Minutes of the UPDF Working Group Meeting
September 11, 2000

Meeting Attendees
Mark VanderWiele	IBM
Geoff Soord		Software 2000
Norbert Schade		Oak Technology

Some months ago the UPDF group announced that it will concentrate on some
very technical work, mainly device font handling, which needs serious
expertise in details. So we expected to be a small group for some conferences and we
got it done as expected.
A side effect was that we competed with UPnP all the time, which did not
bother us too much, as long as we had this specific target of device font
handling. But that time is over now and we are happy not only to be able to
show the results in device font handling, but also in constraints.

Major items discussed
As there were no open font questions so far, the main section of the meeting
was constraints.
With Mark VanderWiele, Geoff Soord and Norbert Schade we were three people
living in driver development for more than a decade each. So we quickly
agreed on the problems to be solved and started working on the proposals
prepared before the meeting.
We eventually agreed on a solution that incorporates several key
requirements, which makes this a very competitive solution to others known in the
market.

* Simple constraints (filter a control, if the setting meets a certain
other condition) are as easy as in other concepts. All other features are
optional. This ensures a kind of backwords compatibility in philosophy.
* Complex constraints can be defined and read like in human language. This
should ease the understanding of the concept.
* The number of conditions to be combined to a constraint is practically
indefinite.
* Operations AND and OR are realized in constraints.
Complex conditions can be combined to one constraint, where it makes sense.
This leads to a more realistic list of constraints than an endless list of
simple constraints.
* Filtering is not the only action resolving a constraint. The format also
supports messages and automatic selections in other controls.
* Constraints can be defined global or operating system specific.

We think that a successful architecture of UPDF must be based on four
columns:
* Parameter Converter (done on the first level)
* Event Handlers
* Constraints (almost done on the first level)
* Predefined sets
With the current spec for constraints we think we can provide a very flat, 1-level model without loosing any flexibility. This is especially true for media handling. 

We are accompanying the DTD with a small specification document, explaining
the ideas behind the concept.
We verified the spec with samples in XML. Some samples will be shared with
the community.

The constraints section needs a few more days of polishing, before we will
share it with the community.
As IPP is short of constraints up to now, we would like that group to study
the spec, too. It is a separate DTD right now, very compact and short.

The remaining time was used to prepare a decision on the overall architecture and final file
structure. This needs further discussion. So far we agreed to only support
long file names.
You may want to watch the email traffic next weeks for that.

We have seen that especially driver developers from printer manufacturers and operating system vendors are contributing much value to the spec.
So we send a special invitation out to driver developers to contribute to the spec and attend the UPDF day at the conference.
This includes not only the complete Windows world, but the Mac and
upcoming concepts under Linux as well.