John Mount
First, thanks for a great and useful package. I am seeing an identifier change case during transform. Possibly this is my fault as I don't know how to tell `sqlparse`...
With `sparklyr` version `0.6.2` (and `dplyr` version `0.7.2`, though I don't know if this depends on version) I am seeing `hive` tables that have small integer types imported as `R`...
This is likely related to https://github.com/rstats-db/RMySQL/issues/37 , but I can't get type warnings to stop when using RMySQL and dplyr::mutate together. Even placing dplyr::compute() in a suppressWarnings() doesn't help as...
Make it easy for categorical variables to share indicator space additively, and same for derived impact columns.
From https://github.com/tidyverse/dplyr/issues/3516 . Notice how violently the value is rounded when round-tripped through the database. The amount of rounding depends on the driver (so may not be a pure `DBI`...
### Problem description I like that the Polars .std() and .var() are, in most cases, computing sample variance and sample standard deviation. I also understand the utility of returning zero...
Implement a pure data algebra training path, or a Polars training shim
Finish turning back on tests where Polars behavior isn't the same as desired SQL ref. Note: for Nulls Polars can tell the difference from NaN, so will not match Pandas-...
To solutions add a regularize records solution. Perhaps 2: one to service "the zero bug" ( https://win-vector.com/2017/02/21/the-zero-bug/ ), and one to narrow and fill out records prior to a cdata...