Skip to content

Conversation

blegat
Copy link
Member

@blegat blegat commented Aug 5, 2021

Closes #1508

@odow odow merged commit 0e64780 into master Aug 5, 2021
@odow odow deleted the bl/moi_mpb branch August 5, 2021 21:10
@odow
Copy link
Member

odow commented Aug 5, 2021

I'm not sure I agree, but anyway...

@blegat blegat restored the bl/moi_mpb branch August 7, 2021 07:14
@blegat blegat deleted the bl/moi_mpb branch August 7, 2021 07:15
@blegat
Copy link
Member Author

blegat commented Aug 7, 2021

I'm not sure I agree, but anyway...

I'm open to rephrase, for LPQP and NLP solvers, MPB was probably simpler. I'm not sure how to phrase it either. The previous phrasing needed to be nuanced though. It really depends on which kind of solvers. Solvers that fitted well in one of the three classes of MPB were simple in MPB but for others, MOI is simpler to implement because there is no bridging to do.
Even for conic solver, it seems more complicated than with MatrixOfConstraints: https://github.com/jump-dev/SCS.jl/blob/master/src/MPB_wrapper.jl#L135-L407
And with that implementation, it does not support rotated SOC.
MOI is maybe more complicated but it's not more work.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Development

Successfully merging this pull request may close these issues.

Transitioning solver interface
2 participants