We ought to recognize the use cases, as much as humanly possible, for all languages or frameworks or platforms (etc.) and simply use the best tools for any given job.
I wouldn't exactly consider Operating Systems to be one of those use cases for Javascript. The flip side of knowing the use cases is knowing the limitations and understanding that "everything" isn't exactly a valid use case, which I believe is what larme was trying to say.
I will say, though, that exploring the possibilities for a language is not a waste of time or mind. From stretching both mind and tool to do new things, new insight can be gained or patterns formed that might never have been considered.
Additionally, nobody ever said an OS has to be the best at everything either. It could be that a javascript-based OS, while perhaps not as efficient at generating machine code, may end up being much more efficient for humans. This would certainly be a win for many tasks such as management that should be optimized for human use.
Well, isn't that a bit of the point the parent makes? It is not necessarily the best tool for _everything_.
(Not criticizing the project, many package managers are written in some scripting language)