Oscar Dowson

Ah it's https://github.com/jump-dev/Xpress.jl/blob/60ae1dd4971cd5cda76d214deaf93aa4c12c3d95/src/MOI/MOI_wrapper.jl#L2910-L2916

I think the issue was that https://github.com/jump-dev/Xpress.jl/blob/60ae1dd4971cd5cda76d214deaf93aa4c12c3d95/src/Lib/xprs.jl#L479-L482 and some users of Xpress.jl are still stuck on quite outdated versions of Xpress.

> But is it not possible to have code that checks the version of Xpress available and adjusts accordingly? Yeah, we can probably do that. Especially now that we have...

CI is failing because it is from a fork that does not have access to the license. What is missing is my time to review, merge, and tag a new...

Just chiming in here with my JuMP hat on. > Does anybody have an opinion on a fix? The highest priority fix is to make this non-modifying and return temporary...

> I'm curious about your use case? If you want explicit immutability This is a solution. But the main issue is that the defaults are the wrong way round: *...

> My intent was that the user would never keep an explicit highs_linear_expression object reference Oh, but they do. All the time. JuMP syntax, since I'm not familiar with highspy,...

See https://sddp.dev/stable/guides/add_integrality/ We do not necessarily converge to an optimal solution if there are integer variables. You could try: ```Julia SDDP.train(model; iteration_limit = 10, duality_handler = SDDP.LagrangianDuality()) # or SDDP.train(model;...

> The Lagrangian returns 1.28 and the Conic 1.2 so no luck in that regard. This makes sense because of the big-M formulation (even if the M is quite tight)....

Closing because this seems resolved. It is not a bug in SDDP.jl, and I think the docs summarize the issue well enough. Please comment if you have more questions and...