Initial guess part update

[PierreMartinon]

Steps for the CTDirect / CTBase updates

• 1. Merge v0.9.0 with local init into main
• 2. Then comment local init and its export, and make release from this branch; set as pre production release
• 3. Then update CTBase and test 0.9.0
• 4. Release updated CTBase
• 5. Update CTBase requirement, delete local init, merge into main and release 0.9.1 normally

[jbcaillau]

@PierreMartinon looks like this is almost done, no?

Regarding HSL.jl, @amontoison confirms it is very light (everything is in the dll) so this can indeed be included (= using HSL) in CTDirect.jl, should be harmful on load time.

[PierreMartinon]

Yup, done and done

[jbcaillau]

@PierreMartinon any clue welcome on #69 (comment)

[PierreMartinon]

@PierreMartinon any clue welcome on #69 (comment)

Haven't tried the HSL option yet. I'll do this when restarting the performance/benchmarking part.

[amontoison]

@PierreMartinon looks like this is almost done, no?

Regarding HSL.jl, @amontoison confirms it is very light (everything is in the dll) so this can indeed be incluhttps://docs.github.com/github/writing-on-github/getting-started-with-writing-and-formatting-on-github/basic-writing-and-formatting-syntaxded (= using HSL) in CTDirect.jl, should be harmful on load time.

HSL.jl only has HSL_jll.jl as dependency so it's very light.
You can only depend on HSL_jll.jl if you want but you will not gain a lot and you need to copy this Julia function here:
https://github.com/JuliaSmoothOptimizers/HSL.jl/blob/main/src/C/libhsl.jl#L1

The main advantage of using HSL.jl is that it takes care of loading an LP64 BLAS / LAPACK for you with the function __init__
https://github.com/JuliaSmoothOptimizers/HSL.jl/blob/main/src/HSL.jl#L16

But in your case you also depend on Ipopt.jl and I implemented the same __init__ function.