Skip to content
This repository has been archived by the owner on May 21, 2024. It is now read-only.

Latest commit

 

History

History
193 lines (136 loc) · 7.18 KB

CHANGELOG.md

File metadata and controls

193 lines (136 loc) · 7.18 KB
Raw

Changelog

7.6.0 (2023-09-18)

  • Support for Pachyderm v2.7.0.
  • Added check to prevent invalid file operations during transactions.

7.5.0 (2023-05-31)

  • Support for Pachyderm v2.6.0.
  • Add datum-batching feature:
    • @python_pachyderm.batch_all_datums
    • python_pachyderm.Client.worker.batch_datum
    • python_pachyderm.Client.worker.next_datum

7.4.0 (2023-02-23)

  • Support for Pachyderm v2.5.0 (Projects)
  • Add concurrency option for PutFileURL.

7.3.2 (2022-07-14)

  • Parity with Pachyderm 2.2.5

7.3.1 (2022-07-01)

  • Fix stable docs (#403)

7.3.0 (2022-06-01)

  • Add support for python3.10

7.2.1 (2022-06-01)

  • Pin protobuf dependency to >=3.17.1,<4.0.0

7.2.0 (2022-05-02)

  • Better handling of gRPC Connection errors when unable to connect to a pachyderm instance.
  • Added checks for pachctl when using mount operations and exposed python_pachyderm.experimental.util.check_pachctl method.

7.1.2 (2022-03-16)

  • Fix bug where server CAs were not expected to be b64encoded within the pachctl config file.

7.1.1 (2022-02-04)

  • Update protobuf files to v2.0.5
  • Fix bug where only 20MiB of data could be retrieved with client.get_file

7.1.0 (2021-11-18)

  • New experimental client (available via import python-pachyderm.experimental, which will contain beta/unstable APIs and features (#352)
  • New Mount() API in the experimental client, which runs pachctl mount in a subprocess to expose pachyderm data on the local filesystem (#353)

7.0.0 (2021-11-01)

  • Improved documentation

7.0.0-rc.1

  • Parity with Pachyderm 2.0.0, including support for Global IDs, and corresponding API changes
  • Support for Pachyderm's new spout semantics (added in Pachyderm 1.12)
    • Spout-manager has been consolidated into core client, as it's no longer necessary with this change.
  • Remove returns from methods whose RPC calls return a negligible protobuf (#328)
  • put_files supports putting individual files
  • Client() constructor reads connection settings from the Pachyderm config
  • API methods that take a commit now accept tuples of the form (repo, branch, commit_id, repo_type)
  • Support Pachyderm root tokens
  • Updated examples

6.1.0

  • Support for constructing a Client from a Pachyderm config file (PR #220)
  • Support for pachyderm v1.11.2 (PR #213)
    • Tweaks to debug service functions
  • Support for build-step enabled pipelines (PR #213)
  • Switched create_python_pipeline to use build-step enabled pipelines (PR #213)
  • Add support for authenticating with OIDC ID tokens (PR #228)
  • Clients automatically authenticate with an ID token in the PACH_PYTHON_OIDC_TOKEN env var (PR #236)
  • Add the metadata field to the response from create_pipeline (PR #234)
  • Add new_from_config to construct a client from a pachctl config file (PR #220)

6.0.0

  • Changes to the interface of enums dynamically generated from protos (PR #207)
  • Changes to the SpoutManager interface to better support commits (PR #206)
  • Added support for putting or deleting many PFS files in an atomic commit (PR #204)
  • Deprecated Client.put_file_bytes with an iterable of bytestrings (PR #204)
  • Added version number to build (PR #203)

5.0.0

  • Support for pachyderm v1.11 (PR #201)
    • Deprecated Client.get_admins and Client.modify_admins
    • Added support getting/setting cluster role bindings
    • Added support for OIDC login
    • Added the full flag to Client.inspect_job
    • Added sidecar_resource_limits for Client.list_job
    • Added Loki backend query support for log fetch functions
  • Better support for working with pipeline spec files (PR #200)
    • Added parse_json_pipeline_spec and parse_dict_pipeline_spec for parsing JSON spec files
    • Added create_pipeline_from_request for creating a pipeline from a parsed pipeline spec
  • Use sh rather than bash in python pipelines for better compatibility with custom images (PR #196)

4.2.0

  • Allow for customization of spout directory for testing (PR #188)
  • Fixed put_file_bytes failing for large files (PR #189)

4.1.0

  • Support file-like operations on get_file results (PR #187)

4.0.0

  • Support for pachyderm v1.10 (PR #178)
    • Added support for specifying pipeline metadata
    • Added support for s3 sidecar instances
    • Added support for secrets
  • First-class support for building spout producer and consumer pipelines (PR #173, #178)

3.1.0

  • Use pachd peer service where available in Client.new_in_cluster, to support TLS-enabled clusters (PR #176)

3.0.0

  • Support for pachyderm v1.9.11 (PR #170)
    • Removed scale_down_threshold from Client.create_pipeline
    • Added job_id to Client.run_pipeline
    • Added Client.run_cron
    • Added Client.batch_transaction

2.6.0

  • Faster put_files by putting multiple files in a single request (PR #166)

2.5.0

  • Added a utility function for creating a pipeline from locally stored python code (PR #163)
  • Added a utility function for recursively putting many files into PFS (PR #163)
  • Removed support for python 3.4 (PR #162)

2.4.1

  • Re-introduced PACH_PYTHON_AUTH_TOKEN, as we need it for JupyterHub integration (PR #161)

2.4.0

  • Synced with pachyderm core v1.9.8 (PR #159)

2.3.0

  • Tweaks to the way Clients can be initialized (PR #143, #157)
    • Removed support for the deprecated PACHD_ADDRESS environment variable
    • Removed support for library-specific PACH_PYTHON_AUTH_TOKEN environment variable
    • Added Client.new_in_cluster, which can be used to create a Client instance running inside a Pachyderm cluster
    • Added Client.new_from_pachd_address, which can be used to create a Client instance from a pachd address/URL
    • Added support for using default system certs
    • Added support for client initialization with a transaction ID
  • Added support for the enterprise service (PR #157)
  • Added support for the auth service (PR #157)
  • Added auth_token and transaction_id properties for better ergonomics (PR #157)
  • Renamed Client.metadata to Client._metadata, since it should be private (PR #157)
  • Some changes to Client.transaction to be more ergonomic and less redundant with other functionality (PR #157)

2.2.0

  • Added support for health (PR #156)
  • Added support for debug (PR #155)
  • Added full feature parity with PFS, PPS and transactions (PR #152, #151)
    • Added health_records to put_file methods
    • Added input_tree_object_hash to finish_commit
    • Added reverse to list_commit and list_branch
    • Added prov to subscribe_commit
    • Added full to list_job
    • Added spec_commit to create_pipeline
    • Added memory_bytes to garbage_collect
    • Added fsck, diff_file, delete_all_transactions

2.1.0

  • Added support for admin (PR #139)
  • Added support for transactions (PR #138)
  • Expose spout parameter for create_pipeline (PR #137)

2.0.0

  • Major, backwards-incompatible refactor. The largest change is to move all functionality into a single Client class. See this PR for more details: #129

1.9.7

  • Synced with pachyderm core v1.9.7
  • Note that this is the last version that will be pinned to pachyderm core versions. Future revisions will rely on semver. See the readme for details.