[DNF2]

I believe the `psort!` code that was shown in the blog post was a 'toy example', meant to demonstrate how to use different parallel techniques, one of them being the use of thread-local state.

There's probably going to be a lot of development of parallelized code going forward. Maybe you could help? :)

[joehl]

Providing Julia Code for in-memory sorting, i.e. a binary mergesort with parallel branches and parallel merge plus a specialized sort for instable inplace sorting (better Quicksort, parallel branches only) is not a huge step from the code above: I could contribute that. If however, the expectation would be that the sort is more general, i.e. k-ary and performs for data actually being on ssd or even on disk (JuliaDB?) thats a complete different story: one would need to experiment with much more complicated algorithms such as cache-oblivious funnel-sort, parallel superscalar samplesort, real disk sorts ... I don't have time for that. How would I avoid duplicating the efforts of others?