for `lifecycle::deprecate_warn` Is there a way to NOT encouraging users to report the issue here?

[chainsawriot]

rio/R/utils.R

Lines 145 to 150 in 447ea1b

	lifecycle::deprecate_warn(
	when = "1.0.3",
	what = "import(trust = 'should be explicit for serialization formats')",
	details = paste0("Missing `trust` will be set to FALSE by default for ", format, " in 2.0.0."))
	trust <- TRUE ## Change this for version 2.0.0
	}

[chainsawriot]

r-lib/lifecycle#187

As well as #450 #432

[chainsawriot]

#452 I think we need to address this before the typical "semester beginning" wave.

[chainsawriot]

tag @schochastics @jsonbecker

I think one solution is to not use lifecycle for this. Just use the good old warning().

[jsonbecker]

Oh wow that's aggressive to report an issue to a maintainer when they've deprecated a feature?

I'm good with swapping this to warning but that's a bad upstream choice.

[schochastics]

That is indeed a bit aggressive. I second that a good old warning is enough.

[chainsawriot]

Maybe should give this suggestion by @olivroy a try: r-lib/lifecycle#187

[chainsawriot]

But how deep in the caller_env() should we go? It is still quite convoluted.

[schochastics]

I think I still vote for an ordinary warning but mostly because I have not yet fully understood the caller_env()

[chainsawriot]

@schochastics OK, I agree with you. And I close this once again.

[olivroy]

Agreed the warning seems fine.

Here is my attempt at explaining caller_env()

wjakethompson/taylor#49 (comment)

This guide can be helpful too.

https://rlang.r-lib.org/reference/topic-error-call.html

I made a few PRs to implement this in different packages.

See lintr for example r-lib/lintr#2602 or gt rstudio/gt#1638