compatibility: mlr3 1.0.0

[be-marc]

Hey, the new mlr3 requires instantiated resamplings to store a hash of the row ids of the task. More info here mlr-org/mlr3#1282. This PR should also work with the current mlr3 CRAN version. Please upload a new mlr3spatiotempcv version to CRAN soon.

[pat-s]

@be-marc

Tests still fail

Error: --- re-building ‘mlr3spatiotempcv.Rmd’ using rmarkdown
--- finished re-building ‘mlr3spatiotempcv.Rmd’
--- re-building ‘spatiotemp-viz.Rmd’ using rmarkdown
Quitting from spatiotemp-viz.Rmd:89-148 [unnamed-chunk-4]
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
<error/rlang_error>
Error in `self$task_row_hash <- task$row_hash`:
! cannot add bindings to a locked environment

(I just got a CRAN mail due to the errors likely related to the mentioned changes)

[be-marc]

Maybe this is a caching issue with tic? It works locally and r-cmd-check-paradox also works.

[be-marc]

Winbuilder: https://win-builder.r-project.org/M8KhjU613KK2/00check.log

[pat-s]

Thanks, I'll have a look.

Maybe this is a caching issue with tic? It works locally and r-cmd-check-paradox also works.

I guess I'll remove tic in the near future.

[pat-s]

@be-marc seems to work now, thanks!

Btw: any reason you're not using use-public-rspm: true in the GHA workflows but instead source package installs?