Vaex

* Features
* Allow casting integers to timedelta64 type [1741](https://github.com/vaexio/vaex/pull/1741)
* When a single task can fail, other can continue [1762](https://github.com/vaexio/vaex/pull/1762)
* Improved rich progress bar support [1771](https://github.com/vaexio/vaex/pull/1771)
* vaex.from_records to build a dataframe from a list of dicts [1767](https://github.com/vaexio/vaex/pull/1767)
* Settings in Vaex can be configured in a uniform way [1743](https://github.com/vaexio/vaex/pull/1743)
* Unique for datetime64 and timedelta64 expressions [1016](https://github.com/vaexio/vaex/pull/1016)
* Copy argument for binby, similar to groupby [4e7fd8e](https://github.com/vaexio/vaex/commit/4e7fd8e154c151323410cc1bedec96dd8a9667cb)
* Performance
* Improve performance for filtered dataframes [1685](https://github.com/vaexio/vaex/pull/1685)
* Fixes
* S3: endpoint override combined with globbing [1739](https://github.com/vaexio/vaex/pull/1739)
* Support having filtered and unfiltered tasks in 1 pass over the data [1761](https://github.com/vaexio/vaex/pull/1761)
* Continue next tasks even when old ones are cancelled [1769](https://github.com/vaexio/vaex/pull/1769)
* Handle empty arrow files [1773](https://github.com/vaexio/vaex/pull/1773)
* Evaluate and evaluate_iterator did not work for empty filtered dataframes [1775](https://github.com/vaexio/vaex/pull/1775)

* Performance
* Dot product with many columns does not use expressions, but dedicated function [1671](https://github.com/vaexio/vaex/pull/1671)

* Features
* OSX Metal support for jitting expressions [584](https://github.com/vaexio/vaex/pull/584)
* Improved progress support, including Rich progress bars [1738](https://github.com/vaexio/vaex/pull/1738)
* Control number of columns and rows being printed [1672](https://github.com/vaexio/vaex/pull/1672)
* Groupby with regular bins (similar to binby) [1589](https://github.com/vaexio/vaex/pull/1589)
* Groupby with a limited number of values, and 'OTHERS' [1641](https://github.com/vaexio/vaex/pull/1641)
* New aggregators: vaex.agg.any and vaex.agg.all [1630](https://github.com/vaexio/vaex/pull/1630)
* Better API for correlation and mutual information [536](https://github.com/vaexio/vaex/pull/536)
* Materialize datasets columns for better performance of non-memory mapping files (e.g. parquet) [1625](https://github.com/vaexio/vaex/pull/1625)
* Avoid using nest_asyncio [1546](https://github.com/vaexio/vaex/pull/1546)
* Multi level cache support (e.g. memory and disk) [1580](https://github.com/vaexio/vaex/pull/1580)
* Do not mutate dataframe when comparing dates. [1584](https://github.com/vaexio/vaex/pull/1584)
* Performance
* Fingerprint for tasks are more stable when the dataframe changes, but not the task description, for more cache hits. [1627](https://github.com/vaexio/vaex/pull/1627)
* Faster conversion between Arrow and NumPy [1625](https://github.com/vaexio/vaex/pull/1625)
* Cache sparse-finding/combining of high-d groupby [1588](https://github.com/vaexio/vaex/pull/1588)
* Allow (lazy) math and computations with aggregators [1612](https://github.com/vaexio/vaex/pull/1612)
* Less passes over the data when multiple dataframes use the same dataset [1594](https://github.com/vaexio/vaex/pull/1594)
* Share evaluation of expressions of selections [1594](https://github.com/vaexio/vaex/pull/1594)
* Delay support for groupby [1594](https://github.com/vaexio/vaex/pull/1594)
* Fixes
* Missing values in groupby were not well supported [1637](https://github.com/vaexio/vaex/pull/1637)
* Groupby over boolean [1632](https://github.com/vaexio/vaex/pull/1632)
* Negative periods for diff and shift [1608](https://github.com/vaexio/vaex/pull/1608)
* Arrow timestamp promotion during concatenation [1551](https://github.com/vaexio/vaex/pull/1551)

* Features
* Filelocks for multi process convert=True cooperation [1573](https://github.com/vaexio/vaex/pull/1573)
* Performance
* Fingerprint speedups [1574](https://github.com/vaexio/vaex/pull/1574)
* Expression.nunique asked unique for Python list (slow) [1576](https://github.com/vaexio/vaex/pull/1576)
* Groupby was slow for particular data (with low bits 0) [1571](https://github.com/vaexio/vaex/pull/1571)
* Blob encoding is faster due to blake3 [1575](https://github.com/vaexio/vaex/pull/1575)
* Stop iterating over dataset when exception occurs when computing [1577](https://github.com/vaexio/vaex/pull/1577)

* Features
* Protect file creation parts with lock files [1541](https://github.com/vaexio/vaex/pull/1541)
* Expression.str.extract to extract parts of strings using regex to a struct [1423](https://github.com/vaexio/vaex/pull/1423)
* Performance
* We now cache Expression.nunique() [1565](https://github.com/vaexio/vaex/pull/1565)
* Hashmaps memory is shared among threads (does not scale linear with number of threads), and avoids a merge phase [1525](https://github.com/vaexio/vaex/pull/1525)
* Hashmaps serialize efficiently [1525](https://github.com/vaexio/vaex/pull/1525)
* Avoid falling back to eval to get the dtype [1514](https://github.com/vaexio/vaex/pull/1514)

* Performance
* Reuse filter data when slicing a dataframe [1287](https://github.com/vaexio/vaex/pull/1287)
* Faster astype('str') [1411](https://github.com/vaexio/vaex/pull/1411)
* Task refactor, which allows for more caching [1433](https://github.com/vaexio/vaex/pull/1433)
* Features
* df.to_records() for output to JSON [1364](https://github.com/vaexio/vaex/pull/1364)
* df.dt.quarter and df.dt.halfyear [1441](https://github.com/vaexio/vaex/pull/1364)https://github.com/vaexio/vaex/pull/1441)
* Arrow struct support [1447](https://github.com/vaexio/vaex/pull/1364)https://github.com/vaexio/vaex/pull/1447)
* Fixes
* df.concat did not copy functions [1287](https://github.com/vaexio/vaex/pull/1287)
* Dropping columns when name was invalid identifier [1434](https://github.com/vaexio/vaex/pull/1434)
* Old dates wrapped due to negative ints and casting to unsigned [1504](https://github.com/vaexio/vaex/pull/1504)
* Timestamp to NumPy datetime64 would ignore units [1513](https://github.com/vaexio/vaex/pull/1513)
* Thread unsafety could trigger issues in Plotly dash [1512](https://github.com/vaexio/vaex/pull/1512)