Use level checking for Fresh instances #23101
Merged
+1,862
−837
Conversation
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
odersky
force-pushed
the
drop-disallow-fresh
branch
4 times, most recently
from
b58327d to
a178c04
Compare
|
Most of the added lines in this PR are in the tests. We have more tests and the existing neg tests explain the problem in more detail. |
natsukagami
reviewed
May 12, 2025
Refactor to avoid generating refsText ahead of time. We need to keep the original capture sets or tree lists so that we can record them as seen.
- Map cap to Result in their original types (was Fresh before) - Map Result to Fresh on access
Turn level checking on under the useFreshLevels config option (which is on by default)
Fall back to previous scheme if there is no recording in Seen. That way, we see what we expect when turning on Ycc-verbose and other printing options.
…bles. The check was not working properly before, which meant that problems went undetected. To make things work again, we need two additional fixes - A more detailed treatment of result mapping for inferred types. In these, we map root.Fresh to root.Result only if the original inferred type was dependent. AppliedType functions are also mapped to RefinedType functions in setup, but this should not count for fresh to result mappings. - A fix in the Fresh to Result mapping of parameterless defs.
When printing types deriving from capability, do print their capture sets even if these are implied. This avoids confusing errors where a capability classes has two different implied Fresh instances which are both elided so the types look the same. Exception: When printing a singleton type (x: T) we do elide in T, since the usually the only part that matters is the `x`. But we don't do the elision under -Ycc-verbose or -explain.
We had the wrong condition in maxSubsumes before. We need to test for isStaticOwner instead of isStatic. The previous test made all fresh capabilities in global classes slip through the net.
odersky
force-pushed
the
drop-disallow-fresh
branch
from
b0f03a0 to
2f135cc
Compare
|
All comments should be addressed now. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
This PR changes the way we treat fresh instances. Instead of disallowing all of them when instantiating type parameters (as well as in a number of other situations), we allow them but check that inner fresh instances do not get widened to outer ones. This makes code like this one compile
Previously, that would not work since the
XandYvariables ofPairwould be instantiated toRef^, which was not allowed.The PR also implements lots of improvements to error diagnostics.
^,cap, or=>.