V/I table checking (from Kellee)

From: Kellee

I see my hopes of a fast update have faded into the end of the week (oh well).

I have to agree that it is far better to take the time and do it correctly
the first time.

On the other hand if we ended up with numerous IBIS models with incorrect tables
that weren't done correctly the first time we have also failed.

1) Relative to the ECL problem. This updated is intended only to address Ver 1.1
so it seems like this is not a problem. We could however check for the ECL
parameter and check for both tables have the same sign direction in this update.
This would get us a head start on the Ver2.0 parser.

2) Relative to several comments about only checking the signs at the two end
points and some devices having only negative or only positive signs. This doesn't
matter, what I proposed is testing if the values were larger or smaller. The
sign is not needed for this test. (-5 is less than -2) (0 is less than 2). Did
I miss something.

3) The IBIS_CHK program is not the specification. It only supports a portion
of the full IBIS 1.1 specification information. What we agreed to is that the
IBIS_CHK program would be the golden parser. That means when there is a question
about who's parser is correct the IBIS_CHK parser is always the one that is correct.

I am willing to make this into a BIRD and add the ECL, or just continue with my
proposal as is. Which way should we go?

I would like to continue as it is and just support version 1.1

I'll check in Sunday and start a BIRD if that is really needed. I still believe
that the IBIS_CHK program is simply the validation of the existing specification
and thus doesn't require a bird. It seems like the need for a BIRD indicates that
we didn't specify this area well enough and that is what the BIRD would need to
address. If we do start requiring a BIRD for each parser change than how in the
world will we ever get the parser done for version 2.0. Do we need a BIRD for each
and every item to be updated in the parser (it could get very lengthy). It seems
to me we just need to agree that a new version is acceptable to everyone and release
it. A set of regression IBIS files seems like a good idea though. How about using
the 82430 and 82420 files as the first set of regression files. Each IBIS
representative should also test the new update with their own IBIS files before
we re-certify the parser. What do you think?

Have a great weekend! (I will am going snow shoeing with my wife on Friday)

Kellee
Received on Thu Mar 17 23:43:58 1994