Phone Conference P1450.1 Working Doc Subgroup
Thurs, June 17, 2004, 10:00 am PDT
Attendees:
Tony Taylor (chair)
Daniel Fan
Greg Maston (scribe)
Bruce Kaufmann
John Cosley
Jose Santiago
Julie Villar
Peter Wohl
Documents
Agenda
Meeting discussion
IEEE meeting clearances
Nothing under discussion or presentation for this meeting
was identified as being proprietary or restricted.
Table 10:
backslash-operator interactions. Added BNF-style definition
in table to clarify interactions.
Some discussion about /S (or a question about /s). /S pulls the "last
value present from a previous S or s event", but there is no escaped
lowercase-s
Group accepted this addition to table 10 without additional comments.
Loop Data, and Shift
- intent: Shift only supports signals with parameters that have Scan
attributes, and supports padding behavior.
Loop Data consumes data, if data expires (on shorter data sets) then
the last specified WFC persists with no other padding behavior.
John identified a clause in rule 3 that indicates Shift may have a
dependency on Scan Length values, which needs clarification or
elaboration on the circumstances of applications. This application is
under LockStep only, John would like additional words to clarify when
the length is applied.
Open for the moment for complete review.
Annex N
Reviewed Bruce's addition. Issue is if there are Vectors after the
Shift that are part of the chain, there is no way to clearly identify
the end of the actual shift operation. Optional X-end statement to
identify end of actual Shift.
Question about whether X statements attach to Vectors, like
labels. Current definition is that the X statement sits between
statements, but the first Vector after an X statement counts as zero
offset from this X statement.
Discussion about whether the Load_Unload_End X statement, occurring at
the end of a procedure, would ever be applied, since the return from
the procedure would cause the upper-level X statements to be visible
again. In this context (the X statement being the last statement in a
procedure), it is doubtful that this statement would be referenced in
failure information, but again the purpose is to identify the end of
the scan operation.
RK-5: issue raised about specific applications interpreting the data;
we have a mechanism to log the data, but not interpret the data. No
change proposed for this comment.
Expressions Issues
Properly raised by the chairperson as an open action item that needs
resolution to complete several issues.
Specific Issues
RK-1: request to eliminate multiple inheritance in the Environment. no
change to what we do; even a single statement still supports an
ordered nesting of Environment that appears unnatural.
RK-4: no change.
RK-6 / Annex F: no change at the moment; haven't identified an example
to change to.
DM: Rules for If, Else, Goto interactions:
After discussion, identified:
- need some rules
- need some examples
for instance, how do these constructs interact; can Goto's jump into
If's and Elses, can they jump out of Whiles? What rules *should* be in
place and what are the consequences of these rules (hence the issues
of how to do certain things - for instance, terminating loops "early").
PW-DM: reviewing complex scan cell descriptions, and implications of
the descriptions in specific contexts. Discussion started, Peter
understands Denis' perspectives.
Meeting ended at 11:30.
Next meeting
Thursday, Junly 1, 2004, 10:00 to 11:30AM, PDT
AIs
Greg - Review and improve clause-6 - expression syntax; get NO-voters buy in
Tony - work with Rohit to resolve the timing expression issue in dot6
Peter - DM-5 "What is the nature of STATE_ELEMENT?"
Tony - Update doc per the above decisions