Woot! Thank you, Sung, for managing the release! I'm very excited about this new version.
I want to highlight the many contributors who have improved PyIceberg since the last release. There have been 34 unique contributors (source <https://github.com/apache/iceberg-python/compare/pyiceberg-0.6.1...pyiceberg-0.7.0rc2>). Also, as seen in issue #511 <https://github.com/apache/iceberg-python/issues/511>, the community has come together to contribute various metadata table implementations to PyIceberg. Onwards and upwards, Kevin On Tue, Jul 30, 2024 at 12:08 PM Sung Yun <sungwy...@gmail.com> wrote: > I'm pleased to announce the release of Apache PyIceberg 0.7.0! > > Once again, this large release includes the following features on a high > level: > > * Write support to partitioned tables with IdentityTransform and > TimeTransform partitions > * Support for deletes using predicates. It will drop whole files when it > is able to based on the Iceberg statistics, otherwise it will perform a > copy-on-write. > * Parallelizing writes for a given partition based on a target file size > * A new API for rendering PyArrow tables that show metadata about the > tables’ manifests, partitions, etc > * Support for evolving table partitions > * Updated schema compatibility check to be more permissive, by supporting > promotable types and subset of schemas on write > * Option to merge manifests on write when number of manifests exceeds a > threshold > * Support staging a table for creation and building a transaction > * A new table scan API to return an Arrow RecordBatchReader as opposed to > a fully materialized Arrow table > * Support for categorical and large PyArrow types on write > * A new API to add existing parquet files to a table without rewriting them > * Support for loading custom catalog > > This Python release can be downloaded from: > https://pypi.org/project/pyiceberg/0.7.0/ > > Thank you everyone again for the amazing contributions and engagement > since the last release. > > Sincerely, > Sung >
