The usual reason given is lack of libraries compared to python. I started a project a while ago with the idea that you could create a ruby library that would implement the C python extension API using the ruby C extension library... I got some basic stuff working, so you could "import" a basic python C extension in ruby (after recompiling it against my headers/library). Then you can use a python -> ruby transpiler (which does exist) to convert the python parts of the code to ruby and maybe have stuff "just work". The more I got into it though, it seemed pretty daunting to get something like numpy working... so I have stopped for now.
I'm not sure what Python library you couldn't find an equivalent for in rubygems. I never had a problem finding what I needed. (Maybe some ML or related fields?)
Depending on the size of the library it might make sense to rewrite it.
Using C extensions is more popular than Python or other languages
Ah, good point. Domain does matter. But that didn't seem to be what the original poster was pointing to.
Based on my ambient knowledge, I wouldn't want to write anything in the machine learning space in ruby. Or for that matter in any language other than python (or maybe scala/java, if you are using spark).
Or are you suggesting that python has more/superior libraries compared to ruby across the board?
