This changelog is deprecated. All changes are documented under releases.

Improved handling of tags with maps for Aspen IP.21.

• Allow tag search with tagname and description.
• Fix fetching of correct unit for tags with map.
• Fix fetching of correct description for tags with map.

• Fix FutureWarning regarding use of keep_tz=False in DateTimeIndex.to_series()

First release on PyPI as tagreader.

• Rename from pyIMS to tagreader to avoid name collision on PyPI.
• Move lots of tests to a separate private repository. Those tests need to be run on-site with functioning server connection to PI/IP.21 servers, and are currently handled in Azure Pipelines.

Last release under the name pyIMS.

• Major rewrite to accommodate for other types of connections than ODBC in the future.
• Breaking: It is now necessary to specify imstype (pi or aspen) when initiating a client.
• Update to new default DAS server address.

• Force keep_tz=False to suppress FutureWarning in Pandas >= 0.24.
• Fix reading of digital states from PI.

• Add get_units() and get_descriptions() methods to fetch engineering units and description for specified tags.
• Enable tag search by description.
• Package version made available in pyims.__version__.

• Improve documentation.
• Add commonly requested documentation, including docstring, for specifying query method for read_tags.

• Fix bug where timezone was locked to Europe/Oslo.
• Discovered that Pandas >=0.23 is required. 0.22 causes a weird timezone bug during DST folds when using cache.

• Aspen now supports avg, min, max, rng, std and var queries. These have timestamps in the middle of the period.

• Start preparation for supporting more read types for both Aspen and PI.
• Define max_rows = 100000 for PI to (hopefully) avoid timeouts.

• Fix indexing issue (duplicate indices appeared in some cases, but are now handled).

• Introduce time zones. Start- and stop time for queries are always performed relative to server time, so we have to know where the server is located. The default time zone is "Europe/Oslo", but this can be changed for e.g. Peregrino by specifying tz="America/Sao_Paulo" upon initialization of the client.

• Cache files produced with versions < 0.0.5 must be deleted since they are not compatible the time zones.
• Interpolated queries for IP.21 now return extrapolated values after last valid datapoint.

• Properly handle transitions to and from DST (Daylight savings time).
• Queries should now always return data for exactly start_time <= time <= stop_time. Off-by-one data (e.g. no data for stop_time or data for stop_time+ts) is a bug (except for PI queries with stop_time close to DST changes, in which case PI may act silly).

• Fix cache sorting issue when reading across two segments that were previously written in nonchronological order.

• Add support for tags with mappings (relevant for SNA and SNB) by specifying tags on the form tag;map
• Fix NaturalNameWarning when using '.' in tagnames

Initial release. Supports the most common tag read queries for IP.21 and PI.