* Create Pooch Cache
* not exactly sure where the cache will be on osx
* print some debug info
* update cache clearing action
* Debug osx location now
* allow errors when deugging
* lets see what is in the cache
* see if the cache actually works
* since the cache is in 2 different locations, we need to include the OS in the key
* test cache
* it works!!!
* only run the intergration tests on AWS CPU runners (not all the tests + intergration tests
* skip tests when editing only AWS runner actions
* fixes#1537
* keep CPU runner behavior the same, only run intergration tests on GPU runner
* I just can't spell
* update to newest ami
* See if pinning to an older cuda-version works
* Just run slow tests on CPU runner
* run slow and normal test on CPU runner
* make it more clear what runs with the GPU tests
* fix flag name
---------
Co-authored-by: Irfan Alibay <IAlibay@users.noreply.github.com>
Co-authored-by: Alyssa Travitz <31974495+atravitz@users.noreply.github.com>
* Let's use python 3.12 on the docker image
* try installing the rdkit version I want to test
* free up diskspace
* include source
* Apply suggestion from @atravitz
* fix indent
---------
Co-authored-by: Alyssa Travitz <31974495+atravitz@users.noreply.github.com>
* install espaloma on long tests
* espaloma != espaloma_charge
* rename HAS_ESPALOMA to HAS_ESPALOMA_CHARGE
To disambiguate between esaplaoma and espaloma charge, we need to track
HAS_ESPALOMA separately from HAS_ESPALOMA_CHARGE
* skip test if espaloma is not installed
* fix HAS_ESPALOMA import
* see if installing espaloma gives us any issue
* import HAS_ESPALOMA
* only install espaloma when python != 3.13
* fix syntax
* Update openfe/tests/protocols/openmm_md/test_plain_md_protocol.py
Co-authored-by: Irfan Alibay <IAlibay@users.noreply.github.com>
---------
Co-authored-by: Irfan Alibay <IAlibay@users.noreply.github.com>
Co-authored-by: Alyssa Travitz <31974495+atravitz@users.noreply.github.com>
* update partial charge tests now that a production NAGL model is available (#1506)
* update tests now that a production NAGL model is available
* fix mock
* clean up mocking
* add nagl test news item
* debug slow python 3.13 tests (#1468)
* Turn off minimization when we do a dry run
---------
Co-authored-by: IAlibay <IAlibay@users.noreply.github.com>
* update quickrun execution docs and add MPS link (#1483)
* add jq cookbook (#1488)
* add jq code blocks
* use tutorial data
* news
* debug docs build (#1489)
* don't build openmm
* mambaforge->miniconda for rtd
* artifically small build for debugging
* dont build env
* just build python
* add channels
* add sphinx packages to env
* add all non-openfe-ecosystem deps
* remove shim channel
* add openfe package
* add everything except openmm and toolkit
* add openmm
* add openff-toolkit
* only openff-toolkit
* try openff-toolkit-base
* try latest openff
* everybody back in
* build without openmm
* build with openff-toolkit-base
* bump rtd
* add pip build
* remove any openff-toolkit
* build w/o openmm
* add pip install
* build everything without pip install
* try no deps
* try custom build
* only python and pip
* try openfe-toolkit-base
* add temp deps for build with openff-toolkit-base
* add temp deps for build with openff-toolkit-base
* mock imports
* dont mock with toolkit base
* bump toolkit version
* Revert "bump toolkit version"
This reverts commit 6e3108a836.
* add mocks
* try pinnint to toolkit-base 0.13.0, no mocks
* back to >= 0.13.0 toolkit to make sure it wasn't a fluke
* mock imports
* add pipe example for jq (#1505)
* add pipe example for jq
* pin to docutil 0.20
* try to avoid memory error
* debug rtd build memory error (#1510)
* try mambaforge build
* Revert "try mambaforge build"
This reverts commit 683e07f1ea.
* try explicit build
* try removing ambertools
* use the right env
* just add gufe pip install back
* add back openfe eco packages
* mock py3dmol
* add note for pip installs
* fix quote
* temporarily resort to SettingsBaseModel to get build
* specific imports
* dont use base settings placeholder
* temp pin numpy (#1542)
* Removing unnecessary limit on resids (#1539)
* Removing unnecessary limit on resids indices gathering
* Updated CHANGELOG for 1.6.1
* add summary
* update docs theme (#1545)
* adding new branding assets
* remove images
* update theme branch
* CantinaPurple accent
* lowercase convention
* updating text
* remove dark mode for now
* fix sidebars
* fix formatting
* capitalization
* don't break spaces
* bump ci
* favicon!
* whitespace :(
* pin to sphinx release
* pin to gufe 1.6.1
* move env info before test imports
* bump single file installer to 3.12.11 to match colab
---------
Co-authored-by: IAlibay <IAlibay@users.noreply.github.com>
Co-authored-by: Iván Pulido <2949729+ijpulidos@users.noreply.github.com>
* clean up conda cron CI
* Revert "clean up conda cron CI"
This reverts commit f12b72ba67.
* run at 5 instead of 7
* switch from --json output to raw output
* add xdist back in
* wait for python 3.13 until we release openfe 1.6.0
* Update .github/workflows/conda_cron.yaml
Co-authored-by: Irfan Alibay <IAlibay@users.noreply.github.com>
* switch to use github version
* fix bracket
* use checkout
* fix checkout call
* fix syntax
* fetch depth
---------
Co-authored-by: Irfan Alibay <IAlibay@users.noreply.github.com>
* bump to setup-micromamba v2
* pin micromamba version to latest
* bump everybody to checkout v4
* removing binary path entries
* explicitly pin micromamba version
* remove micromamba version from action
---------
Co-authored-by: Mike Henry <11765982+mikemhenry@users.noreply.github.com>
* test larger cpu runner
* update omsf runner version
* nvidia-smi throws an error if there is no GPU
* Try a large
* just run slow tets
* try xlarge (should fail but thats okay)
* lets see if the intel flavor is any faster
* see if this changes the runtime at all
* go back to working runner instance type
* bump size of disk
* check time on a t3 med
* try xlarge
* try t3a.xlarge
* now t3a.2xl
* now t3.2xl
* use platforms that actually will work
* need empty set not dict
* pin buggy mypy
* fix mypy pin
* see if large works
* check t3.2xlarge
* make sure we load the marks correctly
* I think we are using the wrong instance family
* don't run itergration tests on the CPU
* fix syntax with conda-lock
* don't run these while we are testing since it just eats up runners
* not sure how I messed up that command
* undo ci skip
* skip running other CI
* test lf env
* fix prefix location
* fix syntax with conda-lock
* don't clone the repo to get the latest tag
* fix yaml syntax
* forgot to include id
* test on osx as well
* osx /home -> /Users
* cleanup language
* update name
* propagate name change
* ci: add support for new start/stop runners for GPU runner job
This commit utilizes the new OMSF start/stop gha-runner action for the
gpu_runner workflow.
* ci: add support for new start/stop runners for CPU runner job
This commit utilizes the new OMSF start/stop gha-runner action for the
cpu_runner workflow.
* see what happens if we remove the flaky mark
* now that we run this one slow test on PRs, we should make sure it gets cancled each time a push is made
