nuances of XML processing. For instance, "string-equal" works two
> arguments of xs:string. In XML, what do you do with leading and trailing
> whitespace or mulitiple lines? Does leading and trailing white space get
> ignored, or reduced, etc.? I'll bet *everybody* had a different approach.
>
> The following functions need specification: (They are also listed in the
> Functions Document I put on the list with [NEED SPECIFICATION]. Here's a
> compiled list.
>
>                    return    1st Arg   2nd Arg
>                    type
> string-equal      xs:boolean xs:string     xs:string
>    Q: leading and trailing spaces, multiple lines, etc.
>
> date-equal        xs:boolean xs:date       xs:date
> time-equal        xs:boolean xs:time       xs:time
> dateTime-equal    xs:boolean xs:dateTime   xs:dateTime
> anyURI-equal      xs:boolean xs:anyURI     xs:anyURI
>
> x500Name-equal    xs:boolean xs:x500Name   xs:x500Name
>   Q:   Do we have an adequate specification on this from pkix?
>
> rfc822Name-equal  xs:boolean xs:rfc822Name xs:rfc822Name
> hex-equal         xs:boolean xs:hex        xs:hex
>    Q: Is the type xs:hex or xs:hexBinary?
>    Q: leading and trailing spaces, multiple lines, etc.
>
> base64-equal      xs:boolean xs:base64     xs:base64
>    Q: Is the type xs:base64 or xs:base64Binary?
>    Q: leading and trailing spaces, multiple lines, etc.
>
> string-greater-than              xs:boolean xs:string  xs:string
> string-greater-than-or-equal     xs:boolean xs:string  xs:string
>    Q: leading and trailing spaces, multiple lines, etc.
>    Q: What if two different international strings?
>
>
> time-greater-than                xs:boolean xs:time    xs:time
> time-greater-than-or-equal       xs:boolean xs:time    xs:time
>    Q: Are these relative to only one day? with Timezones?
>
> date-greater-than                xs:boolean xs:date    xs:date
> date-greater-than-or-equal       xs:boolean xs:date    xs:date
> datetime-greater-than            xs:boolean xs:date    xs:date
> datetime-greater-than-or-equal   xs:boolean xs:date    xs:date
>
> string-match      xs:boolean  xs:string      xs:string
> rfc822Name-match  xs:boolean  xs:rfc822Name  xs:rfc822Name
> x500Name-match    xs:boolean  xs:x500Name    xs:x500Name
>
> Ann has a whole lot of other functions in the compliance tests such
> as:
>
> add-dayTimeDurations
> subtract-dayTimeDuration-from-time
>
> of which I still do not know how they are to proceed. I don't even know if
> we have a type for a datTimeDuration or what it looks like. Does anybody?
>
> I suggest that we leave all of these out, especially if we cannot come up
> with specifications for them. We can add them later when people do figure
> out how they want them to operate.
>
> Cheers,
> -Polar
>
>
>
> ----------------------------------------------------------------
> To subscribe or unsubscribe from this elist use the subscription
> manager: <http://lists.oasis-open.org/ob/adm.pl>
>
> ----------------------------------------------------------------
> To subscribe or unsubscribe from this elist use the subscription
> manager: <http://lists.oasis-open.org/ob/adm.pl>
>