PostgreSQL 10 is supported from now on.

Parallel sequential scan mode is supported.

Indexing speed has been increased.

Log4j logging is implemented. Documentation

New cache parameters are provided for performance tuning of searches. Documentation

Peptide sequence file format was not handled correctly in the searches. From now on the 3-letter peptide format works in searches.

Duplicate search could be slow when the target structure contained charged hydrogen atom.

Speedup when query contains 5-membered ring with ambiguous aromaticity.

Wrong fingerprints were used after copying columns from one molecule subtype to another.

Similarity search did not always return hits in the order of similarity.

Molfiles with sgroups embedded in SRU groups were not handled.

Missing hits could have been experienced in tautomer duplicate search in case of InChi files.

jchem-psql service could crash if erroneous query structure was applied in a query.

Service jchem-psql stop did not work on some new Ubuntu versions.

Known issue present in version 2.8 has been solved.

Changing MapDB backend to MVStore approximately doubles the speed of the CREATE INDEX processes.

Similarity search did not always return hits in the order of similarity.

Molfiles with sgroups embedded in SRU groups were not handled.

Missing hits could have been experienced in tautomer duplicate search in case of InChi files.

jchem-psql service could crash if erroneous query structure was applied in a query.

Service jchem-psql stop did not work on some new Ubuntu versions.

The size of the index files under /var/lib/jchem-psql/store/ have been hugely increased.

Performance of select statements containing conditions for several rows or tables and including substructure search is improved by up to 20 times. Documentation

Missing hits could have been experienced in tautomer duplicate search in the case of molecules with aromatic rings.

Quick relevance sorting of substructure search hits is provided by using a new chemical indextype SORTEDCHEMINDEX. Documentation

Querying with R-group structures did not work without Markush Enumeration license.

New method providing fast similarity search has been implemented. Documentation

Molecules with small rings containing double bonds could miss duplicates when duplicate search was performed.

Tautomer duplicate search could result false negative hits.

Speed of insert into indexed table has been increased.

Index creation threw exception in case of molecules having deuterium atom.

Tautomer duplicate search could result false positive hits.

Reaction search is supported from now on. Documentation

Searching structures containing rings is sped-up.

JChem PostgreSQL Cartridge is supported on a distributed PostgreSQL Citus database. Documentation

Duplicate search could result failing hits in tables indexed with chemindex.

Misleading warning: WARNING: molecule_ops.cpp:205 Cost estimation values were not calculated has been fixed.

PostgreSQL 9.5 is the only supported PostgeSQL version from now on. PostgreSQL 9.4 is only supported in JChem PostgreSQL Cartridge versions up to 2.0. Documentation

Tetrahedral stereo properties were incorrectly evaluated in case of structures where a substituent on the chiral center - e.g., on phosphorus - was connected by double bond.

Searches with joined queries and with limit <n> conditions has been improved due to cost estimation and function mode search enhancements.

Limitation of the number of jchem-psql service sessions is configurable by the administrator. Documentation

Searches combining more than three SELECT statements with logical operators could run into dead lock state.

Relevance calculation threw exception in case of polymer and query structures.

Relevance calculation could throw license exception even if the postgresql license was present.

Custom standardizer.xml applied in type definition generated exception when more than one molecules were inserted in one statement.

Relevance sorting possibility of the search hits is provided. Documentation

Collecting invalid molecules during sdf import and (cx)smiles/(cx)smarts import is available.

New function, is_valid_molecule, is introduced.

Memory size requirement of searches in big tables has been decreased, and further performance tuning possibilities are provided. Documentation

Speed of searches - compared to version 1.6 - has been increased.

Specified requirement regarding glibc version for running JChem PostgreSQL Cartridge is canceled.

From now on, search engine in JChem PostgreSQL Cartridge works on vague bond level half. Previously, it worked on vague bond level 0.

glibc version 2.14 is needed to run JChem PostgreSQL Cartridge.

Tautomer substructure search could result false negative hits in case of aromatic structures.

Index entries were overridden by each other in some cases resulting loss of hits.

The use of chemical terms is implemented. Documentation

The performance of searching with combined queries can be tuned by calibration. Documentation