- From: Eric Prud'hommeaux <eric@w3.org>
- Date: Thu, 23 Jun 2011 12:00:23 -0400
- To: David McNeil <dmcneil@revelytix.com>
- Cc: Souripriya Das <souripriya.das@oracle.com>, RDB2RDF Working Group WG <public-rdb2rdf-wg@w3.org>
* David McNeil <dmcneil@revelytix.com> [2011-06-22 11:05-0500]
> > But, if we want to change, my preferences would be (most liked first):
> > - stay with "use*"
> >
>
> Part of my concern with "use*" is that it seemed to be inconsistently
> applied. That is, some properties started with "use" but others did not,
> presumably because some property names collided with class named (and I
> don't think the solution is to add more "use" prefixes :)
>
>
> > - use abbreviations: useLogTab, useSmap, usePOmap, usePmap, useOmap,
> > useRefPOmap, useRefPmap, useRefOmap
> > - no use, more abbrev: logTab, sMap, poMap, pMap, oMap, refpoMap, refpMap,
> > refoMap
> >
>
> Hmm... I appreciate the drive to shorten the properties, but these
> abbreviations seem too cryptic to me.
two more options, either of which I prefer to the others:
- drop use: drop "use" and leave predicate and class names distinguished only by capitolization.
- revert all: return to old predicate names withou use* and, as collatoral, restore the *Class classes.
The priority should be on the predicate names; those are what people type. The class names are there for documentation.
$ perl -ne 'while (/\G.*?[^a-zA-Z](use[A-Z][a-zA-Z]*)/gms) { print $1, "\n" }' | sort | uniq
useLogicalTable
useObjectMap
usePredicateMap
usePredicateObjectMap
useRefObjectMap
useRefPredicateMap
useRefPredicateObjectMap
useSubjectMap
In order to give us something to evaluate, I did this:
cp Overview.html P1.html
perl -ne 'while (/\G.*?[^a-zA-Z](use[A-Z][a-zA-Z]+)/g) { print $1, "\n" }' | sort | uniq | perl -ne 'm{^(.*)$}; print "perl -pi -e s/", $1, "/", lc(substr($1, 3, 1)), substr($1, 4), "/g P1.html", "\n"' | bash
<http://www.w3.org/2001/sw/rdb2rdf/r2rml/P1#TriplesMap_Properties> has assertions like "rr:logicalTable rdfs:range rr:LogicalTable ." which seem fine to me. The spec seems to hang together just fine with e.g. refrences to <a>LogicalTable</a>. This is my first choice.
--
-ericP
Received on Thursday, 23 June 2011 16:00:53 UTC