Remove DEFAULT_RUST_EDITION #1256
Conversation
dtolnay
force-pushed
the
required2
branch
2 times, most recently
from
8ad26ca to
980d54a
Compare
dtolnay
force-pushed
the
required2
branch
5 times, most recently
from
b47fd54 to
a723865
Compare
|
Sorry about all the force pushes -- it took a bit of iteration with CI to get all the platform-specific and I figure we do not want to set a default edition in the top-level WORKSPACE.bazel because doing so would mask problems such as #1254, so some new |
UebelAndre
left a comment
UebelAndre
left a comment
There was a problem hiding this comment.
I feel like there's something I'm not thinking about regarding the impact of this change but what I've come up with seems like there's good guidance on how to maintain the current behavior. So this looks good to me, give or take a few nits, thanks!
| crate_info = CrateInfo, | ||
| dep_info = DepInfo, | ||
| stdlib_info = StdLibInfo, | ||
| default_edition = DEFAULT_RUST_EDITION, |
There was a problem hiding this comment.
Maybe it's worth leaving this for a little bit to reduce the impact to public API? I'm torn though.
There was a problem hiding this comment.
What do you see as being the intended use of this API?
There was a problem hiding this comment.
I don't have any strong answers. Just noticing we're deleting a value from rust_common and thought maybe there should be more comms about this. But it can also be a note in the next release.
|
2 participants
This is 1 commit on top of #1255. It removes
DEFAULT_RUST_EDITIONas the defaulteditionargument ofrust_repositoriesand various other places. Instead, if the downstream WORKSPACE does not pass aneditiontorust_repositories, theeditionbecomes mandatory in everyrust_*target in their workspace, rather than using an arbitrary fallback default defined within rules_rust (previously 2018).This approach was suggested by @dfreese in #1255 (comment) and +1 by @UebelAndre in #1255 (review). I am ambivalent between them and would be happy with either this or #1255 by itself.