The SPARQL 1.1 parser is exactly the spec. In fact, the HTML and the Java parser are built from the same source. The grammar is LL(1) - i.e. simple so people can implement easily.
The ARQ parser rewrites the INSERT DATA with a better, javacc specific way to parse the same language (dynamically changing the lookahead). But it is not applicable to simple parser generators. https://github.com/apache/jena/blob/master/jena-arq/Grammar/master.jj#L760 Andy On 17 October 2017 at 10:14, Neda Alipanah <[email protected]> wrote: > Hi everyone, > While testing history in the graph I’ve run into occasional stack overflow > when adding the Jena Model to the triplestore using an UpdateRequest. > Initially I was using Syntax.syntaxSPARQL_11 (the default for > UpdateFactory) > but after googling a bit I found > http://thread.gmane.org/gmane.comp.apache.jena.user/8195 which makes me > think it is an inherent issue with the SPARQL 1.1 syntax parser and large > updates. Per the recommendation in 8195 I moved to using Syntax.syntaxARQ > which through initial testing seems to have eliminated the stack overflow. > The javadoc for syntaxARQ simply states it is *The query syntax for > extended SPARQL* which doesn’t really tell me a lot. Does anyone know the > difference between syntaxSPARQL_11 vs syntaxARQ? > > Thanks > Neda >
