use a TypeContext instead of extending the ParserState #6
Conversation
| parse_ts_type(p, context.and_conditional_type_allowed(true)) | ||
| .or_add_diagnostic(p, expected_ts_type); | ||
| p.expect(T![:]); | ||
| parse_ts_type(p).or_add_diagnostic(p, expected_ts_type); | ||
| parse_ts_type(p, context.and_conditional_type_allowed(true)) | ||
| .or_add_diagnostic(p, expected_ts_type); |
There was a problem hiding this comment.
Nit: conditional_type_allowed is guaranteed to be true because of the is_conditonal_type_allowed check above.
| parse_ts_type_parameter_name(p).or_add_diagnostic(p, expected_ts_type_parameter); | ||
| p.expect(T![in]); | ||
| parse_ts_type(p).or_add_diagnostic(p, expected_ts_type); | ||
| parse_ts_type(p, TypeContext::default()).or_add_diagnostic(p, expected_ts_type); |
There was a problem hiding this comment.
TypeContext::default should only used in positions that start a new "type-parsing" context (e.g. because it is parenthesized or because it is in an initializer position). It's otherwise best to pass the context to all parse_ts_type functions
| parse_ts_type(p, TypeContext::default()).or_add_diagnostic(p, expected_ts_type); | |
| parse_ts_type(p, context).or_add_diagnostic(p, expected_ts_type); |
The same applies to many other positions (e.g. mapped types, tuple types etc)
There was a problem hiding this comment.
The only thing that we need to be careful about is to make sure to pass context through correctly and only use TypeContext::default in positions where a type starts a complete new context (because it is parenthesized)
It is really hard for me to pass context through correctly. I tried, but I couldn't do that. There are too many functions which the context should be passed to.
nissy-dev
deleted the
support-extends-constraints-on-infer-type-refactor
branch
Summary
Test Plan
Documentation