SPARQL query: what is still to do

Dear all,

Following last meeting's promise, here is a quick list of what is
still to do for the SPARQL query specification. I allowed myself some
personal notes, take them with a grain of salt, they might be wrong:

Substantive:
- cleanup behavior of built-in aggregation functions (MAX, SAMPLE...)
in presence of errors: (issue #21 and #23) [mostly editorial to make
the behaviors clearly specified]
- cleanup "substitute" behaviors that are currently very unspecified
(the operation underlying FILTER EXISTS):
https://github.com/w3c/sparql-query/issues/21 (issues #21 #22 #72 #73
#74 #75 #76) [Andy has a proposal to solve that:
https://github.com/w3c/sparql-dev/blob/main/SEP/SEP-0007/sep-0007.md]
- clarify the blank nodes may be shared between a consecutive BGP and
property path (issue #23) [some paragraphs were not updated in SPARQL
1.1]
- clarify what the results should be when there is a projection
alongside both DISTINCT and ORDER BY operators (issue #24) [bad edge
case, a small clarification would be nice]
- fix the discrepancy between the specification and the test suite in
presence of an empty GROUP BY and no solution (issue #24 #105) [Andy's
proposal: https://afs.github.io/sparql-agg-group-empty.html]
- integrate RDF-star CG report (issue #53)
- support lang string directions (issue #112)
- deal explicitly with XPath implicit timezone (issue #116) [XPath
functions assume a default timezone, SPARQL does not at all, we need
at least to document explicitly the difference, and maybe consider
using a default timezone in some cases]

Editorial:
- migrates to the RDF-concept definition of dataset (issue #23 #28)
- define "unaggregated variables" (issue #22)
- fix sum definition example (issue #24)

+ lots of formatting improvements

Best,

Thomas

Received on Wednesday, 16 August 2023 19:49:50 UTC