User-Agent: Build Identifier: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.5) Gecko/20031007 Firebird/0.7 The nodeID attribute can be added to a blank node or predicate so that references can be made to them. Patch coming up. Reproducible: Always Steps to Reproduce: 1. 2. 3.

Comment on attachment 140220 [details] [diff] [review] add nodeID support >+ else if (attr.get() == kNodeIdAtom) { >+ hasNodeID = PR_TRUE; >+ nodeID.Assign(aAttributes[1]); >+ } > else if (attr.get() == kAboutEachAtom) { > // XXX we don't deal with aboutEach... > //PR_LOG(gLog, PR_LOG_ALWAYS, > // ("rdfxml: ignoring aboutEach at line %d", > // aNode.GetSourceLineNumber())); > } > } Want to blow away the kAboutEachAtom stuff while you're at it? We never supported it, and it's deprecated now anyways.

Sure, here's an updated patch with the aboutEach stuff removed too

This bug breaks test case http://www.w3.org/2000/10/rdf-tests/rdfcore/rdfms-syntax-incomplete/test002.rdf and the patch appears to fix it.

Comment on attachment 212628 [details] [diff] [review] updated patch for nodeID (does not fix aboutEach problems) You need to request review from someone in particular, by entering their email address in the requestee field on the edit attachment page.

Comment on attachment 212628 [details] [diff] [review] updated patch for nodeID (does not fix aboutEach problems) verified testcase, but as nodeID must be non-empty (the spec says it's an NCName, even, but there's no requirement to actually check that), please drop the hasNodeID and just check for emptyness of nodeID instead.

Good point.

Comment on attachment 212782 [details] [diff] [review] replaced hasNodeID with nodeID.IsEmpty() checks This is good, I'd like to see this on the 1.8.1 branch, too.

Comment on attachment 212782 [details] [diff] [review] replaced hasNodeID with nodeID.IsEmpty() checks sr=shaver