Julia could’ve been better than MATLAB, even in heavily vectorized code, if some of the great projects like ParallelAccelerator.jl were still active. It seems that the countless changes that Julia witnessed recently drove many people away.
I don’t understand that. Julia is much better at vectorized code than before, due to the introduction and improvement of dot-broadcasting.
My impression was that things like ParallelAccelerator became obsolete, because the language development either incorporated or overtook this development. The need for various band-aids simply diminished.
I don’t exactly understand what your opinion of the development actually is, though. To me, it seems amazing. Much better than a glorified Matlab.
11 Likes
Most of ParallelAccelerator is now part of Base Julia (threading) and standard compiler optimizations. I am not sure what’s missing.
7 Likes
Maybe just this: How do I deprecate a package?
2 Likes