Skip to content

Releases: Bears-R-Us/arkouda

Release Notes v2022.09.30

30 Sep 20:56
f5e7940
Compare
Choose a tag to compare

Release Notes 2022-09-30

Major updates:

  • Issue #1814 - Changes behavior of search_intervals and interval_lookup to have closed upper bounds
  • Issue #1810 - Adds get_prefixes and get_suffixes to Strings
  • Issues #1787, #1778 - Add GroupBy.std(), GroupBy.var(), and GroupBy.median()
  • Issue #1808 - Fixes runtime errors in ak.in1d()
  • Issue #1794 - Adds documentation of similarities/differences with Numpy & Pandas
  • Issues #1781, #1820 and PRs #1813, #1807 - Make Arkouda compatible with Chapel 1.28
  • Issues #1709, #1745, #1743, #1767, #1765, #1746, #1761, #1760, #1755, #1763, #1757, #1731 - Update server messages to use JSON arguments

Minor fixes:

  • Issue #1742 - Updates symbol table structure
  • Issue #1774 - Updates message function definitions
  • Issues #1681, #1783 - Move SegArray.__init__ and SegArray.__getitem__ to server
  • Issues #1650, #1751, #1747 - Add chapel version to get_config and splash message and add connect url to server connection info
  • Issues #1752, #1749 - Update README and CONTRIBUTING.md

Auto-generated release notes

Full Changelog: v2022.08.30...v2022.09.30

Release Notes v2022.08.30

30 Aug 20:09
e96c533
Compare
Choose a tag to compare

Release Notes 2022-08-30

Major updates:

Minor fixes:

  • Issue #1613 - Enables regex with patterns that match the empty string
  • Issue #1676 - Fixes Strings.to_ndarray for empty Strings
  • PR #1732 - Adds client mode environment variable
  • PR #1720 - Adds logic to call exit(0) only when there are 2..n ArkoudaServerDaemons

Auto-generated release notes

Full Changelog: v2022.08.15...v2022.08.30

Release Notes v2022.08.15

15 Aug 18:25
e6a584d
Compare
Choose a tag to compare

Release Notes 2022-08-15

Major updates:

  • Issues #1618, #1630, #1631, #1632, #1654 - Improve JSON parameter object multiple dtype support and list processing, add nested JSON message argument support, and add a dispatch table for message argument generation
  • Issue #1659 and PRs #1601, #1602, and #1604 - Refactor arkouda_server to be a driver of ServerDaemons and enables external and metric integration
  • Issue #1639 - Adds aggregations for first, mode, and unique to Groupby
  • Issue #1643 - Adds covariance and correlation methods for pdarray and correlation matrix method for DataFrame
  • PR #1635 - Reduces sort bucket exchange overhead
  • Issue #1666 - Removes support for Chapel 1.25

Minor fixes:

  • Issue #1397 - Replaces current DataFrame display method with server implementation
  • Issues #1648 and #1658 - Update docs to address issues with brew install chapel and add instructions to update environment
  • Issue #1642 - Fixes testAllOperators double counting mismatches
  • Issues #1589 and #1672 - Add test to verify randomness for ak.randint (disabled for the time being)
  • PR #1669 - Adds read support for Parquet uint32

Auto-generated release notes

Full Changelog: v2022.07.28...v2022.08.15

Release Notes v2022.07.28

29 Jul 00:50
afd0735
Compare
Choose a tag to compare

Release Notes 2022-07-28

Major updates:

  • Issue #702 - Adds multidimensional object support for HDF5
  • Issue #1495 - Adds multi-array support for in1d
  • Issue #1168 - Adds DataFrame Index renaming
  • Issue #1620 - Fixes uint and float binops
  • Issue #1607 - Re-enables MultiIndex support for mixed dtypes
  • Issue #1614 - Corrects DataFrame Index error on .append()
  • Issue #1220 - Fixes modulo bias in randint
  • Issue #1519 - Adds strip method to Strings
  • Issue #1598 - Allows for overlapping intervals and adds tiebreak to search_intervals
  • Issue #1605 - Adds to_list method to classes which have to_ndarray
  • Issue #1587 - Adds support for JSON formatted argument for server messages
  • Issues #1637, #1622, #1645 - Correct floor division on edge cases

Minor fixes:

  • Issues #1161 and #1573 - Update make test-chapel to run unit tests
  • Issue #1588 - Fixes bug in __getitem__'s argument validation
  • Issue #1583 - Optimizes string indexing
  • Issue #848 - Limits length of file lines logged
  • Issues #1625 and #1627 - Set sphinx minimum version to 5.1.1
  • Issue #1572 - Updates modular build process to initialize modules on command line
  • Issue #1600 - Fixes bug where SegArray.to_ndarray returned list

Auto-generated release notes

Full Changelog: v2022.07.08...v2022.07.28

Release Notes v2022.07.08

08 Jul 21:41
b4e5984
Compare
Choose a tag to compare

Release Notes 2022-07-08

Major updates:

  • Issues #1543, #1545, #1546, #1547 - Recommend Chapel 1.27.0 and drop support for Chapel 1.24
  • Issues #1525 and #1529 - Update to require numpy>=1.22.2
  • Issue #1301 - Updates CONTRIBUTING.md to specify latest guidelines for contributing
  • Issue #1386 - Adds generic unregister_by_name() method
  • Issue #1527 - Adds is_ipv4() and is_ipv6() for ipaddresses
  • Issue #1548 - Fixes interval_lookup bug and enable multi-array input
  • Issue #1512 - Fixes bug in Series methods with MultiIndex

Minor fixes:

  • Issue #1535 - Optimizes lookup algorithm
  • Issue #1518 - Adds to_title and is_title methods to Strings
  • Issue #1540 - Adds .objType and .dtype to Categorical
  • Issue #1565 - Adds Index.is_unique
  • Issues #1478 and #1465 - Add typechecking to Series and Join
  • Issue #1360 - Updates ak.save_all() to utilize append
  • Issue #1520 - Updates make to recompile Arrow
  • Issue #1511 - Fixes make install-deps inconsistencies
  • Issue #1362 - Fixes truncation of BitVector values in Series display
  • Issue #1353 - Makes size() an alias for the current count()
  • Issue #1536 - Switches to using a lowLevelLocalizingSlice for DF indexing message
  • Issues #1514 and #1555 - Rename hash() to siphash() and remove trailing comma from 1 tuple for compatibility with Chapel main

Auto-generated release notes

Full Changelog: v2022.06.13...v2022.07.08

Release Notes v2022.06.13

13 Jun 17:51
f9a7fa2
Compare
Choose a tag to compare

Release Notes 2022-06-13

Major updates:

  • Issues #1505 and #1506 - Fix Series loss of support for Index and multi-index
  • Issue #1491 - Adds uint64 support to search_intervals()
  • Issue #1077 - Updates dtypes to better support scalars (8, 16, and 32 bit)
  • Issue #612 - Changes minimum supported python version to Python 3.8
  • Issue #1470 - Improves ak.Index (adds ak.Index.values, improves argument validation, etc)
  • Issue #1448 - Cleans up util.py (moves most_common() into GroupBy and deprecates functionality duplicated elsewhere in arkouda)

Minor fixes:

  • Issue #1486 - Adds Flake8 to CI
  • Issues #1480, #1481, and #1479 - Fix deprecation and Pandas cast warnings
  • Issues #1499, #1371 and PR #1504 - Update requirements for Pandas and mypy, and swap channel order in conda yaml files
  • PR #1490 - Updates additional files to avoid using SysCTypes, CPtr
  • PR #1498 - Start timing array transfer again

Auto-generated release notes

Full Changelog: v2022.06.06...v2022.06.13

Release Notes v2022.06.06

06 Jun 19:05
877747b
Compare
Choose a tag to compare

Release Notes 2022-06-06

Major updates:

  • Issues #1390, #1428, and #1471 - Make parquet a dependency
  • Issues #1412, #1415, #1420, #1429, #1431, #1435, #1439, #1457, #1455, #1459 - Apply black and isort to arkouda's client code, benchmarks, and tests. Ensure reformatting meets PEP8 using flake8
  • Issue #1350 - Adds ak.DataFrame.isin method
  • Issues #953 and #1450 - Add Import/Export Functionality between arkouda and pandas
  • Issue #1167 - Adds inplace parameter to drop, rename, and reset_index in Dataframe
  • Issue #1456 - Enables saving with a SegArray column and adds file_format to df.save()
  • Issues #1122, #1337, #1381, #1382, #1389, #1399- Add register and attach functionality to Groupby; is_registered to Series, Segarray, and Index; and generic attach support for Categorical, Segarray, and Series
  • Issues #1302, #1377, and #1454 - Update setup instructions and add YAML files for creating user and developer conda environments
  • Issue #1452 - Adds support for argmin and argmax on bool values
  • Issue #1387 - Fixes bug with relational binops on booleans
  • Issue #1393 - Reimplements assume_sorted shortcut for GroupBy and unique left out of refactor
  • Issue #1394 - Forces ak.concatenate to preserve fancy pdarray types (BitVector, IPv4, ...)
  • Issue #1406 - Fixes bug where ak.lookup assumes ordered input
  • Issue #1401 - Fixes bug in DataFrame save_table
  • Issues #1384 and #1379 - Fix instances of accessing index as as a column, allow making Series from a Strings, and allow indexing a Dataframe with a tuple of column names

Minor fixes:

  • Issue #1363 - Adds more robust argument validation to ak.Series init
  • Issues #1404 and #1446 - Improve radix sort performance and testing
  • Issues #1398 and #1424 - Work on DataFrame Display Performance Issue
  • PR #1444 - Optimizes domain_intersection() when both domains are non-strided
  • Issue #1417 - Limits multi-locale CI testing
  • Issue #1426 - Fixes allocated buffer size for akVersion
  • Issue #1433 - Sets mypy==0.931
  • PR #1438 - Updates Math module to be compatible with changes to Chapel
  • Issue #1437 - Removes CyclicDist as an option for pdarrays

Auto-generated release notes

New Contributors

Full Changelog: v2022.05.09...v2022.06.06

Release Notes v2022.05.09

09 May 16:12
6cc42c3
Compare
Choose a tag to compare

Release Notes 2022-05-09

Major updates:

  • Issues #1365 and #1366 - Create nonexistent Parquet and HDF5 files when writing with mode='append'
  • Issue #1331 - Align Parquet save/load with HDF5

Minor fixes:

  • Issue #1351 - Aligns ak.Series.shape with pandas functionality
  • Issue #1367 - Removes akutil subdirectory since it has been merged into the main arkouda python package
  • Issues #1345 and #1369 - Add DataFrame Display benchmark to nightly performance charts

Auto-generated release notes

Full Changelog: v2022.05.05...v2022.05.09

Release Notes v2022.05.05

05 May 16:55
a073707
Compare
Choose a tag to compare

Release Notes 2022-05-05

Major updates:

  • Issue #1008 - Introduces Multi-dimensional ArrayView objects with indexing and slicing
  • Issue #846 - Enables calling unique on lists/tuples of arrays and significantly improves performance of GroupBy on a large number of arrays
  • PR #1313 - Adds functions for setting explicit categories and standardizing categories across multiple Categoricals and adds an N/A value to Categorical
  • Issue #1280 - Adds multiarray setops support for all groupable types
  • Issue #1086, #1282 - Adds SegArray register/attach and a generic ak.attach method
  • Issues #1273 and #1310 - Incorporate akutil plotting and import akutil updates (Dataframe to handle aggregate ops, additional Series functionality, and addition of Accessor classes)
  • Issues #1311, #1317, #1318, and #1339 - Fix bugs in array dtype on python lists, uint64 pdarray ops, string comparison with empty string, and Dataframe index with logical indexing
  • PRs #1288, #1293, #1307, and #1299 and Issues #1303, #1324, #1229, and #1295 - Unify file reading functions into a single read function; add file_format keyword to read, save, and load functions; add a get_null_indices function for Parquet files; and improves efficiency of save

Minor fixes:

  • Issue #1300 - Adds a candidate DataFrame Display and a benchmark to compare performance with the current method
  • Issues #1275, #1199, and PR #1343 - Update Installation documentation for WSL, add dynamic versioning into docs, and add an arkouda talk to README
  • Issues #1290 and #1325 - Update documentation for ak.histogram and add documentation for segarray setops
  • Issues #1268, #1236, #1230, and #1286 - Clean up code in strings, dataframes, and groupby tests
  • PR #1323 - Adds sanitizer check for C++ compilation in Makefile
  • Issue #1315 - Adds better error handling in DataFrame.dtypes

Auto-generated release notes

New Contributors

Full Changelog: v2022.04.15...v2022.05.05

Release Notes v2022.04.15

15 Apr 14:44
563867b
Compare
Choose a tag to compare

Release Notes 2022-04-15

Major updates:

  • Issue #1218 - Extends pdarray setops to work on multiple pdarrays
  • Issue #1197 - Adds Segarray setops functionality
  • Issue #1234 - Removes ls_hdf in favor of a generic ls which automatically handles hdf5 or parquet files
  • Issue #1265 - Adds to_upper/to_lower and is_upper/is_lower functionality to Strings
  • PRs #1222, #1227, #1233 - Add support for writing string Parquet files and an append mode for Parquet file writing
  • Issue #1272 - Changes ak.histogram to behave like akutil.hist
  • Issue #1256 - Moves akutil.join functionality into arkouda
  • Issues #1133, #1210 and, #1279 - Fix uint64 indexing and broadcasting errors
  • Issue #1260 - Enables GroupBy with one of more boolean pdarrays
  • Issue #1240 - Updates client dtype classes to be uint compatible and enables ak.ip_address to accept python lists
  • Issue #1154 - Adds dtype parameter to ak.array

Minor fixes:

  • Issue #1174 - Fixes randint range bug
  • Issue #1190 - Updates ak.Dataframe to use the Index object
  • Issues #1039 and #1241 - Improves performance for String methods (peel, stick, and substring search)
  • Issue #1251 and PR #1243 - Update README install instructions and adds documentation for --saveUsedModules flag
  • PRs #1219 and #1231 - Improves performance of parquet read and append

Auto-generated release notes

Full Changelog: v2022.03.15...v2022.04.15