OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.

 


Help: OASIS Mailing Lists Help | MarkMail Help

ebxml-transport message

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [Elist Home]


Subject: RE: [Fwd: RE: on the issue of PartyId]


Henry

See comments inline.

David

-----Original Message-----
From: Henry Lowe [mailto:hlowe@omg.org]
Sent: Friday, September 22, 2000 6:43 AM
To: ebxml transport
Subject: RE: [Fwd: RE: on the issue of PartyId]


David,

Being as "context" wasn't defined in the MS doc and "type" wasn't 
defined until your reply, I'm going to change my previous "no big 
deal" to advocacy of "context" where "context" is a domain name (OK 
we probably need a slightly tighter definition than that ;-).

##Henry, I don't think we need a separate context attribute as if we use a
URI, then the domain name (aka context) is included **within the URI**
according to well defined rules. Specificially the URI spec [RFC 2396] says
...

  "A URI can be further classified as a locator, a name, or both.  The
   term "Uniform Resource Locator" (URL) refers to the subset of URI
   that identify resources via a representation of their primary access
   mechanism (e.g., their network "location"), rather than identifying
   the resource by name or by some other attribute(s) of that resource.
   The term "Uniform Resource Name" (URN) refers to the subset of URI
   that are required to remain globally unique and persistent even when
   the resource ceases to exist or becomes unavailable."

... and the URN spec [RFC 2141] says ...

   "All URNs have the following syntax (phrases enclosed in quotes are
   REQUIRED):

                     <URN> ::= "urn:" <NID> ":" <NSS>

   where <NID> is the Namespace Identifier, and <NSS> is the Namespace
   Specific String.  "

... and the URN Namespace Definition spec [RFC 2611] says ...


  "The purpose of this document is to outline a mechanism and provide a
   template for explicit namespace definition, along with the mechanism
   for associating an identifier (called a "Namespace ID", or NID) which
   is registered with the Internet Assigned Numbers Authority, IANA."

... so my conclusion is that a context attribute that identifies a domain is
completely superfluous as the URN namespace identifier serves the same
purpose. Do you agree Henry?
##

You proposal of either UID or "userdefined" more or less says that 
ebXML MS won't work without the TPA -- no TPA, you haven't a clue 
as to what the userdefined value means unless you make an incredibly
lucky guess.  
##No. What I'm saying is that if you use a URI, you can probably work out
who the party is without a Party Agreement by following a namespace
resolution processes (there are RFCs for these two but I haven't read them -
try RFC2483). This would in practice work the same way as how you describe
context below. On the other hand if you use userdefined then there will need
to be some prior agreement between the parties.##

At least with "context" (a momain name), if you don't 
have a TPA you still stand a chance of recognizing the domain name 
and you then know how to handle the PartyID value sting, e.g., if 
context=CORBA then the PartyID is an IOR (and as an added bonus, you 
also know that you want to choose the CORBA mapping for the MS->transport).

With almost no extra effort, you get a whole pile of added info 
and I think you have nothing to loose.  How could you possibly 
disagree with this ;-)
##I can and I do ... see earlier response/explanation on URIs. All you would
need to do to support CORBA using a URI is define and register a URN along
the lines of "urn:corba:<ior>" and you're home and dry.

IANA also has well defined procedured that make this easy to do.Secondly if
we use a context attribute then we would have to define our own equivalent
of RFC2611 to define what it should contain and how it should be maintained.

Can we possibly continue to disagree ;) Well maybe ...

Best wishes. David
##

Best regards,
Henry

PS: the DTD in the next message you sent will have to be updated.
-----------------------
At 05:17 PM 09/21/2000 -0700, you wrote:
>>>>However, isn't "type" identical to "context" in MS 2.0?<<<
>
>Nearly but not quite.
>
>Context identified the domain to which the value of the party id belonged.
>
>Type just tells you whether party id is a URI or a value that is
>"userdefined". It does not identify the domain as this is contained within
>either the URI, or alternatively derived from the Party Agreement if it is
>used defined.
>
>David
>
>-----Original Message-----
>From: Henry Lowe [mailto:hlowe@omg.org]
>Sent: Thursday, September 21, 2000 12:55 PM
>To: ebxml transport
>Subject: RE: [Fwd: RE: on the issue of PartyId]
>
>
>I have no problem with 
>   <PartyId type='userdefined'>123dwqf09u23rnsadp9</PartyId>
>
>As the userdefined type (if there is one, if not default to URI)
>will probably come from the TPA along with the type of transport 
>to map to, the negotiation to establish the TPA will ensure the 
>MS can understand the type in the PartyID.
>
>However, isn't "type" identical to "context" in MS 2.0?
>
>Best regards,
>Henry
>------------------------------
>At 12:11 PM 09/21/2000 -0700, Burdett, David wrote:
>>Again
>>
>>I agree with Chris, using a URI, removes the problem from TRP of defining
>>what can validly be present in a context, and allows us to use IANA
>instead.
>>
>>So I still think we should go with URI as the prime mechanism.
>>
>>On the other hand if we really wanted to be flexible though we could have
>an
>>additional alternative of ...
>>
>><PartyId type='userdefined'>123dwqf09u23rnsadp9</PartyId>
>>
>>where in the latter case, the recipient is expected to recognize the
>PartyId
>>through a previously mutually agreed mechanism. In addition if type is not
>>present, then it defaults to 'URI'. This means you either have:
>>1. a URI, or
>>2. a coding scheme mutually agreed between the parties
>>
>>Thoughts?
>>
>>David
>>
>>-----Original Message-----
>>From: Chris Ferris [mailto:chris.ferris@east.sun.com]
>>Sent: Thursday, September 21, 2000 11:49 AM
>>To: ebxml transport
>>Subject: Re: [Fwd: RE: on the issue of PartyId]
>>
>>
>>A URI provides its own context. This is what I am recommending, not that
>>any particular context shall be used or excluded (although I pointed out
>>that telephone # doesn't qualify as an URN which would seem to me
>>to be the form of URI which would be recommended).
>>
>>By that I mean that:
>>
>><PartyId context="duns">1234567890123</PartyId>
>>
>>is the same as:
>>
>><PartyId>urn:duns:1234567890123</PartyId>
>>
>>and that I would recommend that the latter form be used.
>>That was the jist of my proposal.
>>
>>Cheers,
>>
>>Chris
>>
>>Scott Hinkelman/Austin/IBM wrote:
>>
>>> I agree with Henry. I don't think we should dictate what it is, only
>>> dictate a way
>>> to dictate what it is.
>>>
>>> Scott Hinkelman
>>> Senior Software Engineer, IBM Austin
>>> Emerging Technologies, SWG
>>> 512-823-8097 (TL 793-8097) (Cell: 512-940-0519)
>>> srh@us.ibm.com, Fax: 512-838-1074
>>>
>>> Henry Lowe <hlowe@omg.org> on 09/21/2000 01:37:49 PM
>>>
>>> To:   "Brunner, Eric" <EBrunner@Engage.com>
>>> cc:   "'Christopher Ferris'" <chris.ferris@east.sun.com>, ebxml
transport
>>>       <ebxml-transport@lists.ebxml.org>
>>> Subject:  RE: [Fwd: RE: on the issue of PartyId]
>>>
>>> While being able to say that PartyID is a URI might nice,
>>> why is this necessary?  Especially if we are trying to
>>> ensure that we can use all the different sorts of identifiers
>>> which have been mentioned in Chris' list (below, several
>>> messages down) plus telephone numbers?
>>>
>>> While it's not defined yet in the MS spec., the "context"
>>> will happily allow us to use any of these schemes as long as
>>> we define "context" appropriately.
>>>
>>> Best regards,
>>> Henry
>>> --------------------------------------------
>>> At 01:01 PM 09/21/2000 -0400, Brunner, Eric wrote:
>>> >I concure, and (wearing another hat) am part of the group(s) working on
>>> >making the e.164 numbering plan map to the dns, hence making "telephone
>>> >numbers" map to URIs.
>>> >
>>> >-----Original Message-----
>>> >From: Christopher Ferris [mailto:chris.ferris@east.sun.com]
>>> >Sent: Thursday, September 21, 2000 10:36 AM
>>> >To: ebxml transport
>>> >Subject: [Fwd: RE: on the issue of PartyId]
>>> >
>>> >
>>> >nor this one...
>>> >
>>> >-------- Original Message --------
>>> >Subject: RE: on the issue of PartyId
>>> >Date: Wed, 20 Sep 2000 22:41:53 -0700
>>> >From: "Burdett, David" <david.burdett@commerceone.com>
>>> >To: Christopher Ferris <chris.ferris@east.sun.com>
>>> >
>>> >The bottom line (almost) of Chris' email said ...
>>> >
>>> >>>>Thus, I would propose that we adopt use of a URI as the value of the
>>> >PartyId element.<<<
>>> >
>>> >I agree for the reasons Chris gave. The only caveat I would have is
>that,
>>> in
>>> >the EDI world, Telephone numbers *are* used to uniquely identify a
>>> business.
>>> >I therefore think we also need to support this format.
>>> >
>>> >Chris is right that telephone numbers can be re-allocated to new
>>> >individuals/businesses and therefore do not strictly meet the
>requirement
>>> >for a URN on their own. However a telephone number **at a point in
>time**
>>> is
>>> >unique. Open to alternative suggestions but what might work is to
define
>>> >PartyId with a default type of URI, but allow an alternative of a
>>> telephone
>>> >number with a recommendation that the value be considered in
conjunction
>>> >with the date/time that the message was created in order to assure
>>> >consistent results. In practice though, the probability of a clash is
>>low.
>>> >
>>> >David
>>> >
>>> >
>>> >
>>> >-----Original Message-----
>>> >From: Christopher Ferris [mailto:chris.ferris@east.sun.com]
>>> >Sent: Thursday, September 14, 2000 7:57 PM
>>> >To: ebxml transport
>>> >Subject: on the issue of PartyId
>>> >
>>> >
>>> >I scrounged the archives and came up with the following postings on the
>>> >topic:
>>> >
>>> >http://lists.ebxml.org/archives/ebxml-transport/200008/msg00073.html
>>> >http://lists.ebxml.org/archives/ebxml-transport/200008/msg00079.html
>>> >http://lists.ebxml.org/archives/ebxml-transport/200008/msg00084.html
>>> >http://lists.ebxml.org/archives/ebxml-transport/200008/msg00093.html
>>> >http://lists.ebxml.org/archives/ebxml-transport/200008/msg00095.html
>>> >http://lists.ebxml.org/archives/ebxml-transport/200008/msg00082.html
>>> >http://lists.ebxml.org/archives/ebxml-transport/200008/msg00098.html
>>> >http://lists.ebxml.org/archives/ebxml-transport/200008/msg00107.html
>>> >http://lists.ebxml.org/archives/ebxml-transport/200008/msg00097.html
>>> >http://lists.ebxml.org/archives/ebxml-transport/200008/msg00111.html
>>> >
>>> >I feel that these postings capture fairly well the discussion on
>>> >PartyId. I have omitted the zillion related postings on whether
>>> >to regrep or not to regrep on a message send/receipt as irrelevant
>>> >to the discussion.
>>> >
>>> >Dick's posting made a noble attempt to articulate the various
>>> >points:
>>> http://lists.ebxml.org/archives/ebxml-transport/200008/msg00098.html
>>> >
>>> >Dick wrote:
>>>
>>-------------------------------------------------------------------------
>>> >One requirement:
>>> >
>>> >Any two trading partners should be permitted to use non-standard
values
>>> for
>>> >context and authority. The specs must provide an extension mechanism to
>>> >allow
>>> >this type of customization.  For example MIME permits header extensions
>>> >through
>>> >the X-hhhhhhh option, X12 has the "ZZ" qualifier, etc. Consider a
>>possible
>>> >example, using Amazon:
>>> >
>>> ><PartyId context="X-Amazon" authority="userid">RJB9876</PartyId>
>>> >
>>> >This, I believe, would address David B's concern over having parties
>>> >pre-register with recognized "Name space management organizations".
>>>
>>-------------------------------------------------------------------------
>>> >
>>> >David Burdett's proposal:
>>> >http://lists.ebxml.org/archives/ebxml-transport/200008/msg00082.html
>>> >rings true with what I had been advocating (use of URI syntax and NO
>>> context
>>> >attribute
>>> >since the URI provides its own context. I could conceive of use of an
>>> >attribute
>>> >which allowed typing ala BizTalk (which uses xsd:type to declare the
>type
>>> of
>>> >the
>>> >value, but not necessarily the namespace) as this might be seen as
>useful
>>> in
>>> >parsing the value and would permit non-URI syntax to be used as the
>>> PartyId
>>> >value.
>>> >
>>> >I think that it is important that we understand (and agree upon) how
>this
>>> >header element will be used within the MessagingService (let us leave
>the
>>> >Application(Support) layer aside for the moment).
>>> >
>>> >       - routing: an MS might simply be serving as a "mailbox" server
>(as
>>> >many of the
>>> >       existing implementations handle the initial receipt of a
message)
>>> in
>>> >which
>>> >       case the "From" might be the "mailbox" identifier
>>> >(mailto:companyX.com)
>>> >       into which the MS should place the message received. It isn't
>>> >important
>>> >       that the MS "know" *who* companyX.com *is* (e.g. no reg/rep
>lookup
>>> >need be
>>> >       involved).  The MS might resolve the "From" id to an address
>>> >(mailbox, queue
>>> >       whatever) but all that is needed here is a mapping between the
>key
>>> >(PartyId)
>>> >       and some value (mailbox address).
>>> >
>>> >       - logging: an MS might be required to log all messages
processed,
>>> in
>>> >which
>>> >       case having the "To" and "From" logged (without having to look
>>> >anything else
>>> >       up in say a reg/rep or in a TPA store) along with other
>>interesting
>>> >bits
>>> >       which might be found in the headers would be useful. Again, it
is
>>> >probably
>>> >       not important that the log contain the formal company name
>>> (resolved
>>> >from
>>> >       the logical id) in the log file. This could be processed after
>the
>>> >fact
>>> >       if it was important that this information be known to someone
>>> >"reading" the
>>> >       log much the same as most web sites merely log the IP address
and
>>> >not the
>>> >       DNS resolved hostname.domainname (which is expensive and makes
>the
>>> >site
>>> >       less performant) because this can be resolved later when (and
IF)
>>> >the
>>> >       log entry needs to be analyzed.
>>> >
>>> >       - delivery: if we accept David's scenario (and no, I am not;-)
of
>>a
>>> >       TPA-less exchange, then the "To" might be used to map to some
>>> >physical
>>> >       address using some manner of I2L (URI to URL) resolution (THTTP?
>>> >DNS?). In fact,
>>> >       if the "logical" address were a URL (which is a URI) then no
>>lookup
>>> >would
>>> >       be required at all. Again, from the MS perspective, it is NOT
>>> >important
>>> >       that it "know" *who* the entity/party is. The implementation
>would
>>> >merely
>>> >       have a responsibility to know how to perform the I2L resolution.
>>If
>>> >the
>>> >       "key" (I) were unique (as is the case of any URI) then there is
>no
>>> >need
>>> >       to agree to which code set is used, just that it be unique so as
>>to
>>> >       enable the resolution service to map the URI to a URL. A simple
>>> >hashtable
>>> >       of key/value pairs could easily accomodate the mapping.
>>> >
>>> >       in the case where a TPA *is* present, then the "To" might be
used
>>> to
>>> >
>>> >       resolve the communication protocol specifics within the TPA
>itself
>>> >(map the
>>> >       PartyId of the Header to the PartyId of the TPA or to some other
>>> >Party
>>> >       identifier (possibly, the TPA should have something akin to what
>>> >UDDI
>>> >       has to associate a number of different identifiers with a
>>> >BusinessEntity)
>>> >       ...) Heck, the PartyId in the Header could be the *same* value
as
>>> >the
>>> >       PartyId in the TPA;-) e.g. tpa:partyid:12345.
>>> >
>>> >       - security: again, in the absense of a TPA, one might conceive
of
>>> >       use of the "From" PartyId used by the MS as a key to a
>credentials
>>> >registry which
>>> >       could contain the certificate (or passwd) which the receiving
>>party
>>> >uses
>>> >       to validate the signature (or passwd) with which the message was
>>> >signed.
>>> >       Of course, this information would be in the TPA (one would
>expect)
>>> >in which
>>> >       case the "From" might be used within the context of the TPA to
>>find
>>> >       the certificate/credentials.
>>> >
>>> >There may be other uses, but I for one cannot think of them (it's
>late;-)
>>> >
>>> >The bottom line is that the identifier must be unique within *some*
>>> >namespace
>>> >so that the MS can resolve it (somehow) to something useful in the MS
>>> >processing
>>> >context. It is not necessary (based upon the use cases above) that the
>>> *who*
>>> >be resolved from the registry which allocated the id (ISO whatever)
>>> >
>>> >Thus, I would propose that we adopt use of a URI as the value of the
>>> >PartyId element.
>>> >
>>> ><element ref="PartyId" type="uriReference"/>
>>> >
>>> >with examples such as those (except the urn:tel: example which is to my
>>> >understanding an invalid URN because it is not persistent) identified
>>> >in David's email.
>>> >
>>> >I do NOT believe that we need to specify a set of
>>> >code sets other than possibly as examples (NON-NORMATIVE) as to how
>>> >PartyId may be (should be?) used. That (I believe) should be left
>>> >to the parties to agree to (can you say TPA? I knew you could;-).
>>> >
>>> >Comments?
>>> >
>>> >Chris
>>> >
>>> >
>>> >
>>> >--
>>> >    _/_/_/_/ _/    _/ _/    _/ Christopher Ferris - Enterprise
Architect
>>> >   _/       _/    _/ _/_/  _/  Phone: 781-442-3063 or x23063
>>> >  _/_/_/_/ _/    _/ _/ _/ _/   Email: chris.ferris@East.Sun.COM
>>> >       _/ _/    _/ _/  _/_/    Sun Microsystems,  Mailstop: UBUR03-313
>>> >_/_/_/_/  _/_/_/  _/    _/     1 Network Drive Burlington, MA
01803-0903


[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [Elist Home]

Search: Match: Sort by:
Words: | Help


Powered by eList eXpress LLC