Extending Tibbles - current best practice

What is the current best practice for extending the tibble class, in particular, in regards to grouping with dplyr?

I see there is a current issue open (https://github.com/tidyverse/tibble/issues/275) so I guess this has still not been solved but it would be good to get a feel of what approach people are taking in the meantime.

I've seen some old discussions on this but think they've probably been superseded (e.g. using sloop::reconstruct which no longer exists).

The most comprehensive package I've seen tackling this is tsibble but are there any others I should look in to?

Found the answer in the dplyr 1.0.0 release notes:

dplyr now has a rudimentary, experimental, and stop-gap, extension mechanism documented in ?dplyr_extending