[Fwd: Definition of integer division]

re-sending for Shalom (unclear to me why it bounced)

"Bresticker, Shalom" wrote:
> 
> 1364-2005, final version, adds the following in subclause 5.5.2:
> 
> "Propagate the type and size of the expression (or self-determined
> sub-expression) back down to the
> context-determined operands of the expression. In general, any
> context-determined operand of an operator shall be the same type and
> size as the result of the operator. However, there are two exceptions:
> 
> 1) If the result type of the operator is real, and it has a
> context-determined operand that is not real, that operand shall be
> treated as if it were self-determined and then converted to real just
> before the operator is applied."
> 
> What this basically means is that in general, you first determine the
> size and type of the result, coerce/convert all the context-determined
> operands to that type and size, and only then start to evaluate the
> expression.
> 
> If that were to hold when the result type is real (i.e., when any
> context-determined operand is real), then all the operands would be
> converted to real, and only then would evaluation start.
> 
> But reals are an exception. (See
> http://www.boyd.com/1364_btf/report/full_pr/83.html
> (Mantis 1011) and
> http://www.boyd.com/1364_btf/report/full_pr/282.html.)
> 
> Conversion to real is not done at the beginning, only just before the
> operation is performed. So in the case of "a = 5.0 + 1/2;", 1/2 is still
> performed as integer division, and 0 is converted to 0.0 for the
> addition to 5.0.
> 
> This is the way Verilog-XL was implemented and thus it became legacy
> which everyone copied. I think maybe C also works this way, but even if
> not, it is a fact which is not going to change now.
> 
> With respect to division by 0.0 for reals, I filed an ETF issue on it.
> See http://www.boyd.com/1364_btf/report/full_pr/273.html (Mantis 1225).
> See also http://www.boyd.com/1364_btf/report/full_pr/274.html
> (Mantis 1226).
> 
> Shalom
> 
> > -----Original Message-----
> > From: owner-verilog-ams@eda.org [mailto:owner-verilog-
> > ams@eda.org] On Behalf Of Geoffrey.Coram
> > Sent: Wednesday, December 14, 2005 8:51 PM
> > Cc: verilog-ams@eda.org
> > Subject: Re: Definition of integer division
> >
> > Kevin -
> > OK, I had been searching for "coerc" when I found that one
> > statement.  A few paragraphs above, I found this:
> >
> > - Expression type depends only on the operands. It does ot
> >   depend on the LHS (if any).
> >
> > followed by
> > - For non-self-determined operands the following rules apply:
> >     if any operand is real, the result is real;
> >
> >
> > If I have
> >   real a;
> >   a = 5.0 + 1/2;
> >
> > then I guess "1/2" is an expression (and its type is
> > integer), and thus "integer division" is performed
> > (no coercion is needed).
> >
> > At first I thought "5.0 + 1/2" was the expression.
> >
> >
> > I'm still a little concerned about the sentence (in 4.1.5)
> >   For the division or modulus operators, if the second operand
> >   is a zero, then the entire result value shall be x.
> >
> > because "x" doesn't make sense in an analog context.
> >
> > -Geoffrey