WIP: Prism work, rebase with squashed commits #404
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
Co-Authored-By: Vinicius Stock <[email protected]>
And move all other Prism related code to the new parser/prism component
Co-Authored-By: Alex Momchilov <[email protected]>
Co-Authored-By: Emily Samp <[email protected]>
To make sure our translation layer correctly translates node locations, we create a task that, for every Prism regression test, runs the test with location comparison on.
Co-Authored-By: Alexander Momchilov <[email protected]>
Up to this point, in order to build Prism, we've needed to manually pass the path to the correct RUBY version to the build process because Bazel does not take the user's environment into account and will just use the system's default Ruby version (which is often very old). This is a workaround for our workaround, where we use a Bazel `repository_rule` to get the `RUBY_ROOT` environment variable, write it to a file within the Bazel file system, and then pull the value from that file while building Prism.
This debug config launches the main Sorbet executable to type check the provided file path (with Prism).
In our `downcast` helper, we check that the node argument is actually a Prism node. However, doing this requires us to check the `type` field, which will fail if the node is a `nullptr`. By doing a `nullptr` check in the `ENFORCE`, we guarantee we won't crash by checking for a field on a `nullptr`. Co-authored-by: Alex Momchilov <[email protected]>
By not replacing `/` with `_`, the test naming structure is more consistent with normal regression tests. But more importantly, it allows us to run location tests that are under a folder like `error_recovery/assign`, more easily.
The error location doesn't match what Sorbet currently generates. But the idea is to first make parse trees match, accumulate more recovery tests, and then come up with a strategy to generate errors that match locations.
It turns out that we only need `sourceMap`'s value to be set to the workspace folder with any arbitrary key. So not setting `BAZEL_EXEC_ROOT` is fine.
1. It fixes the case where a hash pattern doesn't have a value, which will generate a `PM_IMPLICIT_NODE` that we currently don't handle. 2. It fixes the translation of hash patterns with a named splat node, like `in **o`.
The logic of `translateHash` was kind of confusing because it was handling both vanilla hashes and also assocs/assoc splat nodes. By isolating one part of that method, which generates key/value pairs, and then calling that method in both cases, the two code paths become cleaner and more obvious.
When the method call is `[]`, the message location should be just the begin location of `[]`.
And exclude failing tests
In cases like `foo.()`, the message would be `:call`, but the message location would have null start and end. In this case, we need to fall back to use the call operator location.
This is not covered in parse tree location tests, but reflected in symbol table tests: when we write `FOO = 1` (or other constant write nodes), the location of `FOO` should just be the name, not the entire assignment. In those cases, we should use `name_loc` instead of `base.location`.
When the symbol is used as `a: 1`, the location should not include the colon.
The main blocker for upstreaming Prism parser to Sorbet is that we currently rely on Ruby to build Prism, as its required by Prism's `rake template` task. But by package the files generated by `rake template` in Prim's release, we will be able to remove the dependency on Ruby while significantly simplify the build configurations for Prism.
When ENFORCE macros fail, they use fmt (via spdlog) to format error messages. Add a formatter specialization for pm_node_type to properly display Prism node types in these error messages. Instead of showing numeric values, the formatter uses pm_node_type_to_str to display human-readable node type names.
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 is a test for upstreaming our Prism work. In this PR, I've rebased on Prism master and squashed many of our commits (398 commits --> 117)
Motivation
Test plan
See included automated tests.