Minutes from the IPP WG Meeting in IETF 47 - Tuesday March 28, 14:15 - 15:15 ============================================================================ Note taker: Lee Farrell Carl-Uno Manros, Chair, led the Internet Printing Protocol (IPP) WG session. Around 30 people attended. Carl-Uno listed the IPP/1.1 documents that are still under review by the IESG: - Model and Semantics - Encoding and Transport - Implementer's Guide He noted that although there have been some recent updates on these documents, the changes have been editorial in nature, the technical content has remained the same since the documents were submitted in June 1999. He also listed the new IPP Notification drafts: - IPP Event Notification Specification - The "mailto:" Notification Delivery Method - The "ipp" Notification Polling Method - The "INDP" Notification Delivery Method - IPP Notification Delivery Protocol (INDP protocol) - Notifications over SNMP Bob Herriot then presented some slides about issues using HTTP "chunking" with regard to notification messages. He first asked the group whether HTTP proxies support chunking: - for POST in requests? - for POST in responses? But none of the attendees could provide an answer. As a brief background, Bob explained that subscriptions for both printer and job events contain the notification recipient URL. The event notification is intended to be delivered to the specified recipient URL whenever the event occurs. Currently, there are four different delivery methods currently under consideration: - ipp-get: - indp: - mailto: - snmp-notify: In the ipp-get method, the Printer saves each event for a fixed amount of time. The client must poll for event notifications within this time interval. When polled, the Printer responds to the client with one or more event notifications and a recommended time (lease time) by which the client should next poll. Bob raised a few issues: - In order for the client to receive events as they occur, should there be an operation with a single HTTP POST, returning event notifications in multiple response-parts spread over a long duration of time (e.g., several minutes)? - Should this operation use HTTP GET instead of POST? Although there was some discussion of the first two items, several people in the group felt that the appropriate experts were not present to adequately address this question. Later, Keith Moore suggested, "Try it - it might work." However, it was noted that intermediaries on the network might prevent it from being a reliable method. - Should each response-part be a separate message body in MIME multi-part? Keith Moore suggested that this should not be done. He recommends that the group should "avoid the complication." - Do the lease and server recommended times make this polling mechanism a reasonable alternative to Printer-initiated event delivery methods (i.e. mailto, indp, and snmp-notify)? The last question caused Keith Moore to raise a concern that if multiple notification delivery methods are used within IPP, it is very likely that a user will experience widely differing behavior - depending on the method employed. He recommended that perhaps the group should concentrate on a single method of event notification. When asked which one, Keith suggested that ippget might have the best chance of working in most environments. He also suggested that ippget would require less code (no additional protocols), but might require more cache/memory - for storing events. This could be lessened by using shorter time intervals, causing fewer events to be saved at one time. Carl-Uno noted that the group consensus seems to be that support of ipp-get is favored as the baseline delivery method. The mailto method might be a secondary increment. Carl-Uno then referenced the Internet-Draft proposal for an LDAP schema for printer services: draft-ietf-ipp-ldap-printer-schema-00.txt. He noted that this document is aligned with the SLP printer template: draft-ietf-svrloc-printer-scheme-06.txt He listed the new Internet-Drafts on additional IPP operations and syntax: - Job and Printer Set Operations: draft-ietf-ipp-job-printer-set-ops-01.txt - The 'collection' attribute syntax: draft-ietf-ipp-collection-02.txt Carl-Uno referenced that at the previous IETF Plenary in November he led a BOF session for IPPExt. At that meeting, the group proposed additional future work for the IPP WG. As a result of that meeting, a proposed Charter was submitted to the IESG for an IPP Extensions Working Group. He said that he has not yet received any response about forming a new WG. The proposed Charter contains two areas of focus for future activity in the IPPExt WG. The proposed Milestone deliverables are: - New Operations - June 2000 - Notifications - September 2000 Before the meeting ended, Carl-Uno expressed his idea of making available "open source IPP client(s)" for full support of IPP. He mentioned that Microsoft is only supporting the bare minimum of the IPP feature set in their client. He would also like to see clients that work on other operating systems besides Windows. If anyone,or any company, is interested in working on this goal, please contact Carl-Uno. IPP meeting adjourned.