Store symbol definitions using NodeKey #11121
Merged
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
|
bf3a182 to
fbea550
Compare
MichaReiser
approved these changes
Apr 24, 2024
MichaReiser
pushed a commit
that referenced
this pull request
Apr 25, 2024
## Summary Indexing definitions of symbols is necessary for fast lazy type evaluation, so that when we see a reference to a name we can figure out the type of that symbol from its definition(s). We only want to do this indexing of definitions once per module, and then it needs to remain available in the cache thereafter for use by lazy type evaluation. Rust lifetimes won't let us use direct references to AST nodes owned by the cache. We could use unsafe code to strip the lifetimes from these references, with safety ensured by our cache invalidation: if we evict the AST from the cache, we must evict the symbol definitions also. But in order to serialize such a cache to disk, we would need (at least) an AST numbering scheme. This may still be something to look into in the future, for improved performance. For now, use `NodeKey`: indirect references to an AST node consisting of a `NodeKind` and a `TextRange`, which we can find again reasonably quickly in the AST. These are easy to serialize, have no lifetime problems, and don't require unsafe code. ## Test Plan Updated tests.
MichaReiser
pushed a commit
that referenced
this pull request
Apr 27, 2024
## Summary Indexing definitions of symbols is necessary for fast lazy type evaluation, so that when we see a reference to a name we can figure out the type of that symbol from its definition(s). We only want to do this indexing of definitions once per module, and then it needs to remain available in the cache thereafter for use by lazy type evaluation. Rust lifetimes won't let us use direct references to AST nodes owned by the cache. We could use unsafe code to strip the lifetimes from these references, with safety ensured by our cache invalidation: if we evict the AST from the cache, we must evict the symbol definitions also. But in order to serialize such a cache to disk, we would need (at least) an AST numbering scheme. This may still be something to look into in the future, for improved performance. For now, use `NodeKey`: indirect references to an AST node consisting of a `NodeKind` and a `TextRange`, which we can find again reasonably quickly in the AST. These are easy to serialize, have no lifetime problems, and don't require unsafe code. ## Test Plan Updated tests.
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.
Summary
Indexing definitions of symbols is necessary for fast lazy type evaluation, so that when we see a reference to a name we can figure out the type of that symbol from its definition(s).
We only want to do this indexing of definitions once per module, and then it needs to remain available in the cache thereafter for use by lazy type evaluation.
Rust lifetimes won't let us use direct references to AST nodes owned by the cache.
We could use unsafe code to strip the lifetimes from these references, with safety ensured by our cache invalidation: if we evict the AST from the cache, we must evict the symbol definitions also. But in order to serialize such a cache to disk, we would need (at least) an AST numbering scheme. This may still be something to look into in the future, for improved performance.
For now, use
NodeKey: indirect references to an AST node consisting of aNodeKindand aTextRange, which we can find again reasonably quickly in the AST. These are easy to serialize, have no lifetime problems, and don't require unsafe code.Test Plan
Updated tests.