On 29/03/16 16:10, A. Soroka wrote:
RDF/XML pretty much has to be loaded all together to be processed
(because any given assertion in the XML depends on a lot of context
which could be from elsewhere in the document), so when Fuseki
receives the stream of XML, it's got to assemble a complete graph
before it can act on the data. You will likely find that using a
streamable format (e.g. N-Triples) will get the data through before
Fuseki hiccups. You can use Jena's command-line tools to convert
formats client-side.
If loading into TDB, the issue is transaction space.
Files can be converted to N-Triples and loaded in pieces if bnodes don't
get in the way.
RDF/XML streams when loaded into TDB. Parsing is optimistic (it does
work before knowing the entire file is valid, unlike JSON-LD). Any error
causes a transaction abort. It is well worth checking data before loading.
"riot --validate"
Andy
