Open xiaodaigh opened 6 years ago
That's right. I think it's possible but it hasn't been a priority so far. Do you have a use case other than just making read-performance awesome?
I want to utilise multi-processing so that each process can process one chunk of data.
Any new thoughts on this? Could the logic from CSV.jl, which also has native multi-threading now, be helpful here?
Even if the reading logic could be modified so that it still scans the whole file but it only allocates memory for the given range, that would be really awesome!
I note the README has this passage
But is it technically possible to jump ahead and read from a certain row? I think it's related to #5