CSE join: canonicalise the primitives in both environments #3182
Conversation
This file contains 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
|
@mshinwell You will probably want to look at this, although we should also discuss with @chambart next Tuesday. |
|
Let's add some more documentation about the subtleties of CSE as discussed. I will approve now, just merge once that's done. |
mshinwell
approved these changes
Oct 29, 2024
|
After some more discussions, it looks like there is a deeper problem that we need to think about properly. Since the bug report doesn't come for real code, we're putting this PR on hold until we have a better idea of how to solve the problem correctly. |
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Outdated suggestions cannot be applied.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.
Fixes #3181.
As explained in a comment there, when joining CSE equations we want to transform equations at the use site into equations at the join site (in the env at fork). There are multiple ways to do that: all aliases of primitive arguments that are in scope in the env at fork could be considered, which would yield the most precise result but risks exploding the number of CSE equations.
Before this PR we canonicalise in the environment at use, which handles the cases where primitives were called on local aliases of a "global" variable.
But we're failing in cases like the one from #3181, where a CSE equation is common to all branches but some branches have equations that change which element is canonical.
So this PR proposes to canonicalise in the env at fork in addition to the env at use. This potentially doubles the number of primitives, but in the common case where the branch doesn't introduce new relevant aliases the primitives are the same (using
EP.Set.tto represent these possibilities means we don't actually have to check that manually).