RE: [sv-ec] fork...join_none in functions, and final blocks

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

>> Oddly enough, the proposal for 1615 would make it illegal for the
>> final block to execute the fork..join_none by calling a function
>> containing it, but legal to execute it directly.
>
>You are not correct. I think we are still having e-mail latency
>problems. My earlier reply said that by definition, final block
>statements have the same restrictions as function statements.

This may be a matter of interpretation.

I am looking at the text of the proposal for 1615.  It says that it
is illegal to call a function that executes a fork..join_none except
from certain contexts.  Executing a fork..join_none directly in a
final block is not calling a function that executes a fork..join_none.
So that restriction is not applicable.

If it were expressed as a restriction on what processes can execute
a fork..join_none, then it would apply to directly executing it from
a final block.  But it is explicitly specified as a restriction on
who can call a function containing the fork..join_none.  So it only
applies if the final block calls a function containing fork..join_none.

One of the earlier sentences is somewhat closer, since it talks about
what the function can do.  But it still talks about the process calling
the function, which seems to keep it from applying to a final block.

Steven Sharp
sharp@cadence.com