Re: [Iverilog-devel] comp1001 r66 isolation

[Cary R. <cy...@ya...>]

--- On Thu, 10/30/08, Stephen Williams <st...@ic...> wrote:

> The ! would and should interfere with the expression width
> calculation
> because its argument is self-determined. In other words, it
> should
> prevent the l-value width impacting the width of the
> argument
> expression.

Yes it should interfere with it's argument, but it should not
interfere with what is calling it as an argument. ! has a higher
precedence so it is bound first. The division is then done using
the resulting expression from ! and the zero variable. The
division should be done at the L-value width not the argument
width. It is this incorrect width calculation that is causing
the incorrect run-time result. My guess is that ! is being
stubborn about resizing itself.

FYI, one of your recent patches fixed the compiler to give the
correct result. Also, replacing the !1'b0 with 1'b1 will make
things work correctly.

Cary