Verilog-AMS committee meeting minutes - 4 Oct 2005

Verilog-AMS Committee Meeting Minutes

Attendees:
Marq Kole, Philips
Jim Barby, Univ of Waterloo
Patrick O'Halloran, Tiburon
Martin O'leary, Cadence
Geoffrey Coram, Analog Devices
Kevin Cameron, Sonics inc
Graham Helwig, Freescale Semi
Sri Chandra, Freescale Semi
[Apologies if i have missed somebody on the list]


* Discussions were mainly on the draft version of 1364-2005-AMS that 
graham had posted on the reflector
* This was just to give an overview of the BNF integration and also 
gives an idea as to the effort required to update individual sections of 
the document to be consistent with 2005.

* The committee will go back and prioritize rest of the open tickets 
(that havent been discussed) to make decision on any of the tickets are 
going to be addressed for LRM2.3 before making too much progress on the 
changes to the documentation. Many of the items in Mantis database 
relate to the BNF is expected to be addressed by this work.

* Graham gave an overview of the changes and the work he has done to 
merge the digital and analog BNFs. Discussion on individual sections 
didnt go into finer points/details to resolve the issues raised. The 
discussions were more to highlight some of the key points that need to 
be addressed when the work is being done and reviewed by the committee:

   - Keywords section
     -- Discussion on how to publish new keywords? How should the 
distincition be made between the existing keywords and the new keywords 
that have been added into the language?
     -- Italicising might conflict with 2005 BNF possibly, as it might 
have been used for comments
     -- underline?

   - Module items (A.1.4)
     -- Need for multiple analog blocks?
     -- The generate construct in digital is at module scope and allows 
for multiple initial and always block. Currently analog restricts analog 
blocks to be just one.
     -- Is there a need for this restriction to apply and deviate from 
the merged BNF for analog?

   - Paramset declaration (A.1.9)
     -- can the statements inside the paramset access internal variables 
of modules. Geoffrey clarified that the intent was not to be able to 
access internal variables.

   - Module parameter declarations (A.2.1.1)
     -- The merged BNF should resolve the conflict in parameter arrays 
with systemVerilog
     -- SystemVerilog has introduced `{} syntax to distinguish array 
literals and vector parameters
     -- this needs to be incorporated in the merged BNF
     -- also need to look at string datatype in SV BNF to ensure 
consistency with systemverilog language.

   - Timing Control (A.6.5)
     -- merged the event control statements of analog and digital
     -- usage of some of the events need to be semantically restricted 
(for example: @(net) restricting of net to a digital net).

   - Task Enable Statement (A.6.9)
     -- Lot of discussions on how system task statements should be 
documented in the language.
     -- 1364-2005 has taken the approach of not documenting this syntax 
in detail in the BNF, instead explained each system task with syntax as 
part of the body of the document. Hence all the system tasks are not 
listed in the syntax section of 1364-2005.
     -- Should AMS follow the same approach (or) actually list the 
system task function formally along with the arguments for the system 
task and move some of the details from the body of the document to the 
Annex A BNF.
     -- Clarification will also be sort from digital as to why this 
approach was taken for 2005.

   - Identifiers (A.9.3)
     -- Analog has two macros __VAMS_ENABLE__ and 
__VAMS_COMPACT_MODELING__. How do these related to the `begin-end 
keyword section of 2005?
     -- Probably should use the same keywords approach in the merged 
2005-AMS BNF. Might be good to identify the compact modeling extensions 
seperately in the keyword section.
     -- Should there be a seperate analog keywords section?

   - Compiler directives (A.10)
     -- This is a new section in the merged BNF.
     -- Digital BNF used similar approach for compiler directives as 
they did for system tasks. These directives were not explicitly stated 
as BNF syntax and were just detailed in the body of the document.
     -- The draft BNF version has all the directives extracted from the 
body and has been explicitly specified.
     -- Need to clarify with digital committee to understand the background.

   - timescale directive (A.10.3)
     -- the subitems for timescale (time unit and precision) are not 
defined in the body of the digital BNF.
     -- there is a \n at the end of the line for the timescale directive.



cheers,
Sri
-- 
Srikanth Chandrasekaran
Design Technology (Adelaide)
Freescale Semiconductor
Ph: +61-(0)8-8168 3592 Fax: x3201