- From: Eric Prud'hommeaux <eric@w3.org>
- Date: Mon, 1 Oct 2012 22:08:15 -0400
- To: Lee Feigenbaum <lee@thefigtrees.net>
- Cc: public-rdf-dawg@w3.org
* Lee Feigenbaum <lee@thefigtrees.net> [2012-10-01 20:51-0400]
> Eric, the assertion about correcting existing tests seems confusing
> given that we have/had multiple passing implementations for all of
> the uncorrected tests. Given that, will you be present at tomorrow's
> telecon so we can go through all of these?
Sure, though I can state this here as well. The corrections come in one
of two flavors: canonical decimals and my mistake.
<http://www.w3.org/2009/sparql/docs/tests/data-sparql11/functions/manifest#ceil01>
asserts that the CEIL("1.6"^^xsd:decimal) is "1"^^xsd:decimal. The
canonical lexical form for a decimal 1 is "1.0" per
<http://www.w3.org/TR/xmlschema-2/#decimal-canonical-representation>.
How would a system know to invent an arbitrary non-canonical form for
a decimal? Perhaps systems were passing these tests by doing numeric
equivalence rather than RDF term equivalence in testing result sets.
As to the sha{1,256,512} hashes for "\u98DF", that was a bug in my
turtle lexer which I've since repaired. Please disregard those.
> Lee
>
> On 10/1/2012 5:50 PM, Eric Prud'hommeaux wrote:
> >SWObjects doesn't implement property paths, but apart from that, here are the data-sparql11 test result diffs. They are annotated (leading character '#' after the patches) here and attached without annotations. I believe these patches correct the existing tests.
> >
> >Index: aggregates/agg-numeric.ttl
> >===================================================================
> >RCS file: /w3ccvs/WWW/2009/sparql/docs/tests/data-sparql11/aggregates/agg-numeric.ttl,v
> >retrieving revision 1.1
> >diff -r1.1 agg-numeric.ttl
> >8c8
> >< :mixed2 :double 2E-1 ; :dec 2.2 .
> >---
> >>:mixed2 :double 2.0E-1 ; :dec 2.2 .
> ># Generated terms must be in some form, let's use the canonical form.
> >Index: aggregates/manifest.ttl
> >===================================================================
> >RCS file: /w3ccvs/WWW/2009/sparql/docs/tests/data-sparql11/aggregates/manifest.ttl,v
> >retrieving revision 1.22
> >diff -r1.22 manifest.ttl
> >338c338
> >< mf:name "Aggregate over empty group resulting in a row with unbound variables" ;
> >---
> >> # mf:name "Aggregate over empty group resulting in a row with unbound variables" ;
> ># Two mf:name arcs -- get rid of one.
> >Index: functions/ceil01.srx
> >===================================================================
> >RCS file: /w3ccvs/WWW/2009/sparql/docs/tests/data-sparql11/functions/ceil01.srx,v
> >retrieving revision 1.1
> >diff -r1.1 ceil01.srx
> >17c17
> >< <binding name="ceil"><literal datatype="http://www.w3.org/2001/XMLSchema#decimal">3</literal></binding>
> >---
> >> <binding name="ceil"><literal datatype="http://www.w3.org/2001/XMLSchema#decimal">3.0</literal></binding>
> >27c27
> >< <binding name="ceil"><literal datatype="http://www.w3.org/2001/XMLSchema#decimal">2</literal></binding>
> >---
> >> <binding name="ceil"><literal datatype="http://www.w3.org/2001/XMLSchema#decimal">2.0</literal></binding>
> >32c32
> >< <binding name="ceil"><literal datatype="http://www.w3.org/2001/XMLSchema#decimal">-1</literal></binding>
> >---
> >> <binding name="ceil"><literal datatype="http://www.w3.org/2001/XMLSchema#decimal">-1.0</literal></binding>
> ># Use canonical form for generated terms
> >Index: functions/floor01.srx
> >===================================================================
> >RCS file: /w3ccvs/WWW/2009/sparql/docs/tests/data-sparql11/functions/floor01.srx,v
> >retrieving revision 1.1
> >diff -r1.1 floor01.srx
> >17c17
> >< <binding name="floor"><literal datatype="http://www.w3.org/2001/XMLSchema#decimal">2</literal></binding>
> >---
> >> <binding name="floor"><literal datatype="http://www.w3.org/2001/XMLSchema#decimal">2.0</literal></binding>
> >27c27
> >< <binding name="floor"><literal datatype="http://www.w3.org/2001/XMLSchema#decimal">1</literal></binding>
> >---
> >> <binding name="floor"><literal datatype="http://www.w3.org/2001/XMLSchema#decimal">1.0</literal></binding>
> >32c32
> >< <binding name="floor"><literal datatype="http://www.w3.org/2001/XMLSchema#decimal">-2</literal></binding>
> >---
> >> <binding name="floor"><literal datatype="http://www.w3.org/2001/XMLSchema#decimal">-2.0</literal></binding>
> ># Use canonical form for generated terms
> >Index: functions/round01.srx
> >===================================================================
> >RCS file: /w3ccvs/WWW/2009/sparql/docs/tests/data-sparql11/functions/round01.srx,v
> >retrieving revision 1.1
> >diff -r1.1 round01.srx
> >17c17
> >< <binding name="round"><literal datatype="http://www.w3.org/2001/XMLSchema#decimal">3</literal></binding>
> >---
> >> <binding name="round"><literal datatype="http://www.w3.org/2001/XMLSchema#decimal">3.0</literal></binding>
> >27c27
> >< <binding name="round"><literal datatype="http://www.w3.org/2001/XMLSchema#decimal">1</literal></binding>
> >---
> >> <binding name="round"><literal datatype="http://www.w3.org/2001/XMLSchema#decimal">1.0</literal></binding>
> >32c32
> >< <binding name="round"><literal datatype="http://www.w3.org/2001/XMLSchema#decimal">-2</literal></binding>
> >---
> >> <binding name="round"><literal datatype="http://www.w3.org/2001/XMLSchema#decimal">-2.0</literal></binding>
> ># Use canonical form for generated terms
> >Index: functions/seconds-01.srx
> >===================================================================
> >RCS file: /w3ccvs/WWW/2009/sparql/docs/tests/data-sparql11/functions/seconds-01.srx,v
> >retrieving revision 1.3
> >diff -r1.3 seconds-01.srx
> >10c10
> >< <binding name="x"><literal datatype="http://www.w3.org/2001/XMLSchema#decimal">1</literal></binding>
> >---
> >> <binding name="x"><literal datatype="http://www.w3.org/2001/XMLSchema#decimal">1.0</literal></binding>
> >14c14
> >< <binding name="x"><literal datatype="http://www.w3.org/2001/XMLSchema#decimal">2</literal></binding>
> >---
> >> <binding name="x"><literal datatype="http://www.w3.org/2001/XMLSchema#decimal">2.0</literal></binding>
> >18c18
> >< <binding name="x"><literal datatype="http://www.w3.org/2001/XMLSchema#decimal">0</literal></binding>
> >---
> >> <binding name="x"><literal datatype="http://www.w3.org/2001/XMLSchema#decimal">0.0</literal></binding>
> >22c22
> >< <binding name="x"><literal datatype="http://www.w3.org/2001/XMLSchema#decimal">3</literal></binding>
> >---
> >> <binding name="x"><literal datatype="http://www.w3.org/2001/XMLSchema#decimal">3.0</literal></binding>
> ># Use canonical form for generated terms
> >Index: functions/sha1-02.srx
> >===================================================================
> >RCS file: /w3ccvs/WWW/2009/sparql/docs/tests/data-sparql11/functions/sha1-02.srx,v
> >retrieving revision 1.1
> >diff -r1.1 sha1-02.srx
> >8c8
> >< <binding name="hash"><literal>d46696735b6a09ff407bfc1a9407e008840db9c9</literal></binding>
> >---
> >> <binding name="hash"><literal>dae30070783bd163def39a3db66fbef89171fa8c</literal></binding>
> ># SWObjects didn't get the same value. I tested with a pretty well-exercised tool and got the "dae3...fa8c" value
> ># perl -e 'use Digest::SHA; print Digest::SHA::sha1_hex("98DF"), "\n"'
> ># dae30070783bd163def39a3db66fbef89171fa8c
> >Index: functions/sha256-02.srx
> >===================================================================
> >RCS file: /w3ccvs/WWW/2009/sparql/docs/tests/data-sparql11/functions/sha256-02.srx,v
> >retrieving revision 1.1
> >diff -r1.1 sha256-02.srx
> >8c8
> >< <binding name="hash"><literal>0fbe868d1df356ca9df7ebff346da3a56280e059a7ea81186ef885b140d254ee</literal></binding>
> >---
> >> <binding name="hash"><literal>06d13a5cb62a47c7bd2c040c6276f322259d246fdbaac73cf3dee402ae923269</literal></binding>
> ># perl -e 'use Digest::SHA; print Digest::SHA::sha256_hex("98DF"), "\n"'
> ># 06d13a5cb62a47c7bd2c040c6276f322259d246fdbaac73cf3dee402ae923269
> >Index: functions/sha512-02.srx
> >===================================================================
> >RCS file: /w3ccvs/WWW/2009/sparql/docs/tests/data-sparql11/functions/sha512-02.srx,v
> >retrieving revision 1.1
> >diff -r1.1 sha512-02.srx
> >8c8
> >< <binding name="hash"><literal>b433ed0e60c818bea72d3aa1a43db89b3ed2b624597407b7912bbb7685f2e45ae5500e092da5f938391d282b26bc43e4035b12460c93ab5e2e1a05d582331d85</literal></binding>
> >---
> >> <binding name="hash"><literal>b43de20b1b0147e07290af96dad226a66c80bad483f9c8388726d2fe67b06d5bc7bc25eaee8d4fee8de5fbe4947bb65e13738881ec0320a5f15366fa132d201b</literal></binding>
> ># perl -e 'use Digest::SHA; print Digest::SHA::sha512_hex("98DF"), "\n"'
> ># b43de20b1b0147e07290af96dad226a66c80bad483f9c8388726d2fe67b06d5bc7bc25eaee8d4fee8de5fbe4947bb65e13738881ec0320a5f15366fa132d201b
> >
> >There's one more which I would have a hard time explaining in a SPARQL tutorial. How does the graph ?g in subquery03 simultaneously constrain the subselect's BGP to be in some (all, really) named graph without constraining the ?g to be that graph?
> >[[
> >select ?x where {
> >graph ?g {
> > {select ?x where {?x ?p ?g}}
> >}
> >}
> >]]
> >Having subselects ignore containing graph constraints seems like a more consistent and predictable language. Thus the above {?x ?p ?g} would be matched against the default graph. If the user wants it to be in a named graph, they can utter the graph constraint. This is trivial to implement (similar to what one does with nested name graphs) and IMO, far more clear to readers/writers of SPARQL queries, as well as 3rd party rewriters.
>
--
-ericP
Received on Tuesday, 2 October 2012 02:08:45 UTC