RE: JTP: Feature List Priorities
Some thoughts:
1. I wonder if JTP is a bit of a misnomer: test patterns are used for
various purposes e.g. OMA, extinction ratio, EMI, risetime,... Can we be
more general, and yet distinguish this from XAUI test patterns, with "STP"
for Serial Test Patterns?
2. As Jonathan and I have pointed out, we are unlikely to find the best
ever test patterns, even if we find the best yet. My take on this is,
rather than demand compulsory programmability for, or even existence of, a
feature which is likely to be used "just once" for acceptance tests between
supplier and customer, and never used out in the field, is that we recommend
good practice but leave the whole thing optional so that industry practice
can evolve to suit differing circumstances in the future. After all, we
don't strictly _need_ any special test patterns at all to make serial links
that work robustly (this statement may be specific to the scrambled line
code, because signal and idle are randomised to have similar properties to
each other). Special test patterns may be nice or convenient or allow us to
avoid some over-engineering in the analogue domain (just as long as we avoid
it in the digital domain) - but they aren't essential to building robust
Ethernet.
3. Should we treat the LAN and WAN separately, with any WAN test
patterns created in the WIS rather than the PCS, or consider creating test
patterns for the WAN in the PCS?
4. > How does Tuesday, 3-Apr, at 1:30 PST sound?
Pretty grim! I hoped you would take note of Gareth Edwards's email so I
didn't repeat it - I'm in the UK too. But "prime time" (5-6pm in much of
Europe, 4-6pm in UK, 8-10am in Pacific timezone) is often cluttered...
Piers