Re: [sv-bc] Tagged Unions Proposal incorporating 12/8 amendments


Subject: Re: [sv-bc] Tagged Unions Proposal incorporating 12/8 amendments
From: Dave Rich (David.Rich@synopsys.com)
Date: Thu Dec 11 2003 - 08:46:22 PST


Nikhil,

in section 3
In the packed tagged union, are the unspecified bits really X, or just
unspecifed? Saying that they are X is specifing them.

Replace "In the extreme case of a void member, only the tag is
significant and all the remaining bits
are undefined."

With

"In the case of a void member, only the tag bits are left justified
and all the remaining bits
are undefined."

In section 7

All of your tagged union expressions examples have parenthesis around
the value expression, but the BNF does not support that, was that your
intention ?

Add to this sentance "An attempt to read or assign a value whose type is
inconsistent with the tag results in a runtime error. No write shall be
performed and the read return value shall be undefined"

 "An uninitialized variable of tagged union type shall contain x's in
all the bits, including the tag bits." What about a 2-state union?

Rishiyur S. Nikhil wrote:

> Attached is a revised PDF of the Tagged Unions and Pattern Matching
> proposal, incorporating the 4 small amendments suggested during
> yesterday's con-call, which are listed below.
>
> Nikhil
> ----------------
> Amendments incorporated:
>
> 1. Move the "tagged" keyword from before to after the "union" keyword
> in definitions of a tagged union type
>
> 2. Use ".*" instead of "$" for wildcard patterns
>
> 3. Use syntax: case/casex/casez (e) matches ... endcase
> instead of: casep/casepx/casepz (e) ... endcase
>
> 4. Show an example of pattern matching that does not involve tagged
> unions (see bottom of page 10).
>

-- 
--
David.Rich@Synopsys.com
Technical Marketing Consultant
http://www.SystemVerilog.org
tele:  650-584-4026
cell:  510-589-2625



This archive was generated by hypermail 2b28 : Thu Dec 11 2003 - 08:47:30 PST