Subject: [sv-bc] Are enum identifiers elaboration time constants
From: Mark Hartoog (Mark.Hartoog@synopsys.com)
Date: Fri Oct 31 2003 - 09:35:43 PST
Section 3.10 says "An enumerated type declares a set of
integral named constants". At the end of the first paragraph
in section 3.10.3 it says "The enumeration values can still
be used as constants in expressions, and the results can
be assigned to any variable of a compatible integral type."
My question is are these elaboration time constants or run
time constants?
Section 5.3 on constants does not discuss enum identifiers. It
says parameters, local parameters, spec parameters, genvars and
constant functions of these are elaboration time constants, while
'const' variables are run time constants.
In the BNF section A.8.4 there is the rule constant_primary, which
does not include enum identifiers. (It does include parameter,
specparm and genvar identifiers). I believe this rule is suppose
to describe elaboration time constants.
This appears to indicate that enum identifiers are not elaboration
time constants. However, note that if they are not elaboration
time constants, then the following is illegal:
typedef enum {false, true} bool;
module m #(parameter bool P = false) ();
endmodule
module top();
m #(true) u1();
endmodule
The value of a parameter must be an elaboration time constant. You would
have to write this as:
typedef enum {false, true} bool;
module m #(parameter bool P = bool'(0)) ();
endmodule
module top();
m #(bool'(1)) u1();
endmodule
This defeats the purpose of having parameters of type enum.
I feel that enum identifiers should be elaboration time constants,
and the LRM should be updated to indicate this. Section 5.3 will need some
changes, and enum identifiers need to be added to the constant_primary.
Mark Hartoog
700 E. Middlefield Road
Mountain View, CA 94043
650 584-5404
markh@synopsys.com
This archive was generated by hypermail 2b28 : Fri Oct 31 2003 - 09:38:45 PST