Fix potential soundness hole when adding references to a mapped capture set#18758
Merged
Linyxus merged 3 commits intoscala:mainfrom Oct 30, 2023
Merged
Fix potential soundness hole when adding references to a mapped capture set#18758Linyxus merged 3 commits intoscala:mainfrom
Linyxus merged 3 commits intoscala:mainfrom
Conversation
odersky
added
the
cc-experiment
Fix soundness hole when adding references to a set that is the image of an
idempotent `tm` map `tm`. If the element `ref` did not come from the source
of the set, we still assumed that `tm(ref) = ref`, so that we simply added
the reference to the set and also back-propagated it to source. But that is not
necessarily the case (although it is the case in our complete test suite,
so I am not sure this case can actually arise in practice. Nevertheless,
it's better to not leave a potential soundness hole here.
In the new implementation, we test whether `tm(ref) = ref`, and only proceed
as before if that's the case. If not there are two sub-cases:
- `{ref} <:< tm(ref)` and the variance of the set is positive. In that case we
can soundly add `tm(ref)` to the set while back-propagating `ref` as before.
- Otherwise there's nothing obvious left to do except fail (which is always
sound.
odersky
changed the title
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
3 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Fix potential soundness hole when adding references to a set that is the image of an
idempotent
tmmaptm. If the elementrefdid not come from the sourceof the set, we still assumed that
tm(ref) = ref, so that we simply addedthe reference to the set and also back-propagated it to source. But that is not
necessarily the case. (Although it is the case in our complete test suite,
so I am not sure this case can actually arise in practice. Nevertheless,
it's better to not leave a potential soundness hole here.)
In the new implementation, we test whether
tm(ref) = ref, and only proceedas before if that's the case. If not there are two sub-cases:
{ref} <:< tm(ref)and the variance of the set is positive. In that case wecan soundly add
tm(ref)to the set while back-propagatingrefas before.sound).