Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

unreleased

0.15.0 - 2025-02-25

Added

• Temporary ._token_refresh_lock asyncio.Lock bound to the new event loop for synchronous request invocations.
• DELETE request to the ChemCloud server after successfully receiving ProgramOutput.
• Adaptive polling interval to FutureOutput.get_async() so that the interval is reset if results are collected.

Removed

• _HTTPClient._httpx_timeout.

0.14.0 - 2025-02-20

Added

• Full async API for chemcloud. This enables Jupyter support. All chemcloud methods have an _async version that can be awaited as follows:

from chemcloud import compute_async

...
output = await compute_async("terachem", prog_inputs)

0.13.3 - 2025-02-19

Added

• FutureOutput.save() and FutureOutput.open() methods to serialize futures to disks and then open the later for retrieval.

Changed

• FutureOutput.single_input to .return_single_output.

0.13.2 - 2025-02-18

Added

• FutureOutput.single_input so that list of len 1 submitted to compute() still get returned as list. May want to refactor into FutureOutput and FutureOutputs, but this solution works for a quick fix for today's presentation.

0.13.1 - 2025-02-17

Added

• Usage examples for main chemcloud functions.
• Tests for main chemcloud functions.
• queue can be passed to configure_client() or CCClient() to specify that all calculations should be submitted to a particular queue.

Changed

• ProgramOutput objects are removed from the FutureOutput.output list as they are returned by .as_completed() to reduce memory footprint for collecting large batches of results.

0.13.0 - 2025-02-16

Added

• High level compute() interface for doing calculations without needing to first instantiate a client.
• More complete TaskStatus enum.
• Parallel task submission and retrieval with ChemCloud server via asynchronous coroutines.
• asyncio.Lock() to auth requests so that token refreshes are not duplicated by multiple coroutines.
• .save() and .open() methods to FutureOutput to serialize to/from disk.

Changed

• Refactored CCClient, _RequestsClient and FutureOutput to have cleaner interface and more clearly defined roles.
• CCClient.configure() renamed to .setup_profile() to improve clarity and reserve the term configure for the new configure_client function.
• compute() and client.compute() are now synchronous by default. If end users want to run calculations asynchronously and have a future object returned pass return_future=True.
• _RequestsClient is entirely async now with a .run_parallel_requests() method that can be used for synchronous tasks.
• Lists of ProgramInput objects submitted to CCClient.compute() are now submitted individually (and concurrently) rather than as a batched, single request. This will prevent the server from having to retrieve large batch calculations which occasionally fills server memory and crashes it.
• _HTTPClient and updated requests methods to reuse these clients across requests rather than creating a new connection for each request. This enables connection pooling and reuse of TCP/SSL connections, greatly speeding up requests and lowering server load.
• Refactored and renamed methods on _HTTPClient for clarity and maintainability.
• Updated documentation to use new compute() API instead of instantiating a client.

Removed

• TaskStatus.complete. TaskStatus has been expanded to mirror celery's states.
• FutureOutputGroup in favor of a single FutureResult object that can handle both single tasks and arrays of tasks.
• Convenience types defined in models.

0.12.5 - 2025-02-12

Fixed

• Retry logic for httpx.RequestError now correctly raises an exception after max_attempts. Requests that fail due to sporadic connection/network issues will now be retried up to max_attempts times.

Changed

• Removed retry logic for 4xx and 5xx errors.

0.12.4 - 2025-02-11

Added

• Retry requests for 4xx and 5xx status codes. This is a short term fix for the sporadic 500 errors related to LookupError: unknown encoding: idna on the server.

0.12.3 - 2025-02-10

Added

• Retry loop for HTTP requests to handle ssl handshake errors or other non-HTTP status code errors.

0.12.2 - 2025-02-07

Changed

• Loosened dependency specifications for qcio and httpx.

0.12.1 - 2024-09-11

Changed

• Using FutureOutput for arrays of inputs of length 1.
• Updated to Python 3.9 typing syntax.

0.12.0 - 2024-08-06

Changed

• Removed black and isort in favor for ruff

Updated

• 🚨 python (>=3.8 -> >=3.9)
• tomli (^1.0 -> ^2.0)
• httpx (^0.23.1 -> ^0.27)
• ruff (^0.0287 -> ^0.5)
• pytest-httpx (<0.23.0 -> >=0.23.0)
• format.sh script dropped black and isort.

0.11.1 - 2024-07-19

Changed

• Updated to qcio 0.11..

0.11.0 - 2024-07-12

Changed

Updated qcio (0.10.1 -> 0.10.2). Structure.ids -> Structure.identifiers

0.10.1 - 2024-07-11

Added

• py.typed file for type checking in projects that depend upon chemcloud.

0.10.0 - 2024-07-10

Changed

• Updated to qcio 0.10.1 which uses Structure instead of Molecule.

0.9.0 - 2024-06-14

Changed

• Upgraded to qcio Generics data structures.
• Upgraded black from ^23.0.0 to ^24.0.0.
• Renamed collect_wavefucntion kwarg to CCClient.compute(...) to collect_wfn.
• Updated the response returned by ChemCloud server to have attributes status and program_output from state and results.
• Rebuilt documentation to reflect new qcio Generics and renamed kwargs.

0.8.3 - 2023-10-20

Changed

• Updated GitHub actions to run on pull_request in addition to push.
• Migrated dependency uiri/toml to hukkin/tomli (Python < 3.11) or the built-in tomllib (Python >= 3.11)
• Updated GitHub actions to test against both Python 3.8 and Python 3.11

0.8.2 - 2023-09-25

Changed

• Removed pydantic==2.4.0 from supported versions due to immediate bug upon this release. When trying to import CCClient a reference count error appears to trigger a KeyError. Bugs are being tracked here: https://github.com/pydantic/pydantic/issues/7617.

0.8.1 - 2023-09-20

Changed

• Updated to qcio>=0.7.0 to account for renaming of DualProgramArgs to SubProgramArgs.

0.8.0 - 2023-09-19

Changed

• FutureResult objects now called FutureOutput to keep in harmony with qcio nomenclature.
• Documentation rewritten to capture API changes with qcio and qcop.
• /examples scripts instantiate a Molecule directly rather than opening h2o.xyz so that code examples can be run directly from the documentation website.
• All mkdocs and associated documentation packages updated to the latest versions.

0.7.0 - 2023-09-08

Changed

• Dropped QCElemental in favor of qcio.
• Updated client to work with v2 of the ChemCloud server using qcio.
• Updated DevOps stack to be in harmony with other qc* packages (poetry, GitHub Actions, pre-commit, etc).
• Updated from pydantic v1 -> v2, added pydantic-settings to dependencies.
• Changed settings chemcloud_default_credentials_profile to chemcloud_credentials_profile.
• Updated a few names from result to output to be more in harmony with qcio nomenclature.

Added

• Publish to pypi from GitHub actions.

0.6.2 - 2022-12-27

Changed

• Updated qcelemental==0.24.0 -> 0.25.1

0.6.1 - 2022-07-19

Changed

• Pegged qcelemental to version 0.24.0 since 0.25.0 introduces breaking changes. Need to keep this version in sync with ChemCloud server version.

0.6.0 - 2022-07-19

Changed

• Updated project name from qccloud to chemcloud

0.5.0 - 2022-07-15

Changed

• Updated project name from tccloud to qccloud

0.4.1 - 2022-05-07

Changed

• Upped the default timeout on http reads from 5.0s -> 20.0s.

0.4.0 - 2022-4-02

Added

• to_file() and from_file() methods to easily save compute job ids for later retrieval.

Changed

• Simplified management of task ids between client and server. Only need to send a single id to server even if a batch computation was initiated.

Removed

• Support for Python3.6. Python3.6 end-of-lif'ed December 23, 2021.

0.3.1 - 2022-03-27

Added

• Decode b64 encoded data returned from server in AtomicResult.extra['tcfe:keywords']

Changed

• Updated config.settings.tcfe_config_kwargs = "tcfe:config -> config.settings.tcfe_keywords = "tcfe:keywords

0.3.0 - 2022-03-26

Added

• Support for AtomicInput.protocols.native_files. User can now request QC package specific files generated during a computation.
• Added support for TeraChem-specific native_files. c0/ca0/cb0 bytes files (or any bytes data) placed in AtomicInput.extras['tcfe:keywords'] will be automatically base64 encoded and sent to the server. The enables seeding computations with a wave function as an initial guess.
• Base64 encoded native_files returned from server will be automatically decoded to bytes.

0.2.4 - 2021-06-07

Added

• Private compute queues to compute() and compute_procedure()

0.2.3 - 2021-06-04

Added

• Batch compute for both compute() and compute_procedure() methods
• FutureResultGroup for batch computations

Changed

• Added pydantic BaseModel as base for FutureResult objects.

0.2.2 - 2021-05-21

Added

• Extended documentation to include a Code Reference section and much more comprehensive documentation of the main objects.
• Added compute_procedure to TCClient for geometry optimizations.
• Added TCClient.version property for quick version checks.

0.2.1 - 2021-03-05

Added

• Changelog
• User documentation
• Website for documentation

0.2.0 - 2021-02-26

Added

• Added TaskStatus enum to hold all task statuses.
• Basic documentation on main classes.
• [core_decisions.md] to document thinking behind architectural choices.

Changed

• FutureResult.get() to return either an AtomicResult or a FailedComputation
• Simplified README.md overview to use dictionaries instead of classes. Results in simpler tutorial with fewer imports.

0.1.1 - 2021-01-22

Added

• TCClient that can manage credentials, submit AtomicInput computations, and retrieve AtomicResult output from TeraChem Cloud.
• _RequestsClient class that handles all network requests to TeraChem Cloud server
• FutureResults object that is created from a task_id and can be used to retrieve a result once finished.