Thanks Lee.
Additional comments, clarification please forward to list.
Regards,
Greg LeClair
greg@erc.epson.com
===============================================================
Teleconference Participants
The attendees included :
Shigeru Ueda Canon
Akihiro Shimura Canon
Takashi Isoda Canon
Lee Farrell Canon Information Systems
Greg LeClair* Epson
Brian Batchelder Hewlett Packard
Alan Berkema Hewlett Packard
Greg Shue Hewlett Packard
Fumio Nagasaka Seiko Epson
Randy Turner Sharp
* 1394 PWG Chairman
Discussion
Nagasaka-san started the meeting by asking how Greg Shue will implement
his SBP-2 approach according to the January proposal. Shue said that his
proposal has not changed since then. (Neither has Alan Berkema made any
changes, although he expects to do so soon.) According to Shue, the
recent e-mail has only been a discussion contrasting the features of
different alternatives. Shue is anxious to avoid confusion between his
proposal and the e-mail discussions.
When Greg LeClair arrived, he attempted to set an agenda. The three
proposal alternatives currently under discussion were identified:
* 1284.4 over SBP-2 (there has not been much
discussion supporting this on the e-mail list, but Brian Batchelder says
he's not yet ready to abandon it completely. He said that most people
are agreeing that 1284.4 over SBP-2 is not very desirable because of the
significant feature overlap. Nagasaka-san says he still likes the use of
1284.4 because of its credit feature. Others are willing to consider
1284.4 over some datalink.)
* SBP-2 "native"
* DPP (Randy Turner wonders how heavy the session
layer will be for this.)
Greg LeClair concluded that today's discussion should cover technical
issues about SBP-2 vs. DPP. Brian suggested that he would like us to do
a trade-off analysis of SBP-2 and DPP at the next PWG meeting.
(Greg LeClair's comments - I felt it was more of a clarification of the technical issues
being discussed on SBP-2 and DPP {not SBP-2 vs. DPP} as these had been the hotter topics
on the reflector in the last few days.)
Both Greg LeClair and Alan Berkema think that Randy's topic of a Winsock
API on top of SBP-2 was interesting. However, Alan wants to see a more
detailed proposal.
Randy Turner suggested that we also consider the DPP command set over a
different transport layer.
Of the several proposals currently under consideration, Randy wanted to
identify an alternative that could be rejected. However, no one could
agree on one to eliminate. (At least one person was interested in each.)
Alan Berkema suggested that we have an e-mail poll to rank the different
alternatives being considered. Greg LeClair will organize this. As part
of the "ranking poll", he will also ask respondents to offer a few words
describing why they rank the alternatives as they do.
Alan asked for a scenario walk-through of how SBP-2 really works to be
given at our next 1394 PWG meeting.
Despite claims from George Chrysanthakopoulos (Microsoft), Brian is very
concerned that SBP-2 really won't work (as supported by Microsoft) for
printing. He is (at least) concerned about enumerating multiple logical
units as separate devices. Neither Brian nor Shue knows which is more
difficult: getting Microsoft to change an existing implementation of
SBP-2 or getting them to add a (new) different protocol.
Brian was confused as to why Nagasaka-san feels that crediting is so
important. Brian says that 1394 should be able to support the flow
control using something other than credit-based control. With SBP-2, if
each channel (logical unit service) has a different login, he believes
this should eliminate any problems that Nagasaka-san is worried about.
Randy Turner is concerned that our requirements-especially regarding the
enumeration of different logical units-may conflict with Microsoft's
plans for their implementation of SBP-2 support. He is very anxious that
we do not compromise our 1394 PWG requirements simply because of current
OS support limitations.
Nagasaka-san (Brian?) said that someone from Microsoft will join the next 1394
PWG meeting. (GL - I've confirmed that Hitoshi Sekine plans to attend)
What should we have on the Austin meeting agenda? Suggestions included
the following:
* Summary of e-mail polls for ranking the alternative proposals
* Comparison of SBP-2 and DPP vs. the requirements list
* Randy's ideas on creating a "best current practice" for layering above 1394
* Walk through scenarios on how a print session over SBP-2 really works:
login, print job, reverse channel communication, etc.
Start with real simple printer-basic example.
We need to work through the assumption of multiple logical units vs. single
device. Greg Shue volunteered do this.
* Multiple logical units and how they are used (may be part of scenario walk-through)
Teleconference adjourned.
Next 1394 PWG Teleconference
None was scheduled. (TBD after Austin - GL)