In the next minor version bump (v0.5.0) of Optim.jl, we have some deprecations being phased out, which is going to result in breaking code. This is annoying for users (Part I) and developers (Part II) alike. In this post I will focus on users.
The Julia Language, and as a result its ecosystem, is still young, and not even v1.0 yet. This means, amongst other things, that the user base is still relatively small and fragile. Breaking changes are to be expected, but that doesn’t make them less annoying. It also means that there isn’t too much information available online, so it is critical that whatever is available, has to be correct.
To help the transition from the old API to the new API, I went online, and searched for posts containing Optim.jl questions and answers. I located the following pages
- Logistic Regression in Julia using Optim.jl
- Optimizing Ad Placement with Julia
- Simple nonlinear least squares curve fitting in Julia
- Julia Libraries (quant-econ.net)
The first one is a stackoverflow question. This means I could simply use the
edit functionality. The first change has been approved,
still waiting for Ian Dunnings post to be updated.
The two next on the list are blog posts. Luckily, they didn’t have comments disabled, so I could leave comments.
The last one is from the quant-econ project. This is more like a book, and this means I cannot simply change, or propose a change. Instead, I contacted Spencer Lyon, who put the small changes needed on the list for their next update.
There are probably more out there, and I would love a heads up if anyone knows of any. You can find me at julia-opt.