Releases: mtzgroup/chemcloud-server
Releases · mtzgroup/chemcloud-server
Release 0.4.0
0.4.0 - 2021-06-04
Added
- Batch compute capabilities to
/compute
and/compute-procedure
Changed
- 🚨BREAKING CHANGE🚨: Updated
/compute*
endpoints to returnTask
objects includingGroupTask
objects that may contain subtasks instead of returning a singlestr
of thetask_id
- 🚨BREAKING CHANGE🚨: Updated
/result
endpoint to acceptTask
objects instead of just astr
oftask_id
. This included updating the endpoint to be aPOST
instead ofGET
endpoint.
Release 0.3.5
0.3.5 - 2021-05-26
Added
xtb
to compute engines- Added
cached-property
package sincekombu
didn't install it by default as a dependency into the linux docker container.
Changed
- Updated
tcpb>=0.8.0
- Updated
celery
version
Release 0.3.4
0.3.4 - 2021-05-21
Added
geomeTRIC
optimizerrdkit
engine (adds force fields which can be used for single point computations or in optimizers)
Changed
- Run compute tests on hydrogen instead of water for faster results
Release 0.3.3
0.3.3 - 2021-05-19
Added
compute_procedure
task for doing geometry optimizations usingpyberny
- VSCode settings file added to repo.
Changed
- Combined two commands from the CI/CD pipeline used against XStream to get fewer auth errors
- Updated
mypy
definitions onSettings
object to passmypy
checks while still allowing tests to run on CircleCi without needing Auth0 configuration. I.e., the web application can be run without needing Auth0 config (for testing purposes). - pre-commit
mypy
now loads from GitHub repo instead of localmypy
install.
Release 0.3.2
0.3.2 - 2021-04-06
Added
- Test assertions to check that results are deleted from celery backend after retrieval by client.
Changed
- Added flags
--without-heartbeat --without-mingle --without-gossip
to celery workers to reduce network overhead chattiness. - Delete celery results from backend upon retrieval by client. This will prevent Redis from becoming a memory hog at times of high computational load.
Release 0.3.1
0.3.1 - 2021-03-10
Added
- Changelog
- Updated
tcpb
package to>=0.7.1
to enable molden file creation usingimd_orbital_type="whole_c"
keyword.
Release 0.3.0
0.3.0 - 2021-01-26
Added
- Added TeraChem Protocol Buffer server to available compute engines.
- XStream deployment for both dev and prod now require GPUs to power Terachem. Deployment config for dev/prod was split into
docker-compose.xstream.dev.yaml
anddocker-compose.xstream.prod.yaml
. - TeraChem license.key docker secret to swarm on XStream
architecture.md
to/docs/development
to document core architectural decisions.- Links on user dashboard to
qccloud
python client,/logout
, and a brief description of how to change password. - Links to
users/dashboard
on the main documentation page. - Forgotten
__init__.py
toterachem_cloud
package. Added__version__
to file. This causesmypy
checks to fail as they were previously not inspecting this package fully due to missing__init__.py
. TaskStatus
enum to hold task status values.
Changed
/compute/result/{task_id}
can return eitherAtomicResult
orFailedOperation
objects.- CircleCi build pipelines to only include a single build step for web and workers instead of a split pipeline for dev/prod
- Can approve CircleCi build and deploy steps upfront without having to wait for tests to pass. Build/deploy will still only occur if tests pass.
- XStream stacks for dev/prod no longer derive from the same template file.
- Changed
CeleryAtomicResult
toTaskResult
and changeatomic_result
attribute to justresult
to note thatresult
may now be anAtomicResult
or aFailedOperation
(a result could either of these data types).
Release 0.2.1
0.2.1 - 2021-01-21
Removed
- Removed
docker_layer_caching
from CircleCi build process
Release 0.2.0
0.2.0 - 2021-01-21
Added
- refresh_flow to
/oauth/token
endpoint
Release 0.1.0
0.1.0 - 2020-12-18
Added
- Three endpoints all available on
/api/v1
:/oauth/token
to get JWTs,/compute
to submit AtomicInput compute request,/compute/result/{task_id}
to request a result delivered as an AtomicResult object.
- Auth provided by Auth0.