@graphy/memory.dataset.fast - npm Package Versions

v4.2.0 (2020-04-20)

This update adds experimental multithreaded reading for N-Triples and N-Quads, called 'Scanners': @graphy/content.nt.scan and @graphy/content.nq.scan. These are deserializers that take a single input stream and read it across multiple workers in parallel, allowing users to take advantage of multiprocessor systems.

@graphy/content.nt.scan and @graphy/content.nq.scan are experimental; they have limited test case coverage in node.js and have not been tested in the browser.

How it works: The main thread populates sections of a preallocated SharedArrayBuffer with chunks copied from the input stream data and allows workers to claim each slot on a first-come-first-serve basis (synchronized using Atomics). For each chunk of input data, the main thread handles reading the leading and trailing fragments that belong to RDF statements spanning across chunk boundaries. Communication about data exchange is done mostly through unidirectional byte indicators on SharedArrayBuffer, with occassional help and synchronization from Atomics when necessary. Error handling and gathering of results is done using message passing, i.e., postMessage, which accepts transferable objects when passing large amounts of data.

How its used: Similar to map/reduce, you must provide some code that will be run on each thread to create the reader instance and send updates or submit the final result(s) back to the main thread. Set the .run property on the config object passed to the constructor, it will be eval'd on both the main thread and each of the workers (eval is necessary to allow access to global context, such as require, while providing consistent behavior across node.js and browser). Tasks are therefore intended only for running trusted code. See scan verb documentation for more information.

🍭 Features

v4.1.0 (2020-03-19)

Added RDF/XML Serializer as Scriber: @graphy/content.xml.scribe.

🍭 Features

v4.0.5 (2020-02-11)

Fixed missing dependency to @graphy/core.class.scribable for all writer packages.

🔧 Fixes

@graphy/content.*.write

• Added scribable class dependency.

v4.0.4 (2020-02-05)

Fixes #22, a bug when validating IRIs in the N-Triples and N-Quads readers.

🔧 Fixes

@graphy/content.n*.read

• IRI validation

v4.0.3 (2020-01-07)

Patch for negative lookbehind RegExp polyfills in all readers that affects certain browsers including Firefox.

🔧 Fixes

@graphy/content.*.read

• Lookbehind RegExp polyfills

v4.0.2 (2019-12-29)

Turtle and Trig reader performance patch.

⚡︎ Performance

@graphy/content.t*.read

• Faster lexing and validation for prefixed names.

v4.0.1 (2019-12-21)

Fixed tag selectors in quad filter expressions, and added a patch for reading large string literals.

🔧 Fixes

graphy CLI

• Tag selectors have been fixed.

@graphy/content.*.read

• Fixed reading very large string literals.

⚡︎ Performance

@graphy/content.*.read

• Optimized string literal lexing.

v4.0.0 (2019-12-11)

This update brings many new features to all the packages, with some necessary breaking changes, several fixes to the readers and writers, and performance improvements across the board.

BREAKING CHANGES