short comments on AM 2001-03-30

 Hello all. 8-)

 I have a few comments on the 2001-03-30 draft of the abstract
model. Generally I think it's great. 8-)

 If somebody else already noticed and posted what I'm posting
right now, please ignore this.

 In the document the very last line says that XMLP_UnitData and
XMLP_Intermediary were merged into one operation and the document
mostly corresponds to this statement, only a few places still
mention XMLP_Intermediary. Most notably section 3.1 lists the
four primitives of the XMLP_UnitData operation:
         XMLP_UnitData.send
         XMLP_UnitData.receive
         XMLP_UnitData.status
         XMLP_Intermediary.forward

 It seems that the last should be UnitData, too.

 Section 4.2 bullet number 6 mentions the id/href mechanism. I'd
rather call it id/idref mechanism unless we _explicitly_ want to
allow for general linking even to different documents. I don't
think that this general linking is needed and it does make life
harder for the implementors.

 Also section 4.2, bullets no. 4 and 7: I think that the special
actor URIs should be URNs - that would show the use of URNs and
it wouldn't connect using actors with the XMLP to the http:
scheme which may be confusing.

 So let's consider something like
  urn:org-w3c-xmlp:final
  urn:org-w3c-xmlp:none
  urn:org-w3c-xmlp:next

 The last comment is only minor: section 5.1.4.1 metions it
"assumes use of the HTTP Connection: keep-alive header" where it
would probably be better to say it "assumes using HTTP persistent
connections" because HTTP 1.1 does not have the Connection:
keep-alive header (I didn't find it in RFC 2616 anyway), it
actually assumes that unless otherwise specified (by Connection:
close) the connection is persistent.

 Best regards

                            Jacek Kopecky
                               Idoox

Received on Wednesday, 11 April 2001 10:44:09 UTC