3/22/2002
STIL.4 phone conference
Attendees:
Dave Dowding
Stuart Nelson
Jim O'Reilly
Tim Micek
Dan
Fan
Jim Mosley
Tony Taylor
Ernie Wahl
Doug Sprague
Dave has coordinated with the IEEE for a P1450.4 reflector. Will announce soon.
Group emails since last
meeting:
Ernie Wahl - regarding state machine
Gordon
Robinson's thoughts
Don Organ's follow-up
Jim Mosley -
regarding binning
Tom Micek - concerns regarding binning
Observations and general discussion
Ernie: some
initial information was top-down, others was bottom-up
Dave: we have 4 slices
of examples, we can soon start in on a data model
Jim O'Reilly: the 1450.4
draft of '99 concentrates on Flow - a broader perspecitve than
what
we've discussed so far - we've so far discussed only 2 of 7
items
Doug Sprague: didn't
get copied on the emails.
Tony Taylor: a
convention - use STIL.4 at the beginning of the subject line to help people
filter
Jim - may need some
terminology
Ernie - such as
soft-binning and hard-binning and terminal bins
Tom Micek - what is
the group's opinion? - should the binmap be separate?
Ernie - my
experience is that you have many more software bins than hardware bins. I think
we should concentrate primarily on the soft-bins, but am in favor of hard-bins
as well.
Tom Micek - the
binmap (from soft to hard bin) is currently external.
Ernie - we can allow
a map - and where there is no map, that could be a default. Would like to allow
a graphical interpretation in addition to the ASCII text that we will
define.
Tom Micek - the test
programmer hands the program off to others - who may need to be able to change
the map (but don't want them to change the program)
Dave - I think this
gets us back to the need for a definition of terms
Jim - what did Tom
mean when he said he wanted the binning as part of the test? In my experience it
is always the FlowNode instance that has the binning information associated with
it.
Tom - within each
test (same as FlowNode) - defines the soft-bin as well as the
hard-bin
Jim - for a probe
program vs final test - may have the same soft-bins but would want
different hard-bins. Having it as a separate entity (block) is probably pretty
important. Could be in the same file - or separate from the main test
program.
Tom - test engineers
want it separate, product engineers would like it in the FlowNode (for clarity).
I'll talk to some of my internal users to see what their opinions
are.
Dave - test options
could be used to indicate which bin-map to use
Tom - how is
agreement reached?
Tony - a consensus
process (everybody has to compromise). Must win approval from a larger ballot
group. Also, if things are too contentious then maybe its not appropriate to be
part of the standard.
Ernie - we need to
break the problem into the right types of pieces. Binning (soft and hard),
Tests, Test parameters, Test FlowNode
Dave - and Test
Method. We may have some time pressues (may 12 -16 months until we have
something to propose).
Tony - I think you
should have goal of being ballot ready in a year
Tom - how do we
prioritize everybodies concerns?
Jim - one way
to speed up the definition process - is to develop categories of topics
(binning, flow, context for equations, how to pass parameters to
tests)
Dave - I'll take the
action item to put out a first list.
Jim O'Reilly -
there's good material on pages 2-5 of the 1450.4 draft of
'99
Dave - perhaps we
can start with some prioritization at the next meeting
so for the next
couple of weeks we're trying to get some initial definitions and some idea of
the scope. We should probably be declarative versus procedural - raises the question of
how much to standardize - versus how much to leave as an implementation
consideration.
Next call is on
April 5th.