Re: Draft SystemC LRM for review - URGENT!

I just wanted to be clear about the semantics of enable to avoid confusion. I could say "As a consequence of the rules given in [quote clause], calling enable does not cause the target process to become runnable" (take out the "shall").

John A

-----"Philipp A. Hartmann" <philipp.hartmann@offis.de> wrote: -----
To: john.aynsley@doulos.com
From: "Philipp A. Hartmann" <philipp.hartmann@offis.de>
Date: 05/05/2011 09:30PM
Cc: sofie_vandeputte@yahoo.com, stanleyk@cadence.com, P1666 Technical WG <systemc-p1666-technical@eda.org>
Subject: Re: Draft SystemC LRM for review - URGENT!

Yes, without the corner cases it would be perfectly fine to add it.  But
not necessarily required, is it?  IMHO, the rest of the rules already
imply that the process is not made runnable just by calling enable.

  The implementation-defined behaviour is not allowed to break this
normative rule.  Maybe we can move it to a NOTE or make it otherwise
non-binding (the "shall" is the problem, I think?).

Thanks,
  Philipp

On 05/05/11 22:19, john.aynsley@doulos.com wrote:
> Hmm. Do you agree that barring implementation-defined corner cases it is
> true? I cannot qualify every single obligation concerning process
> control to say "(unless some implementation-defined behavior related to
> a corner case means that it need not hold)". I could omit the "under no
> circumstances", if that is the issue for you.
>
> John A
>
>
> -----"Philipp A. Hartmann" <philipp.hartmann@offis.de> wrote: -----
>
>     To: john.aynsley@doulos.com
>     From: "Philipp A. Hartmann" <philipp.hartmann@offis.de>
>     Date: 05/05/2011 09:09PM
>     Cc: systemc-p1666-technical@eda.org, sofie_vandeputte@yahoo.com,
>     stanleyk@cadence.com
>     Subject: Re: Draft SystemC LRM for review - URGENT!
>
>     John, all,
>
>     there is in fact an issue with respect to the process control corner
>     cases.  It's in the description of enable (5.6.6.2) on page 83, lines
>     61f, where the draft says:
>
>       "[CHANGE] Calling enable shall under no circumstances cause the
>        target process instance to become runnable."
>
>     This effectively prohibits one of the alternatives for an implementation
>     to define the disable/resume interaction (5.6.6.11, page 90, lines 53ff)
>     in one of the corner cases.
>
>     Are there pressing reasons to add this restriction to the standard?
>
>     The rest looks good to me, also the annex with the changes between
>     P1666-2005 and P1666-2011.  Very useful indeed.
>
>     Greetings from Oldenburg,
>       Philipp
>
>
>     On 05/05/11 18:19, john.aynsley@doulos.com wrote:
>     > Folks,
>     >
>     > I have uploaded a draft P1666 SystemC LRM for review prior to the
>     > recirculation ballot. Stan wants to start the ballot as soon as
>     > possible, so please review the LRM and shout on this reflector if you
>     > see any show-stoppers.
>     >
>     > Every change to the document made since the first ballot has been
>     marked
>     > [CHANGE] and shown in red, so simply search for "[CHANGE]" in your PDF
>     > viewer. You will see that I have replaced Annex D with a fairly
>     > comprehensive list of changes.
>     >
>     > You can find the draft LRM at:
>     >
>     >        
>     > http://www.eda.org/twiki/pub/P1666/WebHome/1666-2011_May_05_11.pdf
>     > <http://www.eda.org/twiki/pub/P1666/WebHome/1666-2011_May_05_11.pdf>

-- 
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 · http://www.offis.de/
Phone/Fax: +49-441-9722-420/282 · PGP: 0x9161A5C0 · Skype: phi.har
-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.