RE: Empty events lists

Let’s not forget that the most likely reason someone has written “wait()” is
that they are using code written for an SC_CTHREAD and potentially are
trying to convert that for use with an SC_THREAD.

In that case an error or warning will at least give them notice that they
need to add a clock event to the list.

-msm-

*From:* owner-systemc-p1666-technical@eda.org [mailto:
owner-systemc-p1666-technical@eda.org] *On Behalf Of *Bishnupriya
Bhattacharya
*Sent:* Tuesday, November 02, 2010 3:02 AM
*To:* david.long@doulos.com; john.aynsley@doulos.com
*Cc:* Philipp A. Hartmann; owner-systemc-p1666-technical@eda.org;
systemc-p1666-technical@eda.org
*Subject:* RE: Empty events lists

Agreed. I think wait(empty_list) should be an error, and wait(10, SC_NS,
empty_list) should degenerate to wait(10, SC_NS) with a warning.

-Bishnupriya

 ------------------------------

*From:* owner-systemc-p1666-technical@eda.org [mailto:
owner-systemc-p1666-technical@eda.org] *On Behalf Of *david.long@doulos.com
*Sent:* Tuesday, November 02, 2010 2:54 PM
*To:* john.aynsley@doulos.com
*Cc:* Philipp A. Hartmann; owner-systemc-p1666-technical@eda.org;
systemc-p1666-technical@eda.org
*Subject:* Re: Empty events lists

I agree with Philipp.

Making

  wait( empty_list );

behave like

  wait();

just seems wrong to me. I think it should throw an exception.

Dave

  From:

"Philipp A. Hartmann" <philipp.hartmann@offis.de>

To:

john.aynsley@doulos.com

Cc:

systemc-p1666-technical@eda.org

Date:

01/11/2010 23:29

Subject:

Re: Empty events lists

Sent by:

owner-systemc-p1666-technical@eda.org

 ------------------------------

John,

On 01/11/10 23:26, john.aynsley@doulos.com wrote:
>
> We have not explicitly discussed the semantics of empty event lists, e.g.

I vaguely remember, that you (or someone else?) suggested forbidding
this corner case some time back.

> sc_event_or_list list;
> assert( list.size() == 0 );
> wait( list );
>
> I propose that "If the event list object is empty, the effect shall be
> the same as omitting the event list argument from the call."

Especially for the "list-only" wait() case, I don't really like this.
The user intended to dynamically wait on an (unfortunately empty) list,
why should the process suddenly be triggered by its static sensitivity?

Consider the following:

 SC_THREAD(...)
   sensitive << ev1;
 // ...
 sc_event_or_list list1, list2 = ev1 | ev2;

 // never triggered by ev2, but why?!
 wait( list1 );

I think, such a bug (list1 instead of list2) might be a pain to track
down, especially in more complex situations. So I would prefer to at
least require a warning in case of an empty list.

 For the timeout variants, ignoring empty lists might not be that much
of a problem, since at least the process is never triggered by an
"unrelated" event. For consistency reasons, we might want to issue a
warning as well, though.

Greetings from Oldenburg,
 Philipp

> In other words, in the example above,
>
> wait( list );
>
> shall be equivalent to
>
> wait();
>
> and
>
> wait(10, SC_NS, list);
>
> shall be equivalent to
>
> wait(10, SC_NS);
>
> Is that acceptable?
>
> John A
>
>
>

-- 
Philipp A. Hartmann
Hardware/Software Design Methodology Group
OFFIS Institute for Information Technology
R&D Division Transportation · FuE-Bereich Verkehr
Escherweg 2 · 26121 Oldenburg · Germany
Phone/Fax: +49-441-9722-420/282 · PGP: 0x9161A5C0 · http://www.offis.de/
-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.
-- 
This message has been scanned for viruses and
dangerous content by *MailScanner* <http://www.mailscanner.info/>, and is
believed to be clean.
-- 
This message has been scanned for viruses and
dangerous content by *MailScanner* <http://www.mailscanner.info/>, and is
believed to be clean.
-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.