Final technical agreement sought for EXPRESS 2

[Wilson, Peter R]

    Please consider the following regarding the dependent rule.

o I recall that within STEP USE (or is it REFERENCE?) can (could?) only be used in IRs and REFERENCE (or is it USE) can (could?) only be used in APs. This was justified by "If you look at an EXPRESS model you can tell if it is an AP or an IR according to its use of USE or REFERENCE". Certainly at one time USE/REFERENCE were artifacts of the STEP documentation architecture and did not reflect any  information modeling first/second class distinction .

o Existence dependence is modeled in EXPRESS via multi-schema models and USE/REFERENCE. The STEP architecture (i.e., a single long form) deliberately threw away that distinction.

o If existence dependence is modeled via global rules, what is the impact on the size and clarity of the model? For example, what would the AP203 long form look like with the proposed dependent rules?

o Will anyone implement such rules? (Does anyone implement any global rule?)

    Personally, as a general principle I think that global rules should be avoided wherever possible and only used as a last resort when all other model constraint methods have been exhausted. I am concerned about Phil's suggestion that a SELECT type would have to be introduced into the long form to enable the dependent rules; it's getting complicated and hints at artificiality.

    I would like to see the dependent rules eliminated from the short to long algorithm, especially as I think that the USE/REFERENCE in STEP is an architectural artifact and does not represent real modeling requirements. If that is not acceptable then I am strongly of the opinion that the rules must be optional and not required.

Peter W. 

Dr Peter R. Wilson
Boeing Commercial Airplanes
PO Box 3707, MS 2R-97, Seattle, WA 98124-2207
Tel: (206) 544-0589, Fax: (206) 544-5889
Email: peter.r.wilson at boeing.com
--------------------------------
Any opinions expressed above are personal;
they shall not be construed as representative of any organisation.
--------------------------------
 

> -----Original Message-----
> From: Phil Spiby [mailto:Phil.Spiby at Eurostep.com]
> Sent: Thursday, April 17, 2003 2:22 PM
> To: Wilson, Peter R; wg11 at steptools.com
> Subject: RE: Final technical agreement sought for EXPRESS 2
> 
> 
> Peter,
> 
> You are correct, I hadn't got there when I sent my message 
> earlier, I had
> just got to the bullet points under G.2 where it states that this
> information may be lost. The bullet needs changing to reflect 
> the current
> situation regarding the generation of these global rules.
> 
> As a secondary issue, the algorithm identified was pointed 
> out as being
> inadequate by Alan Williams, my response to this is given below:
> 
> We should correct this step in the algorithm if we are going 
> to keep it in.
> Comments please!
> 
> Phil
> 
> --------------------------------------------------------------
> --------------
> --
> 
> Response message:
> Hi Alan,
> 
> Glad to see that someone is watching this ...
> 
> Yes you are correct I made up the rule whilst writing the 
> e-mail, and should
> have checked it before sending it.
> 
> Your circular dependent structure would cause problems, 
> thinking a little
> more about this we could set up a select type in the 
> long_form schema which
> contains every dependent entity. Then we could test the 
> TYPEOF each member
> of the result of USEDIN(I,'') against the select type and 
> only allow those
> which are not present. A little more complex but still 
> relatively easy.
> 
> Phil
> 
> > -----Original Message-----
> > From: Alan R Williams [mailto:alanrw at cs.man.ac.uk]
> > Sent: 29 October 2002 14:17
> > To: Phil.Spiby at eurostep.com
> > Subject: RE: Comments on WG11 N197 (EXPRESS Ed2 Annex G)
> > 
> > 
> > 
> > > From: "Phil Spiby" <Phil.Spiby at eurostep.com>
> > > To: "'Wilson, Peter R'" <peter.r.wilson at boeing.com>,
> > > <sc4sec at tc184-sc4.org>,
> > "'wg11'" <wg11 at steptools.com>
> > > Subject: RE: Comments on WG11 N197 (EXPRESS Ed2 Annex G)
> > > Date: Tue, 29 Oct 2002 10:16:22 -0000
> > > X-Priority: 3 (Normal)
> > > X-MSMail-Priority: Normal
> > > Importance: Normal
> > > X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1106
> > 
> > [snip]
> > > Instantiability in long forms:
> > > I would argue that each entity which is interfaced either
> > implicitly
> > > or via a REFERENCE statement should result in a global rule which
> > > ensured that it was used in some role possibly: RULE 
> > independent_x FOR
> > > (x); WHERE
> > >   x = QUERY(i <* x | SIZEOF(ROLESOF(i)) > 0);
> > > END_RULE;
> > 
> > Shouldn't the rule be called "dependent_x"?
> > 
> > Won't the rule as it stands return TRUE for entities which
> > aren't valid?  For example if I have REFERENCE imported the 
> > two entities
> > 
> >   ENTITY x;
> >     a : y;
> >   END_ENTITY;
> >   
> >   ENTITY y;
> >     b : x;
> >   END_ENTITY;
> > 
> > then in my information base I could have a pair of x & y that
> > are related to 
> > each other but not related to any other instances.  These 
> > would pass the RULE but would not AFAIK be valid.  Don't you 
> > need to have a set for all the relationships between the 
> > entities in the importing schema and those REFERENCE 
> > imported?  You then have something like:
> > 
> > RULE dependent_x FOR (x);
> > WHERE
> >   x = QUERY(i <* x | SIZEOF(ROLESOF(i) *
> > intra_schema_relationships) > 0); END_RULE;
> > 
> > As to whether anyone cares about enforcing such restrictions,
> > IMO yes as they should affect, for example, an xsd generated 
> > from a multi-schema model.  In reality, though, the 
> > distinction between first and second class entities is 
> > probably made externally, e.g. in another file, or by 
> > comments in the model.
> > 
> > [snip]
> > > Phil
> > 
> > Alan
> > 
> > Alan Williams, Room IT301, Department of Computer Science,
> > University of Manchester, Oxford Road, Manchester, M13 9PL, U.K.
> > Tel: +44 161 275 6270      Fax: +44 161 275 6280
> > 
> 
> > -----Original Message-----
> > From: Wilson, Peter R [mailto:peter.r.wilson at boeing.com] 
> > Sent: 17 April 2003 17:44
> > To: Phil.Spiby at eurostep.com; wg11 at steptools.com
> > Cc: Wilson, Peter R
> > Subject: RE: Final technical agreement sought for EXPRESS 2
> > 
> > 
> > Phil,
> > 
> >     You seem to be reviewing a different version of Ed 2 than 
> > I am. I'm looking at the version with change marks that 
> > Jochen emailed on 28 March. In this the first paragraph of 
> > clause `G.4.3 Secondary population' describes a global rule 
> > for maintaing REFERENCE semantics.
> > 
> >     Personally, I think that due to the artificial 
> > constraints put on the use of USE and REFERENCE by 
> > Yang/Danner that first/second class distinctions are moot in 
> > STEP models. Don't have any such global rules in the mapping. 
> > In other words, I was dissapointed to see the gloabl rule in 
> > the algorithm.
> > 
> >     Regarding comments, I strongly urge that any comment 
> > mapping be left to implementors of the algorithm, and they 
> > should ignore all comments. A comment is a comment is a 
> > comment and is white space as far as any EXPRESS processor is 
> > concerned.
> > 
> > Peter W.  
> > 
> > 
> > > -----Original Message-----
> > > From: Phil Spiby [mailto:Phil.Spiby at eurostep.com]
> > > Sent: Thursday, April 17, 2003 6:46 AM
> > > To: wg11 at steptools.com
> > > Subject: Final technical agreement sought for EXPRESS 2
> > > 
> > > 
> > > Folks,
> > > 
> > > There are two technical issues on the short to long form
> > > generator that, as
> > > far as I can, see there was no recorded agreement.
> > > 
> > > 1. Distinction of first and second class entities in the 
> longform. 
> > > This was proposed and there were discussions around how 
> to form the 
> > > required global rule(s). This is not currently in the draft, but I
> > > haven't seen where
> > > the agreement to pull it from the draft was taken.
> > > I would prefer to see this in the document, but if there has 
> > > been agreement
> > > to remove it, or if there is no agreement on how to do it 
> > > available in the
> > > next week, then I suggest we leave things as they are. 
> > > Although I feel this
> > > is another opportunity to do things correctly which has 
> > gone begging!
> > > 
> > > 2. Mapping of comments.
> > > This has also been discussed and Peter pointed out that since
> > > there was no
> > > association of comments it was impossible. However, the 
> > tagged comment
> > > capability provides an association, which could be mapped through.
> > > I suggest we map through tagged comments, and all other 
> > > comments are dropped
> > > (as done at present).
> > > 
> > > Phil
> > > 
> > >