W3C home > Mailing lists > Public > public-ws-policy@w3.org > May 2007

RE: ACTION-294 Ashok to open a CR issue with text to define "collection"

From: Ashok Malhotra <ashok.malhotra@oracle.com>
Date: Fri, 4 May 2007 10:11:30 -0700
To: "Fabian.Ritzmann@Sun.COM" <Fabian.Ritzmann@Sun.COM>
CC: "public-ws-policy@w3.org" <public-ws-policy@w3.org>
Message-ID: <20070504101130347.00000009940@amalhotr-pc>

Fabian:
The WS-Policy specification does not talk about implementations.
Implementations are free to do what they wish, and will generally use specialized data structures.  But that not the point.  The spec says a policy is an unordered collection of policy alternatives which can include duplicate alternatives.

All the best, Ashok

> -----Original Message-----
> From: Fabian.Ritzmann@Sun.COM [mailto:Fabian.Ritzmann@Sun.COM]
> Sent: Friday, May 04, 2007 5:49 AM
> To: Ashok Malhotra
> Cc: public-ws-policy@w3.org
> Subject: Re: ACTION-294 Ashok to open a CR issue with text to define
> "collection"
> 
> To point 3, at least when it comes to WSDL, I don't think duplicate
> subjects (i.e. WSDL elements with the same fully qualified name) make
> sense. They certainly are not ordered.
> 
> Regarding points 1 and 2, wouldn't it be an implementation decision if
> duplicates are maintained (bag) or discarded/collapsed (set)? I can't
> find any necessity to keep duplicates.
> 
> Fabian
> 
> 
> Ashok Malhotra wrote:
> >
> > My point was that the word 'collection' in Computer Science is used as
> > a generic term to cover several types of collections.
> >
> > See definition from Wikipedia below. Thus, it would be better to be
> > more precise about the exact type of collection we use in WS-Policy:
> >
> > In _object-oriented programming_
> > <http://en.wikipedia.org/wiki/Object-oriented_programming>, a*
> > collection class* is any _class_
> > <http://en.wikipedia.org/wiki/Class_%28computer_science%29> that is
> > capable of storing other _objects_
> > <http://en.wikipedia.org/wiki/Object_%28computer_science%29>.
> > Collection classes usually implement some kind of _data structure_
> > <http://en.wikipedia.org/wiki/Data_structure>, such as a _list_
> > <http://en.wikipedia.org/wiki/List_%28computing%29>, _map_
> > <http://en.wikipedia.org/wiki/Associative_array>, _set_
> > <http://en.wikipedia.org/wiki/Set_%28computer_science%29>, _array_
> > <http://en.wikipedia.org/wiki/Array>, or _tree_
> > <http://en.wikipedia.org/wiki/Tree_%28data_structure%29>. A collection
> > class is usually able to store an arbitrary number of data items, i.e.
> > the size of the collection is adjusted automatically.
> >
> > In the framework document we use 'collection' in three contexts:
> >
> > 1. "A _policy_ <http://www.w3.org/TR/2007/CR-ws-policy-20070330/> is a
> > collection of policy alternatives." Since alternatives can be
> > identical and there is no order between them, a policy is a bag of
> > alternatives.
> >
> > 2. "A _policy alternative_
> > <http://www.w3.org/TR/2007/CR-ws-policy-20070330/> is a collection of
> > policy assertions." Here, again, the assertions in an alternative can
> > be duplicates and are unordered, so bag seems to be the correct term.
> >
> > 3. "A _policy scope_
> > <http://www.w3.org/TR/2007/CR-ws-policy-20070330/> is a collection of
> > policy subjects to which a policy applies." Here I am not sure. Can
> > there be duplicate subjects in a policy scope? Aren't policy subjects
> > ordered?
> >
> > If the policy subjects in a policy scope can be duplicates and are not
> > ordered then all uses of 'collection' mean 'bag' and we can add a
> > definition such as "In this document the word 'collection' refers to
> > what is known in the literature as 'bag'. But I thought we should
> > clarify point 3 first.
> >
> > All the best, Ashok
> >
> 
Received on Friday, 4 May 2007 17:12:45 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:38:34 UTC