Phone Conference P1450.1 Working Doc Subgroup

Thurs, Mar 25, 2004, 10:00 am PST

 

Attendees:

Tony Taylor (chair)
Greg Maston (scribe)
John Cosley
Jim Teisher
Doug Sprague
Dan Fan
Peter Wohl (later)
Rohit Kapur

 

Documents


http://grouper.ieee.org/groups/1450/private/P1450.1-D18.pdf (ballot draft)

http://grouper.ieee.org/groups/1450/private/P1450.1-D19.pdf (updated 3/25)
http://grouper.ieee.org/groups/1450/private/P1450.1-D19-ballot-resolution.pdf (updated 3/25)

http://grouper.ieee.org/groups/1450/private/P1450.1-votes.pdf (ballot results)

subclause 13.2 and 13.3 (see D19, updated 3/25)
subclause 22 (see D19, updated 3/25)

clause 15 and Annex L (see D19, updated 3/25)
 

Agenda

1) expression sub-group - update on progress

2) parallel-pat-list tiling example
- subclause 13.3 - updated from last wg meeting
- DM-22Z - suggest to remove Extend from "PatList"

3) DM-22Y - subclause 13.2 - improved example of pattern burst statements

4) DM-22T - suggestion to change name of SignalVariable to WFCVariable

5) DM-22U - removal of Values on Integer variable definitions (p27)

6) STIL.4 - debrief of 2 day face-to-face (if wg is interested)

7) assignments and AIs for next week

8) subclause 22 (if time)
 


Meeting discussion

 

0. IEEE meeting clearances
Nothing under discussion or presentation for this meeting was identified as

being proprietary or restricted.

 

1. expression subgroup
Reviewed two of the last subgroup's decisions,

DM-2 which requested an understanding of why there are two assignment
operators. This opened up for discussion the entire issue of
differentiating variable-operations from signal-operations, which
resulted in yet some more proposals on how to handle
variable-operations, for instance, typing the LHS explicitly or
allowing variable expressions only in a special statement (like A
{}). All of these solutions result in the same issue: you need to
understand how to manipulate variables as distinct from signals in the
language. The subgroup proposed to expand on the explanation for the
current construct, but do not have a better solution to this problem.

DM-4 which identified confusion on the 'logical expression'. The
subgroup further identified that these expressions are ambiguous in
the STIL (.0/.1) context as well, and expanded on "proper expressions
for evaluation in the STIL context". For instance, the logical
expression 'A&B' is meant to represent a hardware/circuit behavior or
construct, and is NOT intended for evaluation.

Rohit identified that the primary application for these constructs is
CTL, in which the hardware context is being represented. But because
of the 'ScanEnable' statement in this standard (added in support of CTL
requirements), AND because we want to localize expression types to a
single dot, the definition is currently here.

A proposal was made to consider renaming 'logical expressions' to
'logic definition' or 'logic relation' - to remove the implication
that these constructs are meant to be interpreted outside of a circuit
or hardware context. The subgroup will consider this and augment the
current explanation to attempt to resolve this point.

Tony mentioned that Don Organ has strong interest in participating in
the expression subgroup, and Greg indicated that Don will be contacted
before the next meeting.

2. pattern tiling
The group reviewed fig. 1 in Tony's distribution of clause 13.3. After
some discussion about the use of zero-length patterns to provide a
default on unused signals at the start of a block of patterns (be it a
patlist, patset, parallelpatlist or what-have-you), and the
application of extend at the end of the patterns, the WG agreed to
accept this as-is.

A request was made to reword the points in subclause 13.3 b; patterns
may not all start at the same time...

The group then reviewed Tony's proposals to clause 13.2. A request was
made to clarify the behavior of 'if' and 'while' checks on
pattern-references inside the patternburst. The expectation is that
these clauses are evaluated *only* at the start of the
referenced pattern or burst that they are attached to (and not during

the  referenced pattern or burst). They do not change
execution within  this block if the condition changes after this point; for
instance you do not start that reference if a conditional suddenly
changes to true during another pattern execution, or you do not stop
execution of a pattern-reference if a 'while' changes to false, at
least not until the end of that pattern-reference execution (at which
point the decision to start another execution of that reference is
made).

CellGroups
It was hoped at this point that both Peter and Rohit would be present,
to discuss the requrement fo the ScanCellGroups block. Peter had two
issues in particular:

a) is this neccessary in dot1? - nothing uses or can use this data.
b) in separate block? - if needed, why not within scanstructures?

Denis Martin also raised issues with this block. Tony recommended a
subgroup of Peter, Rohit and Denis (and others if interested) be
formed to discuss and propose a solution to the WG on these issues.

4. DM-22T
The Working Group agreed to rejected the request to change the name
SignalVariables, adding the elaboration that this name was previously
used and already changed, and that the connection of SignalVariables
to Signals was seen to be a valuable association when the name was
changed previously.

5. DM-22U
There is confusion in the current spec, where the term 'Values' is
defined for variables but not presented in the syntax. This construct
was removed but the definition was not - Tony will clean this up.

Peter had two email issues reviewed and wanted to close on them:

6.1. Peter agrees with the rewording that Tony sent him for clause 15,
--- para. 3 to remove scangroup ref.

6.2. dm-22
---
A question was raised about the use of the term 'skewed load' (on pg
42 of Draft 19), that this term is never defined.

After some discussion and review between clause 15 and annex L, the
proposal was made to move the words introducing clause 15 into annex
L, where they are not a normative part of the spec, and to put some
generic words in the intro for clause 15 referencing this annex as
justification for the constructs added here.

7. assignments:
Tony added issues dm-5,dm-19,gr-4,jc-3&4 to the expression subgroup.
Tony asked Peter to review dm-21.

dot4 meeting review
Tony summarized the recent 2-day face-to-face meeting of the p1450.4
WG as: they've got a handle on the conceptual model for the flow
behavior, and are moving to syntax. As part of the syntax issues, they
are reviewing previous (1999) proposals, a 3rd-party environment
called OTPL, and Inovus Stylus. Currently the efforts are focusing on
presenting these 3 environments under a common context to make
relating between them understandable.

Meeting adjourned at 11:34 pm PST.
 


Next meeting

Thursday, April 1, 2004, 10:00 to 11:30AM, PST

 

AIs