LRM Committee Meeting Minutes - 21st March 2006

Attendees:

* Geoffrey Coram
* Jim Barby
* Peter Liebmann
* Patrick O'Halloran
* Marq Kole
* Jonathan David
* Martin O'leary
* Sri Chandra
* David Miller
* Missing name

Current Status:
   - LRM2.3 merged syntax is available through the Accellera webpage. 
This has been merged with IEEE 1364-2005
   - An initial review was done few months ago when graham completed an 
initial version of the merged syntax. Need to do a full formal review
   - Couple of pending items on syntax for compiler directives and 
system tasks/functions - take the 1364 approach of explaining syntax in 
the individual sections or expand the BNF to make it complete and formal
   - Work has also been done on merging the lexical sections; probably 
this has not yet been posted to the reflector.
   - No work on the individual sections/chapters have yet happened. 
People assignements for the various chapters was done end of last year 
once the Mantis items addressed for 2.3 was finalized
   - The individual chapter updates include reflecting the new syntax in 
the individual sections and its very likely that lot of semantics would 
have to be documented in the individual chapters. Since the merged 
syntax allows all possibilities the semantics need to be detailed and 
also existing semantics for the current version is not complete to do a 
consistent implementation of Verilog-AMS
   - Need to ensure that editing is done in frame6 and assigned owners 
have an editable version
   - Plan to use 1364-2005 by reference rather than editing the IEEE 
version for LRM2.3 to have the consistency between LRM2.3 and 1364-2005 LRMs


Where to from here:
   - Current deadline for LRM2.3 was scheduled for June/July 2006. This 
is going to be very difficult with the amount of work on updating the 
chapters and the review process that need to be done for all the changes
   - LRM2.3 completion may take more realistically towards end of the year.
   - Briefly discussed possibility of working on SV integration 
parallely but currently there is no bandwidth to take up this work in 
the current committee
   - The current plan of 1364-2005 integration seems agreeable to 
proceed  with, tho' we are currently behind schedule
   - Some of the mantis items for quick fix and corrections and these 
already have proposals as part of the mantis items that could be 
ratified and approved to close some of the items for 2.3. Geoffrey has 
already sent an initial list with regards to this.
   - Sri needs to discuss with Accellera chair (Johny Srouji) to update 
with regards to current status and plans
   - Individual chapter owners to send rough timeframes of when the 
individual sections will be available for review. Probably it may be 
worthwhile not to focus too much effort on documenting all semantics 
before review, and this can be updated/documented as part of review process?

* New issues:
   - Discipline vs Signal name conflicts: usage of Q in std cells for 
flip-flops conflicts with the charge value. Use of namespaces as 
detailed in section 3.10.2 can be done to resolve this conflict?
   - Pass-By-Reference vs Copy-in-Copy-out: Currently AMS supports only 
CICO rather than pass-by-reference. This works equivalent for most 
cases, however when arguments are repeated in the arglist and set inside 
the UDF this can lead to indertiminate values as pointed by David. 
Subsequent to the meeting Geoffrey has posted an example.
   - Valid numeric suffixes: Need to come to an agreement what is 
required. Probably will be done as SV integration and not for LRM2.3 
(Jonathan needs a mantis login to be able to log the item)

Next call: Will meet on weekly basis. Next call will be same time, 
however need to reset the time once the different cities go in/out of 
daylight savings.

Regards,
Sri


-- 
Srikanth Chandrasekaran
DTO, Tools Group
Freescale Semiconductors Inc.
Ph: +61-(0)8-8168 3592 Fax: x3201