Unable to set seed=NULL in simulate residuals

[jmniehaus]

Hello,

I am attempting to write my own simulate function for a package I am implementing that will rely on dharma. I am trying to use dharma's getRandomState function at the beginning of my simulate function, similarly to dharma's simulateResiduals function. However, when I use seed=NULL, I get an error due to a logical problem in getRandomState, I think.

Reprex:

y = rnorm(100)
x = rnorm(100)
mod = lm(y~x) 
simulateResiduals(mod, seed=NULL)

> Error in if (is.logical(seed) & seed == F) { : argument is of length zero

Because seed=NULL, NULL==F returns a zero-length vector.

[florianhartig]

OK, but why is that a problem? What would be your desired behaviour for seed = NULL?

[florianhartig]

OK, sorry, I see that the help specifies that this should work. I will have to dig into that!

[florianhartig]

This was indeed a bug, thanks for reporting this. I am just running the tests and will push this to master in a second (for how to install, see start page).

About your simulate function - is this a regression package, or a package that requires simulations? If the former, note https://github.com/florianhartig/DHARMa/wiki/Adding-new-R-packages-to-DHARMA, if the latter, note that DHARMa as internal S3 wrappers for the supported package in file compatibility.R, see ?DHARMa::getSimulations