Skip to content

Commit

Permalink
Updated the change log file.
Browse files Browse the repository at this point in the history
  • Loading branch information
rhdekker committed Oct 11, 2015
1 parent eb98584 commit 7c25cb2
Showing 1 changed file with 30 additions and 2 deletions.
32 changes: 30 additions & 2 deletions changelog.txt
Original file line number Diff line number Diff line change
@@ -1,11 +1,27 @@
Release 1.7
- Major rewrite of the matching phase of the Dekker algorithm. The MatchTable approach used in versions 1.2 - 1.6 of
CollateX would match the tokens of the witness to be aligned against the tokens present on the vertices of the variant
graph one witness at a time. This approach did not scale very well and it was unable to look ahead. The new
implementation of the matching phase builds a token index of all tokens of all witnesses before the alignment phase
based on sorting the tokens. This approach is faster and returns a complete overview of all the possible matches
including 1) the number of tokens present in a block, 2) in how many witnesses the block occurs and 3) how frequent
the block occurs in the complete witness set.
- Optimizations of the alignment phase of the Dekker algorithm. A priority queue is now used instead of a multiple
value map to select the best possible match. The detection of possible overlap between possible matches now has a fast
path in case of complete overlap or no overlap between two possible blocks.
- The alignment phase of algorithm (including the transposition detection) has stayed the same. There are same quality
improvements however based on the fact that the possible matches are now based on the tokens of the complete witness
set.

Release 1.6.2
- Added servlet module for easy deployment on servlet containers such as the Apache Tomcat webserver. Implementation is based on JAX-RS and Jersey 2.
- Added servlet module for easy deployment on servlet containers such as the Apache Tomcat web-server.
Implementation is based on JAX-RS and Jersey 2.

Release 1.6.1
- JSON processing: fix regression bug due to which "tokenComparator" and "algorithm" turned into mandatory fields

Release 1.6
- new aligorithm based on Greedy String Tiling
- new algorithm based on Greedy String Tiling
- Java 8 now required
- provide our own implementation of variant graphs
- remove optional and seldomly used integrations with Neo4j and Apache Cocoon
Expand Down Expand Up @@ -43,3 +59,15 @@ This approach improves the alignment quality greatly in case of longer witnesses
- New ColateX Tools package, containing a web-service based on JAX-RS.
- New Command-line interface.

Release 1.2
- Beta release of the MatchTable approach (only released internally to beta testers).

Release 1.1
- First release to allow a custom matching function by supplying a comparator<Token> function to the aligner.

Release 1.0
- First release to align against a variant graph instead of an alignment table. The alignment table is now a
visualization of the variant graph.

Release 0.9
- First public release (beta quality).

0 comments on commit 7c25cb2

Please sign in to comment.