search

marklogic / marklogic-jena

Adapter for using MarkLogic with the Jena RDF Framework
Other
5 stars 11 forks source link

Refactor write buffer to use INSERT DATA rather than mergeGraph #48

Closed grechaw closed 7 years ago

grechaw commented 8 years ago

In the interest of code path consolidation, we'll use SPARQL update rather than mergeGraph to accomplish the write buffering, which is always inserts and never a locking update.

grechaw commented 8 years ago

Note to @xquery first attempt has led to memory exhaustion on the server.

grechaw commented 8 years ago

Reason for that is because cache.size() returns the number of graphs in the cache not the number of triples, so you can bust the cache by adding a large number of triples to one graph.

grechaw commented 8 years ago

Implemented a solution to my last comment. Tests all pass with insert data now.

grechaw commented 8 years ago

In PR for both 1.0 branch and develop

grechaw commented 8 years ago

This refactoring has been done, no functional changes should surface.