The intersection of the sets “people who use Windows” and “people who develop Julia” is essentially empty.
Do you mean core julia, or julia packates?
Core. I believe there are a handful of package devs on Windows.
How does this affect julia windows stability? Is it “good enough”?
It doesn’t, and yes? The majority of Julia downloads were for Windows last time I heard numbers. There are plenty of Windows bug reports and people have and gone to extensive lengths to fix issues (speaking both personally and from many more observations). Some devs do maintain 2nd Windows machines or VMs.
If you have unrelated questions, next time please start a new thread.
Ok thanks. sure will do.
That seems to be the case. While there is a clear majority of Mac/Linux devs even in the package ecosystem, we exist. I am a very active package dev and my main dev platform is Windows (and CentOS 7). DifferentialEquations.jl has enough dependencies that if something isn’t working on Windows I’ll open issues / PRs pretty quickly. So anything in that general domain (diffeq, Optim, Plots, NLSolve, IterativeSolvers, etc.) I can vouch for. As you go further over to the machine learning / stats stacks? I can’t vouch for it, and there have been times (specifically due to Rmath, which was solved awhile ago) where I have had problems with it.
The other dev that I know works on Windows is @musm , and he’s done a pretty good job making sure things like MATLAB.jl are supported on Windows. If he’s touched a package he probably fixed up all the Windows issues.
Generally, if a package is all native Julia, it will work well on Windows. If it’s a package which has binary dependencies, look for AppVeyor tests. If there are no AppVeyor tests enabled, it’s a sign that you may be entering untested waters. That doesn’t mean it won’t work: but it’s a sign that likely the dev is working on Linux and only testing on Linux.
1 Like
Thanks for the insight chris.