Robert Kruszewski
Robert Kruszewski
I will take over this pr
This now compiles but there's still a lot of tests to fix
I need to figure out all the slow downs - seems like arrows append_packed_range is faster than bitvec implementaion and forcing iterators to be owned adds a bunch of drops...
Will make a new pr
This should work imho. Right now the reader is dumb and just assumes unique set of columns is being read
I think this is the wrong level to make it work at. This should be unique set of columns to read. You can duplicate column at higher level
Interval types? Are they just an extension dtype over different kinds of ints? For completeness it would be nice to have u128 primitive array though all you really need is...
In order for us to return dictionaries by default we need to move code around so arrow conversion doesn’t live in the main crate. I will do that today which...
I attempted the move and it's a lot more intertwined/involved than I originally assumed. There's some more work required in order to decouple compute from arrays. In the meantime to...
I was halfway through doing this but realised there's a bit of API design change necessary. What we want to do is move compute functions out of the main crate...