Subject: [sv-cc] Meeting Minutes: SV-CC Weekly Meeting - 10-22-03
From: Stickley, John (john_stickley@mentorg.com)
Date: Wed Oct 22 2003 - 10:16:20 PDT
Attendees:
Joao Geada
Doug Warmke
Swapnajit Mitra
Ralph Duncan
Francoise Martinolle
Michael Rohleder
Ghassan Kouri
Avinash
John Stickley
Face-to-Face meeting
sv-bc meeting on Nov. 10 at ICCAD (Nov. 10-14)
sv-cc can schedule face to face that week
Swapnajit looking for suggestions on when to schedule
-- Please send date preferences
VCD donation
Need to organize acceptance
Basam has sent out .pdf file of latest draft of donation
Please everyone have a look and review
Swapnajit will send out ballot
We have 1 week to respond
Basam could not attend for Q/A
Will join us next week for extended Q/A
VPI Donation
Next week we will also have Q/A for VPI donation
Disables and DPI interaction
Michael sees still 2 disagreements between Doug & Joao
What is the right mechanism to notify DPI tasks about disables ?
Doug had sent out modified proposal
New proposal has no callbacks
Joao says we've ironed out differences and proposal looks pretty good
Action: Doug to do updates and put into PDF text in donation form
Michael: Return value should be 0 for normal 1 for disable, others reserved
Action: Doug to add this in to proposal
Michael: How do we enforce protocol by user ? Michael's proposal
suggests API call
Doug: says return code accomplishes the same thing but with simpler
protocol and less interface complication
Joao: Also agrees that new calls are not needed
Michael: Wants to make sure user catches the fact that disable has
occurred and that there's a clear way for user code to act upon it.
He needs to think a little more to make sure that will work
w/ Doug's proposal
Francoise: What happens to output values written before disables ?
JohnS/Doug: This is a general Verilog issue not just DPI
Doug: Should be up to user control
Michael: Output data modified after disable should not take effect
This should be described to user though not enforced
Swapnajit: Asked Francoise to clarify disagreement in DPI imported
task ballot
Francoise: Agrees now about how issues were resolved but still
questions the importance of the feature in general - has concerns
about implementation difficulty
JohnS: Pointed out prior outside input that it is an important
feature (SuperLog, Cliff Cummings)
Further discussions about justification for imported/exported
mechanism
Francoise concerned about the difficulty of allowing time passage
in C - should restrict all time consumption to HDL
Doug, Joao pointed out that that is essentially what we're doing
by saying the only way C can pass time is to call exported HDL
tasks - C itself does not pass time
Other Issues
Swapnajit: BC-77 issue from Vasilio - rang no bells with anyone
as to who submitted this
Joao: BC-77 - we're trying to find out who asked for it
Swapnajit: Wants to move to SV-AC
Joao: For any issues, we need to know source and justification - otherwise
should remove issue
Action: Joao end out e-mail about issue of behavior of output arguments
after disable
Joao: Would like everyone to make a point to read through Novas VCD
proposal for next weeks vote
Doug: Would like to propose that we don't require imported tasks to
be context - suggests that user should be explicit about this
Joao: Thinks that it is generally erroneous for an imported task not
to be context and user should be alerted - i.e. not having context
is likely a user mistake
Action: Doug to send out strawman proposal for how use of context
with imported tasks will be dealt with
-- johnS
__
______ | \
______________________/ \__ / \
\ H Dome ___/ |
John Stickley E | a __ ___/ / \____
Principal Engineer l | l | \ /
Verification Solutions Group | f | \/ ____
Mentor Graphics Corp. - MED C \ -- / /
17 E. Cedar Place a \ __/ / /
Ramsey, NJ 07446 p | / ___/
| / /
mailto:John_Stickley@mentor.com \ /
Phone: (201)818-2585 \ /
---------
This archive was generated by hypermail 2b28 : Wed Oct 22 2003 - 10:18:25 PDT