broccoliSpicy
broccoliSpicy
this function seems a lot of work and I am not sure I can finish it before v0.5 release
this behavior is conformant with the floating point number definition
this is not part of our v0.4 plan,
I think I will implement this in v0.5
there are two solutions to these: 1. implicit cast decimal to float64 2. drop decimal's support for all math functions notice: it is impossible/stupid for us to implement all math...
regarding `Sentinel encoding for nulls`, for datatype `boolean`, i guess we can chose whatever value that is not `false`, `true` for datatypes like `timestamp`, `Date32`, `Date64`, `Time32`, `Time64`, `Duration`, `Interval`,...
so excited to see your ideas on struct/list encodings! @westonpace
a few thoughts about [the doc]( https://docs.google.com/document/d/19QNZq7A-797CXt8Z5pCrDEcxcRxEE8J0_sw4goqqgIY/edit?usp=sharing): 1. during integer encoding, if we group integers in size of 1024, then when the bit-width is `>10` bits, it is guaranteed to...
> during integer encoding, if we group integers in size of 1024, then when the bit-width is >10 bits, it is guaranteed to find a sentinel through constant * 1024...
> > sorry, after rethinking about this, I think this is not feasible using only constant * 1024 runs of XOR operations, there might be many missing numbers > >...