W3C home > Mailing lists > Public > public-ws-addressing@w3.org > November 2004

Re: IOR Questions

From: Tom Rutt <tom@coastin.com>
Date: Mon, 08 Nov 2004 14:08:30 -0500
Message-ID: <418FC42E.3000402@coastin.com>
To: David Orchard <dorchard@bea.com>
CC: public-ws-addressing@w3.org

Some answers inline

David Orchard wrote:

> To all the IOR experts,
>
> I've been reading through the IOP/IOR stuff, and I had some questions. 
> Hopefully you can answer..
>
> My understanding of IORs is that an IOR consists of a type value + 
> sequence of profiles. Profile consists of a tag + profile specific 
> data. The profile specific data contains version, host and port 
> information. The optional profile data consists of a buncha 
> components. These components are roughly policy information (security, 
> transaction) and address/identifier information (alternative IIOP 
> Address, object key, endpoint id position).
>
> 1. Why is it called a tagged profile? Is the XML analogy a "typed" 
> profile? Why not just call it a profile?
>
The profile contains multiple tagged components, which use well know 
integer tags, along with data haveing a format specific to that tag. 
This would be similar to having well know properties for an epr property 
list.

> 2. Is the multicomponent profile simply a way of saying that there's a 
> bunch of components/properties in the profile, ie multi-component 
> profile roughly equals EPR scope as it stands today?
>
yes, but for one protocol's access to that endpoint.

> 3. Why does the IOR spec say so much about how intermediaries deal 
> with the profile transmission? It seems like there are 2 sets of data 
> in an IOR, the "standard" which must always be forwarded and then the 
> extensions which might not be forwarded. And would you expect that 
> this IOR material would go away, that it would be best to forward all 
> the IOR data?
>
Vendors put their own stuff in these profiles sometimes, the tag space 
has vendor specfiic regions of integers. While
you can ignore them, getting rid of them risks one of their "aware" 
clients not being able to use these vendor specific components.

> 4. What is the relationship between corbaloc URIs and IORs? I think 
> that the corbaloc URI schema is a subset of the IOR in that it doesn't 
> serialize the profile data but provides a URI form for the 
> version/host/port info. But it could be a full serialization of the 
> IOR into string format using the IOR->CDR->String rules in 13.6.9 help?
>
corbaloc is a way to serialize an IOR as a URI.

> 5. what else am I missing in my cluelessness?
>
> Cheers,
>
> Dave
>

-- 
----------------------------------------------------
Tom Rutt	email: tom@coastin.com; trutt@us.fujitsu.com
Tel: +1 732 801 5744          Fax: +1 732 774 5133
Received on Monday, 8 November 2004 19:11:47 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 2 June 2009 18:34:59 GMT