RE: $finish and final_step

Hi All,

As far as I can see now we need $finish to behave nicely - the user should
be aware that a lot of execution is still happening after $finish which is
not the impression he would get from the current LRM text. If a user wants
to stop immediately - for instance to prevent a division by zero or not to
execute a long calculation that is known to be wrong beforehand - something
akin to $fatal should probably be used.

Also in case of a $finish in a digital block the analog parts should be
notified so they can all be executed once, including the final-step event.
Do we need to extend the VPI to pass on such a notification? This might
better be picked-up by the mixed-signal subcommittee and not to be
hurriedly pushed into 2.3 release.

By the way - what should happen if $finish is used in an final_step event?
From the discussion I guess it should it be ignored - apart from passing
the notion to the user that the simulation ended from this instance of the
$finish statement if no earlier occurrence of the $finish statement had
triggered the execution of the final_step event.

Additionally, there is talk about the final block in SV: it is not present
in 1364-2005 standard. I don't think it should be introduced prematurely in
the Verilog-AMS standard. LRM 2.3 was about synchronizing with 1364-2005,
not yet with 1800-2005.

Cheers,
Marq

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