OssiLehtinen

Here is an example generated with reprex: ``` r library(dplyr, warn.conflicts = F) library(dbplyr, warn.conflicts = F) mf % arrange(x) %>% group_by(y) %>% summarize(max_x = max(x, na.rm=T)) %>% remote_query() #>...

Wow, looks amazing! You sure work fast with these :) A quick first note on your question: I would vote for having the convert_array parameter in noctua_options. This would allow...

Looks great after some testing too! It is a shame that athena returns the weirdly formatted stuff with structs and maps. We tried brainstorming a bit with some colleagues on...

Sounds reasonable. It also helps that data.table alerts the user if casting fails. > Possibly will just document this behaviour so users are aware of what will happen if they...

Ah, sorry, you are right: it should be 'timestamp' and not 'datetime'. The fractional seconds are optional, however, but can have at maximum three digits. Timestamps of the format '2020-01-31T11:59:25Z'...

Yes! If one wants similar handling of timestamps, there is an additional hurdle: dbplyr formats timestamp variables in the above mentioned format (e.g., '2020-01-31T11:59:25Z', https://github.com/tidyverse/dbplyr/blob/38cdafd74303baaee02e545831d242de6c656f4c/R/escape.R#L76) so one would need to...

FYI: the development version of dbplyr has now generics for sql_escape_date and sql_escape_datetime, which allow, for example, Athena-specific handling of date-variables as inputs. Also simple AthenaConnection-methods exist there now. https://github.com/tidyverse/dbplyr/pull/391...