Maximilian Girlich

@jennybc I saw you are one of the authors of bigrquery. Maybe you want to review and merge this? I think this would make people very happy 😄

Sorry for the late reply. Completely forgot about this... Now this does not rely on purrr anymore and should hopefully work.

I created a [PR](https://github.com/tidyverse/dbplyr/pull/749) in `dbplyr` that exports the generic `sql_random()` After this is merged it should be sufficient to define ```r # registered onLoad sql_random.BigQueryConnection

I've opened a similar issue (r-lib/fs/issues/237) for the `fs` package. A consistent syntax for both packages would be good in case such a feature would be included in both packages....

There is already a tidyr issue https://github.com/tidyverse/tidyr/issues/1071 for this. In dbplyr we simply added `dbplyr_uncount()` for now...

It might take some time for https://github.com/tidyverse/tidyr/pull/1101 to be merged (the interface of most of these functions is not completely stable yet). Therefore, I created a separate PR https://github.com/tidyverse/tidyr/pull/1358 for...

@markfairbanks thanks for your notes and thoughts on this! Do you know any nice way to handle `NULL`? ``` r library(data.table) nest_df

bare numbers are also valid json ([see standard](https://www.json.org/json-en.html) and try out at [jsonlint](https://jsonlint.com/)). My workflow is something like in the following code ``` r x [ #> 1, #> 2...

slightly modified example with an array as input: ``` r jqr::jq(c('{"a": 1}', '[2]'), ".a") #> 1 ``` Created on 2020-01-27 by the [reprex package](https://reprex.tidyverse.org) (v0.3.0)

This still sounds nice but requires a bit more work than initially expected, mostly because the CTE names have to be unique.