From: Steven Sharp (sharp@cadence.com)
Date: Thu Aug 26 1999 - 14:29:44 PDT
B2: Section 3.9.1 (p28) still states that arrays of reals are illegal,
while 3.10.3 (p29) states that they are legal.
B27: There are problems with the statement in section 4.1.5 about complex
results. It doesn't cover all situations in a practical fashion. The
general case of real powers calls for taking a log of the base, which is
not possible for negative numbers. The likely implementation would call
the pow() function from the C math library, which behaved as follows when
I tested it. It returns NaN for a negative base if the exponent is not an
integer. It also returns Inf or -Inf for negative powers of zero. For
0.0**0.0, it happens to return 0, though this could be regarded as an error
condition. We could specify these results, or simply state that the behavior
for a negative base with a non-integer power or zero with a non-positive
power is determined by the underlying implementation.
B19: The description of the >>> operator in section 4.1.12 does not agree
with the implementations in Verilog-XL and NC-Verilog (I realize that we
probably provided these sections). In the implementations, >>> results in
an arithmetic shift for a signed operand and a logical shift for an unsigned
one. In other words, it is the way >> should have worked in the first
place, but that was kept a logical shift for backward compatibility.
BE74: In section 4.1.14 (p47), the first expression should be described
as "positive" or "non-zero unsigned" rather than just "non-zero". Or it
could say that the count is treated as unsigned.
This archive was generated by hypermail 2.1.4
: Mon Jul 08 2002 - 12:53:29 PDT
and
sponsored by Boyd Technology, Inc.