vec_is(..., new_datetime()) too strict

[krlmlr]

The tzone attribute is set for Sys.time() but not for new_datetime(), this is treated as a meaningful difference.

vctrs::vec_is(Sys.time(), vctrs::new_datetime())
#> [1] FALSE

^{Created on 2020-03-11 by the reprex package (v0.3.0)}

Related: #561.

[DavisVaughan]

This strictness is part of the reason we have "partial" types. i.e. for factors the partial type "learns" the levels from the other input

library(vctrs)

vec_is(factor("x"), factor())
#> [1] FALSE

vec_is(factor("x"), partial_factor())
#> [1] TRUE

vec_is(factor("x"), factor("x"))
#> [1] TRUE

^{Created on 2020-03-11 by the reprex package (v0.3.0)}

I'm not sure what the best long term solution is here though. It might be something other than partial types

[lionel-]

To me this doesn't feel connected to partial types because I see timezone as a "detail" parameter of the type.

I think we'll need to bite the bullet and introduce a new generic for the is-a relationship.

[krlmlr]

Can we treat types equal if their proxies are equal?

[lionel-]

We can't, e.g. factors and integers have the same memory representation but represent different values.

[krlmlr]

I now see:

vctrs::vec_is(Sys.time(), vctrs::new_datetime())
#> [1] TRUE

^{Created on 2021-03-18 by the reprex package (v1.0.0)}

[lionel-]

This is probably incidental. vec_is() is currently not guaranteed to work this way, it has design issues that are not solved yet.

[krlmlr]

Let's add a test while it lasts :zany_face: