Stop one unparseable type from aborting the whole typecheck run - #1263
Open
apiology wants to merge 4 commits into
Open
Stop one unparseable type from aborting the whole typecheck run#1263apiology wants to merge 4 commits into
apiology wants to merge 4 commits into
Conversation
…or boundary Two issues found while reviewing PR castwide#1259: - ComplexType::TypeMethods#generate_substring_from's Hash fallback branch unconditionally emitted the 2-parameter `<K, V>` notation, even when key_types/subtypes held more than one type (from a comma-separated union on either side of a Hash{} literal, or from generics substitution rebuilding a Hash-named type with parameters_type out of sync with its key_types/subtypes). Reparsing the resulting 3+-parameter string raised Solargraph::ComplexTypeError, and since ApiMap#get_method_stack reparses a receiver type's rooted_tag unguarded, this could crash method resolution for that receiver. Now falls back to the `{K => V}` notation, which reparses correctly regardless of how many types are on either side. - TypeChecker#call_problems ran chain.infer (and argument_problems_for) for every call node in a file with no rescue around it, so any exception raised while inferring a single call's type aborted TypeChecker#problems entirely, silently losing every diagnostic for the rest of the file. Each call site's inference is now wrapped in a rescue that logs the error and reports it as a Problem scoped to that call, so one bad call site degrades to one reported problem instead of killing the whole run. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01H7Hb7H69hfyzgHJqFiEjgQ
apiology
force-pushed
the
fix-1259-typecheck-crash
branch
from
August 5, 2026 13:56
6dab664 to
c3cc59d
Compare
apiology
marked this pull request as ready for review
August 5, 2026 14:22
apiology
added a commit
to apiology/solargraph
that referenced
this pull request
Aug 5, 2026
apiology
marked this pull request as draft
August 31, 2026 17:29
Vince's pending review on PR 1263 flagged four comments as too long
or as changelogging (narrating a regression's discovery instead of
stating a still-true fact). Shorten the rescue comment in
type_checker.rb to one line, and drop or shrink the PR-comment-URL
narration in the two new complex_type_spec.rb examples and the new
type_checker_spec.rb example, since the it descriptions already say
what each test covers.
Two other comments on the same PR (type_methods.rb:201 and :206) ask
whether generate_substring_from should always use {K=>V} notation
instead of branching on <K,V> vs {K=>V} -- a design question, not a
style fix, left for Vince to decide.
The <K, V> form has room for exactly one type on each side, so a
union in either position produced a tag that failed to reparse.
Rather than choosing between the two notations at render time,
always emit {K => V}, which holds a comma-separated list on either
side. Hash<K, V> is still accepted as input; only the generated tag
changes.
This collapses the branch into the existing hash_parameters? case.
fixed_parameters? moves ahead of it so Hash(A, B) still renders as a
tuple.
The two round-trip specs now assert that reparsing the generated tag
yields the same tag, rather than only that it does not raise.
apiology
changed the title
apiology
marked this pull request as ready for review
September 5, 2026 14:51
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
1 participant
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.
Problem:
solargraph typecheckcan die partway through with a backtrace, leaving the rest of the workspace unchecked.No summary line, no problem count, and every file after the failing one goes unvisited — one bad type annotation costs the entire run.
Solution:
TypeChecker#call_problemsnow rescues each call site's inference and reports the failure as aProblemscoped to that call, and Hash types always render as{K => V}, which — unlike<K, V>— can represent ComplexTypes as keys and/or values.🤖 Generated with Claude Code
https://claude.ai/code/session_01RALWyFBErKxFodRMRzLnBu