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

 


Help: OASIS Mailing Lists Help | MarkMail Help

xri message

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


Subject: RE: Minutes for XRI TC Telecon Thursday 2/23 4-5:30PM USA / Friday morning Asia


[Apologies - the previous minutes were incomplete before I hit the "send"
button. These minutes are complete.]

Following are the minutes for the 2006-02-23 unofficial XRI TC telecom.

PRESENT

Gabe
Drummond
Les
Wil
Steve
Victor


AGENDA

We reviewed the following three lists (technical open issues, editorial open
issues, and to-dos) pertaining to XRI Resolution 2.0 Working Draft 10. All
line references are to ED 06:
	
http://www.oasis-open.org/committees/download.php/16853/xri-resolution-v2.0-
wd-10-ed-06.doc

Key:
* = discussion
# = decision
### - action item (look for your name!)


TECHNICAL OPEN ISSUES

### Drummond is responsible for all updates in this section unless otherwise
marked. 

1) Line 238 - (Victor) How should we show XML namespaces in the example XRDS
document? And what should we recommend about how the default namespace
should be represented?

* We agreed that we cannot control this issue; it is an XML processing
issue. 

# We will add the following text: "It is a recommended best practice for XRI
authorities to use XRD as the default namespace for XRDS document in order
to promote interoperability. However implementers are encouraged to use XML
parsing tools that support XML namespaces to avoid problems."

2) Line 655 - Should we allow only the versioned type value?

# We will use only the versioned value because we don't need a generic
(non-versioned) option until we have more than one version.

3) Line 906 - (Drummond) Should there be a separate error code than 206
REF_NOT_FOUND if the Ref is not found during service endpoint selection vs.
authority resolution?

* We had a long discussion about error processing and error messages.

# It was agreed that we should drop the the reference processing errors
CYCLIC_REF and 205 REF_NOT_FOUND and replace them with AUTH_RES_NOT_FOUND
(new) and SERVICE_NOT_FOUND (existing).

# We will drop 101 SUCCESS_NFR as we don't really need to inform the
consuming application that further reference processing is necessary.

# Change 203 TOO_MANY_REFS to LIMIT_EXCEEDED.

# Provide examples of the types of error message strings might accompany
LIMIT_EXCEEDED (too many XRDs, too many refs, too large an XRDS document)

# Add another 3xx error code for TIMEOUT.

### Will to send Drummond an updated error table reflecting above changes
and renumbering error codes as he recommends after receiving these minutes.


4) Line 913 - (Steve) Should the output of SEP selection also include a list
of CanonicalIDs? If so, how would this be returned in proxy res?

# Yes, output of SEP selection should be a list of both URIs and
CanonicalIDs. 

* We had a lengthy discussion about how to handle this type of return in
proxy resolution. The leading suggestion was to make it a separate media
type. One proposal was to use XRD (not XRDS) as the XML format since we
wouldn't have to define a new return. This XRD would include the
CanonicalID(s) and URIs in the selected Service.

### Drummond to send concrete proposal to the list.

5) Line 935 - (Steve) Are we okay with use of "only" value of match
attribute?

* Long discussion of the fact that the "only" value is an exception to the
rule that all values of match attribute only apply to what should be matched
for that element. The leading proposal was to add another attribute
(possibly Boolean) called "select" that controlled whether the parent
element should be automatically selected if there was a match.

### Drummond or Steve to send a concrete proposal to the list.

6) Line 967 - (Gabe) Are we okay with path match being case insensitive?

# Yes, given that we can be more inclusive with regards to comparison in
this domain than for XRI comparision itself.

### Drummond to add text clarifying what "case insensitive" means precisely
(including escaped hex strings).

7) Line 1035 - (Gabe) How do we want to manage a list of second-level "xri."
Domains?

# Keep this as open issue for the time being.

8) Line 1119 - (Gabe) Is this logic to prevent circular Refs correct?

# We will eliminate this in favor of the LIMIT_EXCEEDED error.

### Drummond to update in text and flowchart.

9) Line 1130 - (Gabe and Steve) Do we agree that standard SEP selection
should be applied when doing Ref processing to locate an authority
resolution service? Should we require that the type explicitly be
"xri://$res*auth*($v*2.0)"?

* We had a long discussion concluding that, for simplicity...

# Local part of any Ref processed for authority resolution will be ignored.

### Drummond to update text.

10) Line 1208 - (Steve) Should we drop error code 202 CYCLIC_REF?

# Yes.

11) Line 1223+ - Are the error instructions in this section correct?

# The first bullet is correct. With the second bullet, anything but an HTTP
302 should be recognized as an XRI resolution error, however to distinguish
that from an HTTP error, the return should be a 2xx that indicates what the
error is using plain text or HTML or whatever other content type is
appropriate.

12) Line 1316 - (Gabe) Confirm that extension architecture still works and
that the XML namespaces are handled properly.

### Gabe to check this section and update text as needed.

13) Appendix A - Is there an xml:idref attribute? Should we just use
xrd:idref?

* We didn't have time to discuss.

### Drummond to send email to the list.


EDITORIAL OPEN ISSUES

1) Line 27 - Are we going to have a separate Implementer's Guide and who is
going to write it? If not, or even if so, how and where should we show
examples of authority resolution and service endpoint selection?

# No, we are going to include examples in appendicies.

### Drummond to add the appendicies.

2) (moved to #1)

3) Line 146 - Where should we put XRI resolution architecture overview and
how should we represent the two phases, i.e., the fact that service endpoint
selection does not actually involve any further network calls?

# In Introduction section.

### Drummond to do a first pass.

4) Line 296 - Should we only use @attribute notation in XPath? And is it
@ns:attribute or ns:@attribute?

* Did not have time to discuss.

5) Lines 317+ - Clarify that LocalID, CanonicalID, and Ref are all types of
synonyms.

### Drummond to add text.

6) Line 1158 - same issue as Line 238 in terms of formatting of XRDS example
wrt XML namespaces.

# Same conclusion as above.

### Drummond to update.
 
7) Line 1210 - Do we need section 9.2 and is Steve going to write it?

### Steve will draft text for this section.

8) Appendix A - Who is going to produce and who will cross-check updated XML
schema?

### Drummond to do first pass, followed by Gabe checking it for his
extensibility architecture, followed by Les/Wil/Steve checking Gabe's
version.

9) Appendix with an example IDL for an XRI resolver in a generic interface
definition language notation.

### Steve to send an email to the list with a proposal for this appendix
and, pending any input, do a first draft.


TO DOS

### The following are all for Drummond on the next pass. In addition he will
incorporate Gabe's other suggested edits from his 2006/02/22 email to the
list.

1) Line 238 - Fix wording to make sure section 3.2 is normative but clarify
that normative schema is Appendix A.

2) Line 451 - Move section 3.3.4 to section 6.3.1.

3) Lines 520-528 - Move to architecture overview section.

4) Update flowcharts to reference correct error numbers/ranges.

5) Add info about state to flowcharts to make it clear what it being passed.

6) Line 810 - clarify the value of the Name attribute of the SAML Attribute
element.

7) Line 1061 - Need to update for new TOO_MANY_REF error.

8) Line 1285 - Reword per Gabe's comment.

9) Line 1332 - Add para explaining need for dsig to still span elements.

10) Line 1369 - Add credit for language to SAML spec.

11) Line 1397 - Update about DNS security per Gabe's comment.

12) Fix prefix of XRD element in XRDS examples.





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