Minutes of the 9/30/02 SV-BC Meeting.

This is my list of attendees and voting status - please submit corrections:
(a-aaaaa) Cliff Cummings (Sunburst Design) *
(aaaaaaa) David Smith (Synopsys) *
(--aa--a) Heath Chambers (HMC)
(aaaaaaa) Karen Pieper (Synopsys) *
(aaaaa-a) Kevin Cameron (NSC) *
(--aa---) Medi Mohtashemi (Synopsys)
(--aa---) Paul Graham (Cadence)
(--aaaaa) Peter Flake (Co-Design) *
(--aaaa-) Simon Davidmann (Co-Design) *
(-aaaaaa) Stefen Boyd (Boyd Technology) *
(aaaaa-a) Steven Sharp (Cadence) *
(---aaa-) Dave Kelf (Co-Design) *
(aa-aaa-) Dennis Brophy (Model Technology) *
(a--aa--) Mike McNamara (Verisity)
(---aaaa) Tom Fitzpatrick (Co-Design) *
(----aaa) Vasisilios Gerousis (Seimens) *
(a----aa) Francoise Martinolle (Cadence)
(------a) Don Mills (LCDM Engineering)
(aa----a) Gord Vreugdenhil (Synopsys)
(aa-----) Brad Pierce (Synopsys)
* indicates eligible to vote on consensus issues

Minutes from the 9/16 Meeting (can be found at http://www.eda.org/sv-bc).
Kevin moves that we approve the minutes.  Steve seconds.  No opposed.  Mac abstains.  Passes.

Action Items:
Karen forwarded stuff we designated for sv-cc to them.  She also let Peter know
about the issues on his plate.  He will get back to us.

Steve needs to drive the creation of the document with the arguments for deleting
the "static" keyword.  He will work with Gord on the document.

SV-BC17d:  Karen to find a location in the manual.

Issues:
SV-BC17d:  In the definition of the names, if you exceed the number of bits made 
	available, what is supposed to happen.  In Verilog, extra bits are ignored. 
	Gord: we should let it be treated by the duplicate value rule, and the 
	numbering wraps.  Kevin: would prefer it to be an error.  Steve: user 
	probably doesn't expect it to wrap.  Francoise: rounding doesn't seem 
	intuitive.  Should deal with default initial value and auto-increment.

Proposals:

Gord:  "Any enumeration encoding value that is outside the representable range 
	of the enum shall be an error."

	We will vote on this proposal next time.  Karen will find a location for 
	this sentence.

SV-BC17e:  Kevin prefers C-style messages (warn on too large value assigned to too 
	small an object).  The manual indicates strong typing.  Arbitrary values 
	create efficiency issues in either time or space for runtime checking.  
	Gord:  Give a tool the option of producing an error or not.  In VHDL, the 
	LRM allows sparce ranges, but most simulators do not check at runtime for 
	assigned values outside of the listed enum values.  Cliff:  Note that these 
	conditions may be checked, but are not required.  Gord:  Behavior is 
	implementation dependent if it is not checked.

Proposals:

Gord:  An implementation may create an error on a conversion of a value to an 
	enumerated type where the value does not correspond to a member of the 
	enumerated type.  If an error is not created, the behavior is not defined.

Steve:  All operations done on enumerations are performed as the underlying integer 
	type.
	David:  The testbench donation does not do this, because it only allows +, -.  
		All other operations require a type conversion.

Steve:  Remove the "strongly typed" reference.

Because of overlap with the enum declaration, we can make a proposed implementation
that we don't have to create language until we see if there is agreement with the 
testbench donation.

SV-BC17b:  The case should be an error.

Steve:  "No explicit datatype or a 2 state data type" instead of no explicit datatype.

SV-BC17f:  Gord: argument isn't compelling.  Tools have to do a whole lot of checking
	at elaboration time now, and we can't avoid that.  Steve: Given a desire 
	to be like C, this is just unfortunate, but unfixable...  

	We agree with the observation, we are not going to do anything about it.
	Steve agrees to close this one.

SV-BC18a:  Steve: Many suggestions for SystemVerilog were extensions to 1995 
	rather than 2001, 2001 chose not to so this SystemVerilog should not.
	Also, we don't need the keyword because the ability to declare up one
	scope exists.
	The static keyword is used to allow static initialization of variables
	in named blocks whereas 2001 did not allow it because of potential 
	confusion with assignment statements.
	Karen: Vassilios has limited our charter to not removing functionality,
	but instead to clarifying the functionality that is there.
	
	Straw pole:  Do you want to remove the keyword "static"?
	Yes: Steve, Gord, Mac, Cliff, Francoise, David 
	No: Karen
	Abstain: Kevin, Dennis

	How do we raise this with Vassilios?  The Chair or Co-Chair should 
	take the concensus and arguments to Vassilios.  Steve will drive the 
	creation of a document explaining the issues.

Next meeting is in two weeks, on 10/14/02.