Vitalie Spinu

`force_tz` gained new argument to control this behavior `roll_dst` ```R > dt force_tz(dt, tzone = "America/New_York", roll_dst = c("NA", "pre")) [1] "2020-11-01 01:30:00 EDT" > force_tz(dt, tzone = "America/New_York", roll_dst...

Would be nice to add conversion from logicals as well. Primarly for the sake of all NA vectors.

Indeed, it does not make sense to check all of them.

No idea, was it working with the older version? As the message suggests your tzone database might not be in the location as suggested. Our tzone db lookup is [here](https://github.com/tidyverse/lubridate/blob/main/R/tzdir.R)...

Our [results](https://cran.r-project.org/web/checks/check_results_lubridate.html) are good on debian and fedora, so if indeed an issue it's FreeBSD specific.

Sorry, I misread. This is R base error and not lubridate specific. Strange for sure. It's probably tzone database related but the error is confusing. It might be worth reporting...

Hi @jakemanger, there is a failure in pkgdown. I think you need to declare it in the index. More importantly, Should we add inequalities as well, for full partial order...

Thanks @jmoralez. Fit+predict is surely an option but it would require fitting the models twice. Also there are some implementation differences between fit+predict and forecast (ex. progress bar, fallback model)....

> Why would you need to fit the models twice? Once for forecasting and once to get the parameters from `fit`, or once for CV and once for parameters. I...

We are no longer able to use statsforecast v1.5 on our spark cluster due to the dependency conflicts of the fugue dependencies. They require [antlr4](https://github.com/fugue-project/fugue/issues/509) higher than that what spark...