Phone Conference P1450.1 Working Doc Subgroup
Thurs September 25, 10:00 am PDT
Attendees:
Tony Taylor (chair)
Daniel Fan
Greg Maston (scribe)
Peter Wohl
Bruce Kaufman
Jim Teischer
Jason Doage
Doug Sprague
Documents
http://grouper.ieee.org/groups/1450/dot1/minutes-2003-09-11.html (minutes of last meeting)
http://grouper.ieee.org/groups/1450/private/p1450.1-D16.pdf (latest working draft)
Agenda
1. new clause 6 - feedback and comment?
2. clause 10 - Variables - need "HoldValue;" statement
3. clause 13 - add "Else" stmt?
4. clause 19 - remove (LABEL:) references - not needed
5. clause 22 - FailReport
a) start at 0 or 1?
b) Tom's issue about tests run
c) Jim's issue about fail addr vs. count
d) need example of X-placement for loops and calls
6. ITC monday @ 1:00pm
a) final issues
b) ballot decision
c) ballot list
Meeting discussion
0. IEEE meeting clearances
Nothing under discussion or presentation for this meeting was
identified as being proprietary or restricted.
6c: ballot list
Tony and Greg will assemble a list of current reviewers, WG members, and
interested parties (including the p1450 reflector members) to be ready
at the ITC meeting next week.
clause 6:
Tony identified that this was a re-write of the current clause 6 to
address several review concerns about clarity. The group was requested
to do a thorough review of this and we would discuss it at the ITC
meeting.
clause 10:
Tony has a concern that there is not currently a strategy to cause an
integer variable value to persist between blocks, as it is reset to
the InitialValue value each time the variable comes into
scope. Therefore there is a proposal to define a 'HoldValue' attribute
on the variable. This discussion was tabled to the ITC meeting.
clause 13:
Question: why is there an If construct in the patternburst, but no
Else? This is inconsistent with other occurrences of the If.
Greg responded that the If statement in the patternbust is not a
flow-control construct as much as an 'attribute' on the selected
pattern or burst reference. The syntax in this block is constructed to
perform selective execution of bursts, but at the individual burst
level and not in a generalized 'control flow' structure. This was
defined this way because selective execution of bursts was seen as a
requirement for this standard, but generalized control-flow is a
separate dot standard, and attempting to do robust control-flow here
will interfere with other efforts (and is in excess of the requirements
here). No change to the current syntax. Tony agreed to improve
the explanation of the If statement needs clarify this point.
clause 19:
Question about the presence of the (label:) constructs on all
statements here. Perhaps it is not necessary to put that label
statement here.
Greg identified that someplace in this standard needs to identify that
the label statement is present on all statements, and placing it here
makes that issue unambiguous. Tony will add the (label:) as an option
to the X statement as well [AI1] as Jason stated it was perhaps
not just uniform, but necessary.
clause 22:
The Working Group agreed that the fail offset count starts at
zero. Zero specifically indicates a failure AT the label, which is a
'label+0' position.
Other issues will be held to the Monday ITC meeting.
Greg agreed to generate some examples on x-ref statements before and
inside loops to identify the *expected* behavior when x-refs are
placed at these locations [AI2].
clause 3.2
Doug identified extra acronyms in this section to be removed now that
BistStructures was no longer part of this draft ([AI3] to Tony).
LockStep discussion with P1450.6 working group
Tony reported that he has had discussions with Rohit with regard to the
use of LockStep in CTL. The concept being adopted in CTL is to have
a common macro that is called by multiple patterns in "lockstep". The
macro then explicitly references the data from each of the calling
patterns and applies it appropriately, thus allowing complete reuse
of the pattern data with no modification.
At this time, no additional syntax is required in 1450.1 to support this.
The definitions in 1450.6 will be something like the following (exact
details will come from the 1450.6 wg):
- a top level block like LockStepMacros {}to contain these special macros.
- variable domains used to identify each pattern's signal variables
- dereference the pattern variable in the lockstep macro by:
V { signals = \Wdom1::varname \Wdom2::varname; }
Next meeting
AT ITC, Monday, Sept 29, room 204, 1:00 pm.
Meeting adjourned at 10:40 pm PDT.
AIs
new
old
9/11 - Jim - present vector-index referencing as part of fail-location data.
8/14 - Greg augment section 18.2 to contain syntax and semantics.