Closed tapastro closed 2 months ago
All modified and coverable lines are covered by tests :white_check_mark:
Project coverage is 66.18%. Comparing base (
84279cf
) to head (55ce7e8
). Report is 5 commits behind head on main.
:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.
I also suggest we leave https://github.com/spacetelescope/jwst/issues/8580 open as this PR doesn't fully address all the numpy 2.0 issues with jwst.
Agreed on keeping the issue open, and I'll get a regtest run soon. I have an stcal PR out as well, and I suppose I'll need a JWST PR to get the numpy upper limit out of the way.
Started a regression test run here:
EDIT: make that https://plwishmaster.stsci.edu:8081/job/RT/job/JWST-Developers-Pull-Requests/1674/
Last set of regtests were failing due to a day of Jenkins issues. Started a new set against jwst/master to check for compatibility issues with numpy<2: https://plwishmaster.stsci.edu:8081/job/RT/job/JWST-Developers-Pull-Requests/1691/
Partially addresses JP-3664
Closes jwst/#8578 Partially addresses jwst/#8580
This PR addresses a bug exposed by testing numpy 2.0 - the irs2 datamodel specifies arrays to be of type uint8, but operations in refpix promoted them to uint16 implicitly in numpy<2.0. Numpy 2.0 does not implicitly type promote in this case, leading to an integer overflow. This PR updates the specified array datatype to uint16.
Checklist
CHANGES.rst
(either inBug Fixes
orChanges to API
)