Oxigraph

Next step is a 0.6 release. Main goals are cleaning up some APIs and internals and improving SPARQL conformance.

Oxigraph got a few new features this February: - JSON-LD 1.1 - RDF/XML 1.2 - RDF-canonicalization 1.0 The work for JSON-LD 1.1 has been significant because Oxigraph JSON-LD parser provides a streaming mode, allowing it to not buffer the full file into memory during parsing.

@fils Thank you! I wanted to know if people were doing exactly this kind of queries. By "with a variable" I mean your second query with "GRAPH ?graph" (opposite to the first one were ex:somethine is a set graph name). I don't see an obvious way of making it better.

Curious if you could define what you mean by "with a variable". I am definitely using graph ?g on a dataset with a LOT of named graphs. I am doing things like SELECT ?resource WHERE { GRAPH ex:somethine { ?resource ex:someProperty "someValue" . } } or SELECT ?graph ?resource WHERE { GRAPH ?graph { ?resource ex:someProperty "someValue" . } FILTER(CONTAINS(STR(?graph), "substringToMatch")) } If there is a better way to do that filter I'm interested. Thanks

Quick check to maybe simply Oxigraph and reduce the storage footprint: is there anyone using the SPARQL `GRAPH ?g` syntax (with a variable) on a dataset with a lot of named graphs?

The Rust transactions are now accessible using a type and are not restricted to a closure anymore. The Rust SPARQL execution API has also been rewritten to allow injecting more options (custom aggregate functions, default prefixes...).

The main change is the support for current W3C working drafts RDF 1.2 and SPARQL 1.2. It replaces RDF-star support. The bulk loader now parses NTriples and NQuads files in parallel if they are not compressed.

We released last sunday Oxigraph 0.5, a short year after the 0.4 release. This release is dedicated to the memory of @gkellogg, author of multiple specifications implemented by Oxigraph. 1/

@triyvn Yes! Oxigraph already provides one

@oxigraph I know some graph stores (like JanusGraph) have a 'bulk' mode you can turn on for the purpose of loading data. IIRC, it involved restarting the service to change modes.

It seems that most Oxigraph users use it OLAP-style, mostly doing bulk loading the data or applying large updates. I am considering making Oxigraph concurrency control weaker to make writes significantly faster and blurry the difference between load and bulk_load.

Is it a problem for anyone?

It will mean that transaction isolation guarantees will be weaker: for example if a SPARQL update clears a graph (U1) while another one inserts a new triple (U2) in it one might end up with the graph only containing the new triple, even if U1 is committed after U2.

Update on this: an automated migration code has been written and will be included in the next Oxigraph release. It will be automatically executed when a database is opened for the first time by Oxigraph 0.5+.

GitHub issue: github.com/oxigraph/oxigr… For reference, triple terms is the "<< s p o >>" syntax, if you don't know what it is, you are very likely not affected.

The upcoming RDF-1.2 specification breaks things related to triple terms compared to RDF-star. Oxigraph is planning to migrate to RDF-1.2 but is unsure if it's worth it to provide a costly migration path. Please reach out if you have a lot of RDF-star data in Oxigraph.

Many thanks to the German Historical Institute in Rome for sponsoring the work.

To use the streaming parser use the if instructed to do so using the application/ld+json;profile=#streaming" target="_blank" rel="nofollow noopener">w3.org/ns/json-ld#str… media type. Streaming JSON-LD: #streaming-profile" target="_blank" rel="nofollow noopener">w3.org/TR/json-ld11-s…

Oxigraph 0.4.10 provides a new JSON-LD 1.0 parser and serializer. It can do streaming parsing on documents conforming to the Streaming JSON-LD profile.

Big thank you to the sponsors who made a bunch of these improvements possible! 5/5

- "GRAPH ?g {}" now properly lists all named graphs (v0.4.7) - Some GeoSPARQL functions are now implemented (beware, very naive and slow implementation) (v0.4.2) - SPARQL query results parsers now support async I/O (v0.4.0) 4/

Some long overdue updates from the past Oxigraph releases: - New Turtle/TriG/NTriples/NQuads/N3 parser named oxttl that supports async I/O and a much faster lenient mode (v0.4.0) - New facade oxrdfio crate to parse/serialize RDF in a bunch of different formats (v0.4.0) 1/