search

jump-dev / JuMP.jl

Modeling language for Mathematical Optimization (linear, mixed-integer, conic, semidefinite, nonlinear)
http://jump.dev/JuMP.jl/
Other
2.17k stars 390 forks source link

Bump compat of DimensionalData #3707

Closed odow closed 3 months ago

odow commented 3 months ago

We can fix the printing. But https://github.com/rafaqz/DimensionalData.jl/issues/664 is a blocker.

codecov[bot] commented 3 months ago

Codecov Report

All modified and coverable lines are covered by tests :white_check_mark:

Project coverage is 98.35%. Comparing base (d7c45a9) to head (7686c35).

Additional details and impacted files ```diff @@ Coverage Diff @@ ## master #3707 +/- ## ======================================= Coverage 98.35% 98.35% ======================================= Files 43 43 Lines 5709 5709 ======================================= Hits 5615 5615 Misses 94 94 ```

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.

rafaqz commented 3 months ago

I forgot to up the lower bound. Its fixed and the other version will be yanked.

odow commented 3 months ago

I forgot to up the lower bound

I don't really understand why this would help? It's failing on nightly, not on an older version like 1.6

the other version will be yanked.

There is no need to yank patch versions

odow commented 3 months ago

Your master is still failing: https://github.com/rafaqz/DimensionalData.jl/actions/runs/8224367439/job/22488102829

rafaqz commented 3 months ago

Thats nightly failing, for multiple reasons

odow commented 3 months ago

Thats nightly failing

Sure, but I don't want to bump the compat because then JuMP will fail tests.

It looks like you added the problematic https://github.com/rafaqz/DimensionalData.jl/pull/585 only yesterday? We shouldn't touch Base internals.

rafaqz commented 3 months ago

To clarify the yank is for 1.8 and 1.9... they will also break for unrelated reasons (like effects system macros). I just forgot about them.

I misunderstood your comment as nightly will fail anyway for other reasons even if I fix that particular bug.

rafaqz commented 3 months ago

DD unavoidably hits a few internals where Base doesnt let us pass through dimension types.

But that new one was probably a mistake, its just very tricky to get dispatch working. Im working around it now.

(Also this version bump is huge, like 40 closed issues. So might be messy for a few days as dust settles.)

odow commented 3 months ago

I misunderstood your comment as nightly will fail anyway for other reasons even if I fix that particular bug.

It's okay if your tests fail because of printing, etc. But ideally, they shouldn't cause the basic tests in JuMP to fail:

https://github.com/jump-dev/JuMP.jl/blob/d7c45a979e01c6d91cd489873bbd8982ad2518d1/ext/test_DimensionalData.jl

rafaqz commented 3 months ago

For sure. I will fix that asap.

Hopefuly the other ones out if my control dont affect your tests.