Closed jameslamb closed 2 months ago
Contributes to https://github.com/rapidsai/build-planning/issues/72.
Similar to https://github.com/rapidsai/pynvjitlink/pull/90, proposes replacing a use of .get() in the conda recipe with [], to ensure that we get a loud build error if the "project" table is not present when conda-build reads pyproject.toml.
.get()
[]
"project"
conda-build
pyproject.toml
I know that at some time in the past, the .get() was introduced as a possible solution for errors observed at build time: https://github.com/rapidsai/pynvjitlink/pull/33#discussion_r1449525892
But based on testing over in https://github.com/rapidsai/pynvjitlink/pull/90, that seems to no longer be an issue.
Contributes to https://github.com/rapidsai/build-planning/issues/72.
Similar to https://github.com/rapidsai/pynvjitlink/pull/90, proposes replacing a use of
.get()
in the conda recipe with[]
, to ensure that we get a loud build error if the"project"
table is not present whenconda-build
readspyproject.toml
.Notes for Reviewers
I know that at some time in the past, the
.get()
was introduced as a possible solution for errors observed at build time: https://github.com/rapidsai/pynvjitlink/pull/33#discussion_r1449525892But based on testing over in https://github.com/rapidsai/pynvjitlink/pull/90, that seems to no longer be an issue.