- From: Peter F. Patel-Schneider <pfpschneider@gmail.com>
- Date: Fri, 31 Mar 2017 07:18:32 -0700
- To: Andy Seaborne <andy@apache.org>, public-sparql-exists@w3.org
On 03/23/2017 08:11 AM, Andy Seaborne wrote: > https://w3c.github.io/sparql-exists/docs/sparql-exists.html > > I've made some updates to reflect recent discussions and also to add > descriptive text. What I'd really like feedback on is where less can be > said. Just adding text to explain points risks putting too much in. > > Andy > I see that the domain of PrjMap has been expanded. This doesn't affect EXISTS because the argument to EXISTS can't result in a projection (because internal projections are always inside a ToMultiSet). There appears to have been some changes to the section on limitations on assignment. This section is still quite unclear. I guess that it means that there cannot be an AS binding to the variables coming from outside anywhere in the argument to EXISTS. This is far too broad, as it prevents AS binding of disconnected variables of the same name. It also prevents AS binding in places that are non-problematic, such as AS binding in the select clause. There also appears to be a requirement that VALUES not use variables coming in from the outside. I don't see any particular need for this limitation. It may be that this is in response to a problem that I pointed out where the example used such a VALUES construct, but the problem has nothing to do with assignment and instead has to do with there not being a place in the resultant algebra where values insertion takes place. Proposal B adds in a new kind of technology to SPARQL, namely replacement of variables by fresh variables. This new technology needs to be carefully examined before it becomes a part of SPARQL. A detailed explanation of it is needed, including demonstration of its safety, so that this examination can be done effectively. Peter F. Patel-Schneider Nuance Communications
Received on Friday, 31 March 2017 14:19:06 UTC