IPP>PRO minutes of 4/28 IPP protocol teleconference

IPP>PRO minutes of 4/28 IPP protocol teleconference

IPP>PRO minutes of 4/28 IPP protocol teleconference

Robert Herriot Robert.Herriot at Eng.Sun.COM
Tue Apr 29 17:38:49 EDT 1997


The following people attended the IPP protocol meeting of 4/28


Roger deBry
Robert Herriot
Scott Isaacson
Jay Martin
Randy Turner
Peter Zehler
Steve Zilles


Most of the meeting was spent discussing the protocol proposal from 
Robert Herriot


There was a discussion about using a MIME type for the document-format
attribute. Steve Zilles will contact the owners of document-format types
in the Printer MIB and tell them to register the document type as a
MIME type if they wish.


During the discussion of headers, Steve suggested that we consider 
whether we should use HTTP rather than invent a new protocol because
of the similarity of the proposed IPP protocol to HTTP.  We have
discussed this issue before.  This time, Randy will produce a document
which compares the Herriot IPP protocol proposal with HTTP origin server
so we can see how close they are.


We discussed how MIME differs from the HTTP MIME-like protocol. There
was a concern that the HTTP version of MIME doesn't support 
Content-Transfer-Encoding, though we think that we probably could add
such an entity-header as an extension and support it through a CGI
script if necessary (though we aren't sure about this).  There was
also a question about how to send binary data in a multipart/mixed,
especially in the chunked case because there is no way to know if 
a CRLF in the midst of binary data is really a CRLF. Thus it is hard
to find the boundary string. We believe that chunked applies to the
entire multipart/mixed entity and cannot be used for one of the sub-entities
alone.  Thus there is no length to mark the boundary of a sub-entity.


In our discussion of application/IPP, Steve suggested that the
syntax for the values conform to the MIME rules to keep parsing simple.
Bob will check the syntax agains the MIME rules to see if there are
any problems.


Randy suggested a change to the GetJobs operation. After much discussion,
we decided that the operation should return the list of JobURLs and
the ordinal position of each in the queue.  The user and state parameter
would remain as filters for determining which JobURLs to return, but the 
result would no longer contain requested attributes for each job in
the queue. We didn't discuss the printer object returned in GetJobs, but 
presumbably we meant to remove it too.  Scott Isaacson will update the model 
document.



More information about the Ipp mailing list