Release Notes

1.3.0

Release date: 2024-09-30

Enhancements

  • Support for PostgreSQL 9.6, 10, and 17: This adds support for the long EOL’d PostgreSQL versions 9.6 and 10. It is definitely discouraged to use these unsupported versions, but in case you’re for whatever reason unable to perform an upgrade you can now use the rdf_fdw.

1.2.0

Release date: 2024-05-22

Enhancements

  • Pushdown support for Math, String and Date/Time functions:

    • abs, ceil, floor, round
    • length, upper, lower, starts_with, substring, md5
    • extract(year from x), extract(month from x), extract(year from x), extract(hour from x), extract(minute from x), extract(second from x)
    • date_part(year, x), date_part('month',x), date_part('year', x), date_part('hour', x), date_part('minute', x), date_part('second', x)

    When used in the WHERE clause these functions will be translated to their correspondent SPARQL FILTER expressions.

Bug Fixes

  • Bug fix for WHERE conditions with “inverted” arguments - that is, value in the left side (T_Const) and column in the right side (T_Var): This fixes a bug that led the pushdown of WHERE condiditions containing “inverted” arguments to fail, e.g "foo" = column, 42 > column. Now the order of T_Const and T_Var in the arguments is irrelevant.

1.1.0

Release date: 2024-04-10

Enhancements

  • USER MAPPING support: This feature defines a mapping of a PostgreSQL user to an user in the target triplestore - user and password, so that the user can be authenticated. Requested by Matt Goldberg.

  • Pushdown suuport for pattern matching operators LIKE and ILIKE: these operators are now translated into SPARQL FILTER expressions as REGEX and pushed down.

  • Enables usage of non-pushable data types in FOREIGN TABLE columns: This disables a check that raised an excepetion when data types that cannot be pushed down were used. This includes non-standard data types, such as geometry or geography from PostGIS.

Bug Fixes

  • Empty SPARQL SELECT clause: This fixes a bug that led some SPARQL queries to be created without any variable in the SELECT clause. We now use SELECT * in case the planner cannot identify which nodes should be retrieved, which can be a bit inefficent if we’re dealing with many columns, but it is much better than an error message.

  • Missing schema from foreign tables in rdf_fdw_clone_table calls: This fixes a bug that led the rdf_fdw_clone_table procedure to always look for the given FOREIGN TABLE in the public schema.

  • xmlDoc* memory leak: The xml document containing the resulst sets from the SPARQL queries wasn’t beeing freed after the query was complete. This led to a memory leak that could potentially cause a system crash once all available memory was consumed by the orphan documents - which is an issue for rather modest server settings that execute mutliple queries in the same session. Reported by Filipe Pessoa.

1.0.0

Release date: 2024-03-15

Initial release.

Support for PostgreSQL 11, 12, 13, 14, 15 and 16.

Main Features

  • Pushdown: LIMIT, ORDER BY, DISTINCT, WHERE with several data types and operators
  • Table copy: This introduces the procedure rdf_fdw_clone_table(), that is designed to copy data from a FOREIGN TABLE into an ordinary TABLE. It provides the possibility to retrieve the data set in batches.
  • Proxy Support for SERVER: quite handy feature in case the PostgreSQL and SPARQL endpoint servers are in different networks and can only communicate through a proxy.