For the last few months, I have been coming up to speed on both Linear Programming with Jump.jl using Clp.jl, Cbc.jl, and GLPK.jl. I have made some minor PRs to a few of the projects in order to improve the capability to solve my own problem.
But as a dig into each of those problems, I keep getting the feeling that Julia needs a native LP solver. Initially, this would most likely take the form of some flavor simplex solver. But with that foundation, I think a lot of progress could be made. From there a MILP solver could be built, and after that, there already exists higher level julia solvers. But to be performant it is clear that you need very low-level interoperability to start building those blocks. Just like Cbc has its own interfaces inside Clp.
Now a few years ago it appears that @miles.lubin started on an initial pass with jlSimplex. Though Julia has progressed a ton since then and I feel like with v1.0 out the capabilities are there finally there to make a native state of the art solver.
At the end of the day, I ask this because it is far outside my capabilities to start a project like this, but I would love to contribute. I am sure many of you have had a similar thought and it would be great to hear your opinions.