Keep S7's := ahead of rlang and data.table - #718
Open
t-kalinowski wants to merge 15 commits into
Open
Conversation
hadley
reviewed
Jun 25, 2026
| } | ||
|
|
||
| env <- as.environment(paste0("package:", pkgname)) | ||
| env[[".conflicts.OK"]] <- TRUE |
Member
There was a problem hiding this comment.
I think we should only ever apply this to S7 itself, and I think we can move it up to .onAttach() so we use it unconditionally when conflictRules() is not available.
hadley
reviewed
Jun 25, 2026
Comment on lines
+42
to
+45
| conflictRules <- get0("conflictRules", envir = baseenv(), inherits = FALSE) | ||
| if (is.null(conflictRules)) { | ||
| return(invisible()) | ||
| } |
Member
There was a problem hiding this comment.
I think it would be clearer to use an explicit version here
t-kalinowski
force-pushed
the
suppress-s7-warnings
branch
from
August 17, 2026 13:35
bdaee29 to
54ebeeb
Compare
Inline the former activate_attach_compatibility() helper so the flag is only ever applied to S7's own attached environment, per review.
conflictRules() has been in base since R 3.6.0 and S7 now requires R >= 4.2.0, so the get0() availability check was dead code.
.conflicts.OK makes library() skip its conflict report wholesale, which also swallowed messages about genuine, unrelated conflicts. Re-emit the report ourselves, minus S7's deliberate masks (@ over base, := over rlang and data.table), reusing base's .maskedMsg formatter and message catalog so the output is identical to what library() would print.
A strict conflicts.policy errors on undeclared conflicts and ignores .conflicts.OK, so register mask.ok rules for S7 (@ over base, := over rlang and data.table). library() reads conflictRules() before loading the namespace, so this takes effect once S7's namespace is already loaded, e.g. imported by another package; a cold library(S7) under a strict policy still requires user-declared rules, as that policy intends.
S7 has no expectations about these conflicts; they are simply conflicts other than the masks S7 itself creates.
The mask filtering is an implementation detail already documented in the function comment; the qualifier only invited ambiguity.
hadley
reviewed
Aug 17, 2026
| data.table = ":=" | ||
| ) | ||
|
|
||
| # Re-emit the conflict report that library() would have produced (see |
Member
There was a problem hiding this comment.
Given we can switch to conflictRules(), I think we could drop this.
Member
Author
There was a problem hiding this comment.
conflictRules() only solves the case when rlang or data.table are loaded after S7. We also need a solution for when they're loaded before S7.
The .conflicts.OK binding is the only lever within our control there, and unfortunately, it's quite coarse: it only lets us opt into disabling all warnings. In which case, if there are other conflict warnings that we don't want to suppress (meaning, not :=), then it's our responsibility to report them.
t-kalinowski
changed the title
:= ahead of rlang and data.table
t-kalinowski
changed the title
t-kalinowski
marked this pull request as ready for review
August 24, 2026 15:39
hadley
approved these changes
Aug 27, 2026
| env[[".conflicts.OK"]] <- TRUE | ||
| # S7_at already supplies @ without conflicting with base::@. | ||
| env <- as.environment(paste0("package:", pkgname)) | ||
| rm(list = "@", envir = env) |
Co-authored-by: Hadley Wickham <h.wickham@gmail.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Fixes #697.
S7, rlang, and data.table all export
:=. Before this change, the version found on the search path depended on the order in which the packages were attached, and R could print a masking message.This PR makes S7's
:=win in either order. R no longer prints an attach-time message saying that one package's:=masks another's. Messages about other masked objects are still shown.When rlang or data.table is already attached, S7 changes only the
:=binding that the package added to the search path. It does not change the package namespace, so explicit access withrlang::`:=`ordata.table::`:=`is unchanged. The original search-path binding is restored if S7 is detached.The package vignette also shows how to import
:=from S7 when a package imports rlang or data.table.Testing