Repeated calls to df.iloc[n]
go through fast_xs which will call interleaved_dtype
each time with the same arguments. We could cache the result of that and save some time. Would need to be careful about cache invalidation.
xref #61747
Repeated calls to df.iloc[n]
go through fast_xs which will call interleaved_dtype
each time with the same arguments. We could cache the result of that and save some time. Would need to be careful about cache invalidation.
xref #61747