1. Fail feedback - Doug Sprague

Doug reviewed his proposal, which is represented in the following code example:

as proposed:

ABIST-30 SO5 22 52 \r20 00 A \w L; // data capture @ cycle 22: HHLT ...80 L’s... TTH

as modified (after WG discussion):

"ABIST-30" SO5 22 {52 \r20 00 A \w L}; // data capture @ cycle 22: HHLT ...80 L’s... THL

issue #1 - The label "ABIST-30" was intended to be a reference to an X label and needs the double quotes around it. Note: Doug was not proposing an address offset of 30.

issue #2 - The string of characters representing the result data in hex is not possible to distinguish reliably from the cycle offset. It was decided that the result data be required to be in a braced block. The braces are required, even if only one result char. The braced block is optional.

issue #3 - Doug re-raised the issue of needing a way of specifying that a pattern was truncated. He would like to do this without having to modify the original pattern (as was proposed at the last ITC meeting). Tony brought up the point that in a burst (STIL.0) it is possible to execute a partial pattern by use of Start/Stop on a pattern or burst. The WG decided to allow the Start/Stop on a Pattern in a PatternFail block. The WG also decided to add a cycle count option. The syntax is as follows:

    Pattern PAT { Stop 5000; }   // stop after 5000 cycles from start of pattern

    Pattern PAT { Stop LAB1 4000; }   // stop after 4000 cycles from label LAB1

    Pattern PAT { Start LAB2; }   // start at LAB2

2. Cell Groups - Rohit Kapur, Peter Wohl

Since Rohit was not present this issue was deferred. However the WG re-emphasized the position that this feature should be dropped from STIL.1 if no one from dot6 presents a case for keeping it.

3. Pattern Tiling - Tony Taylor

The solution proposed last week was awkward for the following reason - There is a "Fixed" statement in the PatternBurst block, but to fix signal within a list, one was required to create a dummy pattern with only a "Fixed" statement in it. Tony proposed an alternate solution which used the burst level "Fixed" statement with revised semantics. New semantics are as follows:

- PatternBurst -> Fixed - is in effect from the point in the burst where it appears until the end of the burst or until another "Fixed" statement is issued.

- A new "Fixed" statement serves to un-do a prior "Fixed" statement.

- An empty "Fixed" statement means that no signals are fixed.

4. Process and progress - Tony Taylor

Tony reported the following statistics with regard to the set of ballot issues:

number of issues = 81, number resolved = 5 (6%)

number editorial comments = 219, number resolved = 31 (14%)

The WG is concerned at the time it is taking (and will take) to work through all these issues. Several proposals were made to speed up progress:

1) hold an all day face-to-face at DAC (june6..10). We will decide on this at the next meeting.

2) separate the big and little issues and then finish up the little issues (to get a better idea of the work required to finish)

3) WG should be more aggressive at making assignments

We discussed whether to continue have weekly phone meetings and the WG unanimously decided that we should.

Meeting adjourned at 11:00 AM, PST.