RE: [sv-bc] clarification request

Back to the issue of mixing procedural and continuous assignments on
disjoint bits of packed variables.


>From: "Rich, Dave" <Dave_Rich@mentor.com>

>These restrictions are based on existing restrictions on the
>force/release and assign/deassign constructs, which do not allow slices
>of variables to be continuously assigned, while leaving other parts of
>the variable alone.

Dave,

I don't see how the reasons for the restrictions on force/release and
assign/deassign apply to ordinary continuous assignments.  There is a
big difference between a simple static check, and the more complex and
expensive dynamic tracking that would be needed for force/release.

I could imagine someone not fully understand those reasons and
believing that they applied, and thus proposing this restriction.
But Gord says that he worded it, and I know that he would know better.


>I think the intent of these restrictions was to keep packed objects
>treated as a singular variable and not have to split them up into
>individual bits like you would a wire.

But the rules still allow driving different bits with different
continuous assignments, which already splits them up in that sense.
And the static checking to allow some bits to be written with procedural
assignments instead is a simple extension of the static checking that
no bits are driven with multiple continuous assignments.

I can imagine that some implementation might be able to leverage some
existing analysis to check continuous assignments more easily than
a mixture of cont assigns and procedural assigns.  But that is not an
inherent reason why the language should have such a restriction.

Steven Sharp
sharp@cadence.com


-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.