v0.7.1

New features:

• Allow to configure spark hashing job using environment variables #207

v0.7.0

Breaking changes:

• Remove empty files from hashing #201
• Exclude very small files from file similarity hashing #201

New features:

• Remove duplicated repositories #200

Fixes:

• hash/query: Use floats instead of doubles in wmh #196
• hash: Filter files with null language #198
• hash: Fix segfault on macOs Mojave #204
• report: Multiple performance optimization #197

Also documentation explains current limitation, performance tips and list of known bugs now.

v0.6.0

Breaking changes:

• New version is incompatible with old feature-extractor

New features:

• feature-extractor: New Extract method to run multiple extractors on the same UAST in one call #188
• feature-extractor: Use multiple processes for extractors #193

Fixes:

• Filter out too long feature names #192
• Update jgit-spark-connecter to support remote standard/bare repositories
• Performance improvements on hashing part #194

v0.5.0

Breaking change:

• Hashing excludes vendor files now

New features:

• Number of workers in feature-extractor service is configurable now

v0.4.0

Breaking changes:

• DB schema is changed and incompatible with the previous one

New features:

• add support for Amazon Web Services S3 #183
• require empty DB keyspace to run hash #184

Bug fixes:

• change DB schema for document frequencies to avoid too large mutation #185

v0.3.0

Breaking changes:

• remove --format flag from report
• add new --cassandra flag which behaves like --format=group-by before

New features:

• add support for Google Cloud Storage #182
• add json output format for report #181

v0.2.0

The main features

Function level similarity

Other improvements:

• Update jgit-spark-connecter (get list of supported languages from bblfsh)
• Keep document frequencies in DB instead of JSON file
• Drop go client

v0.1.0

The main features

Apache Spark application performance tuning on actual cluster:

• hashuses more performant Apache Spark conf defaults for running on cluster
• hash more performance improvements: JSON parser, broadcast DocFreq
• hash CLI has Feature Extractor error reporting summary

Other improvements:

• CLI options parsing refactored and is shared among all commands
• UX: CLI lists actual .siva files to be processed by engine
• Golang example of querying for duplicates updated
• Smaller bug fixes and improvements

v0.0.4

The main features

Packaging of the release as .jars and not using SBT to run it any more

Other improvements:

• better logging on -v

v0.0.3

The main feature:

This release brings file similarity to hash.

Other improvements:

• docker-compose added to make it easier to try Gemini
• use new parameters for file similarity provided by ML team
• documentation is slightly improved