0.6.22

What's Changed

• fix: notebook outputs by @imathews in #20

Full Changelog: 0.6.21...0.6.22

0.6.21

• fix: require arrow >= 16.0, fix build issue on m-series macs

0.6.20

• perf: improve memory efficiency for loading in-memory tables

0.6.19

What's Changed

• refactor: notebook output tables by @imathews in #19

Full Changelog: 0.6.18...0.6.19

0.6.18

• fix: set upper parallelization limit when loading tables

0.6.17

• fix: implement retry handling for broken connections when reading records

Full Changelog: 0.6.16...0.6.17

0.6.16

What's Changed

• fix: parallel execution by @imathews in #18

Full Changelog: 0.6.15...0.6.16

0.6.15

What's Changed

• perf: improve memory allocation and parse performance for in-memory datatypes by @imathews in #17

Full Changelog: 0.6.14...0.6.15

0.6.14

• perf: reduce memory consumption for multiprocess threading

0.6.13

• refactor: don't write to disk for in-memory data structures
  • Previously, we would stream data to disk, and then load it into a tibble, etc (except for the to_arrow_dataset method, which is never loaded into memory). This simplified a lot of the logic, especially when performing parallelization, but could ultimately eat up a lot of disk, and may cause issues when a system's memory exceeds the available disk. This issue has now been resolved.
• fix: fix bad parameters for to_arrow_table()