- From: Dave Reynolds <der@hplb.hpl.hp.com>
- Date: Sun, 14 Sep 2008 20:59:07 +0100
- To: Rule Interchange Format Working Group WG <public-rif-wg@w3.org>
This issue was discussed at the last telecon with only myself expressing any reservations. The chairs expressed a desire to either resolve this or at least draft a resolution at the next telecon. Since I won't make that telecon I'm putting my position in email. Summary: I'll withdraw my concerns and accept unrestricted equality in rule bodies in Core. Comments: There was some discussion on Tuesday about this being purely identity or syntactic quality. That's not true in the presence of data types as Michael later pointed out. I expressed a preference that this should be handled via a generalized equality builtin rather than a syntactic construct. The primary reason being to allow us to use any binding pattern machinery we might devise (to express safety or conformance restrictions) to equality as well. I realize that there is not yet any acceptable proposal for binding patterns so this is a theoretical reservation and I'll accept that it's not strong enough to justify deviation from BLD/PRD. Dave Rule Interchange Format Working Group Issue Tracker wrote: > > ISSUE-76: Equality in Core? [Core] > > http://www.w3.org/2005/rules/wg/track/issues/76 > > Raised by: Jos de Bruijn > On product: Core > > There is already a widespread consensus that Core will not allow equality in the rule heads. > > The question which remains: Will Core allow equality in the rule bodies? > It is known that introducing equality in rule bodies does not increase expressiveness: rules with equality in the body can be straightforwardly equivalently (i.e., they have the same models) rewritten to rules without equality. So, equality in the body is simply a syntactic shortcut. > > It is argued in [1] that a restricted form of equality should be allowed in the rule bodies to facilitate external function calls. However, as mentioned above, this equality is not necessary. It is simply a syntactic shortcut. > > > So, we basically have three choices: > a) allow equality in rule bodies > b) allow only a restricted form of equality in rule bodies, as argued in [1] > c) not allow equality in rule bodies > > > To me personally, (b) does not make sense. Why allow equation of variables and functions, but not variables and variables? > So, I argue that we should either (a) allow or (b) disallow equality in rule bodies. > > >
Received on Sunday, 14 September 2008 19:59:53 UTC