Conflict with conflicted library, remove uses of apropos()?

[emstruong]

Hello,

It appears that Rstan's use of the apropos() function is causing conflicts with the conflicted library, so I was advised to open an issue here to see if all uses of apropos() could be removed?

It doesn't seem like changing apropos() is feasible and it's not clear if it would be a good idea or even possible to change the conflicted library to make things work. Also, conflicted seems like a really important package for computational reproducibility and considering R's nature.

Please see the attached discussion for further context. Here

Michael

[jgabry]

Hi, thanks for reporting this. I didn't even know RStan called apropos but I guess many years ago it was used here on line 82:

rstan/rstan/rstan/R/rstan.R

Lines 77 to 97 in 9b8d8fe

	# find possibly identical stanmodels
	model_re <- "(^[[:alnum:]]{2,}.*$)|(^[A-E,G-S,U-Z,a-z].*$)|(^[F,T].+)"
	if(!is.null(model_name))
	if(!grepl(model_re, model_name))
	stop("model name must match ", model_re)
	S4_objects <- apropos(model_re, mode="S4", ignore.case = FALSE)
	if (length(S4_objects) > 0) {
	e <- environment()
	stanfits <- sapply(mget(S4_objects, envir = e, inherits = TRUE),
	FUN = is, class2 = "stanfit")
	stanmodels <- sapply(mget(S4_objects, envir = e, inherits = TRUE),
	FUN = is, class2 = "stanmodel")
	if (any(stanfits)) for (i in names(which(stanfits))) {
	obj <- get_stanmodel(get(i, envir = e, inherits = TRUE))
	if (identical(obj@model_code[1], stanc_ret$model_code[1])) return(obj)
	}
	if (any(stanmodels)) for (i in names(which(stanmodels))) {
	obj <- get(i, envir = e, inherits = TRUE)
	if (identical(obj@model_code[1], stanc_ret$model_code[1])) return(obj)
	}
	}

But I'm not sure what it should be changed to. Is there a recommended alternative to apropos that wouldn't cause this problem? Looking at the RStan code here, it seems like it's trying to find existing S4 objects that are stanmodel objects identical to the one being created, in which case it returns the existing one instead of creating a new one and compiling again (@bgoodri is that right?).