spice-to-ibis update

Here is an update on our (North Carolina State University) progress on the
next spice-to-ibis converter (s2ibis).

We are able to undertake this project
thanks to the generous support of Cadence.

I am putting AR items in the next paragraph so it does not get lost but it
may not make much sense until you have read the whole document.

AR We are using the golden parser header as distributed some 6 months ago. We
AR would very much like to use the current golden parser header file.
AR This would be a good topic for the next open forum meeting. The advantage
AR of course is maintainability and to ensure that we complete.
AR One set of code we are missing and have not started on is code to write an
AR *.ibs file from the ibis data structure. Has anyone written this?

We have defined a grammar for the s2ibis command input file.
We have also written a yacc/lex parser to read this grammar.
The Grammar is very similar to the ibis grammar.

The basic strategy is to parse an s2ibis input file; set up internal data
structures completely; and then initiate analyses. These analyses will be
parsed and the results used to set up the internal ibis data structures. Then
the *.ibs file will be generated from the data structure. This is a departure
from the way version 1 of s2ibis works. This was essentially an interpreter
than fired off analyses as the input file was parsed. The new scheme seems
to be the only reasonable way to allow a lot of flexibility, to provide
meaningful message reporting, and to provide relatively painless
future expansion.

We estimate that we have put 4 man-weeks in to the coding so far. We are
hoping to have a working prototype of the software by June 12. This will,
however have many of the desirable features missing but it will run.

The whole project is like writing a reverse golden parser.

Michael Steer
Received on Thu Jun 1 18:39:43 1995