Austin Harris
Austin Harris
Been thinking about this a bit, and I think the easiest place to start would be by taking a webassembly file that binds to map/reduce, and then puts those results...
Two implementations I found are by @migueldeicaza and one by @bytecodealliance. https://github.com/migueldeicaza/WasmerSharp https://bytecodealliance.github.io/wasmtime/lang-dotnet.html Objectives - [ ] Define metadata / schema for how we look for plugins - [ ]...
The RDF XML format is one of the popular RDF file formats. Using a basic XML grammar could do. https://github.com/antlr/grammars-v4/tree/master/xml You could follow the approach taken for the Ntriples importer...
The turtle file format is one of the popular RDF file format. There is a G4 grammar for it here: https://github.com/antlr/grammars-v4/blob/master/turtle/TURTLE.g4 You could follow the approach taken for the Ntriples...
This is to allow an improved sequential scan of data from disk.
https://www.microsoft.com/en-us/research/uploads/prod/2018/03/faster-sigmod18.pdf See section 3.1 > Each hash bucket entry (Fig. 2) consists of three parts: a tag (15 bits), a tentative bit, and the address (48 bits).
This will help with consistency of terms used in our product and the papers we are referencing.
We should have a sensible default number of partitions, say 1000. These partitions with be assigned to nodes, and can be reassigned to other nodes and the cluster grows.
Because nodes come in fragmented across multiple writes, we need to collect attributes from all the linked fragments.