Maximilian Paulisch

Idea: change `Path` to a `Context` class that carries more information: path to element of `x`, path to field spec, where the error was thrown e.g. when applying `transform` or...

~I was just about to write that this works when using the first 4 packages. But somehow this sometimes works and sometimes doesn't...~ @krlmlr The issue is that `spec_guess()` has...

This is because the code path for colmajor is used when the input is a data frame. This makes the error message indeed quite confusing. Regarding the errors themselves: 1....

But it is also a bit annoying that all examples work with the same spec if using `simplifyDataFrame = FALSE`.

I basically follow the [design principles](https://design.tidyverse.org/dots-prefix.html) of the tidyverse here. E.g. look at `dplyr::slice(.data, ..., .by = NULL, .preserve = FALSE)` vs `dplyr::slice_head(.data, ..., n, prop, by = NULL)`. I...

Some time ago I added a parser for an OpenAPI spec (`parse_openapi_spec()`). So this basically supports JSON schema. It surely makes sense to add proper support for JSON schema.

Maybe just add as an example in the vignette: ```r tib_bigint

Supporting some kind of id columns sounds great to me. But sometimes one also has integer ids. So it might make sense to have something similar for integers. Does `pillar`...

@krlmlr pillar also supports some kind of focus columns, right? It might make sense to set `tib_id()` as focus columns. What do you think?

I think IDs are usually integer (resp. big int) or character. It would make sense to support both, e.g. `tib_id(format = c("character", "integer"))`