Add optionality to mapped type substitutions based on existence of an optional modifiers prop #57493
Conversation
… optional modifiers prop
typescript-bot
added
the
For Uncommitted Bug
|
|
||
| function hasOptionalModifiersProperty(mappedType: MappedType) { | ||
| const modifiersType = getApparentType(getModifiersTypeFromMappedType(mappedType)); | ||
| if (!(modifiersType.flags & TypeFlags.Object)) { |
There was a problem hiding this comment.
TODO: might also have to handle intersection (that's what isWeakType is doing) and I need to recheck behavior with unions
There was a problem hiding this comment.
recheck behavior with union types later
There was a problem hiding this comment.
@Mateusnasciment Please try to keep comments actionable and on-topic
|
|
||
| type StrictExtract<T, U> = T extends U ? U extends T ? T : never : never; | ||
| type StrictExclude<T, U> = T extends StrictExtract<T, U> ? never : T; | ||
| type A<T> = { [Q in { [P in keyof T]: P; }[keyof T]]: T[Q]; }; |
There was a problem hiding this comment.
I'm not sure what's the best strategy to avoid circularity here. I didn't look into it yet. I'm hitting the 🛌 but maybe somebody would chime in with some suggestions in the meantime ;p
| const templateMapper = combineTypeMappers(objectType.mapper, mapper); | ||
| return instantiateType(getTemplateTypeFromMappedType(objectType.target as MappedType || objectType), templateMapper); | ||
| const type = instantiateType(getTemplateTypeFromMappedType(objectType.target as MappedType || objectType), templateMapper); | ||
| return addOptionality(type, /*isProperty*/ true, hasOptionalModifiersProperty(objectType)); |
There was a problem hiding this comment.
TODO: this should get the list of known optional properties and check if index is assignable to that, if the substitution happens using a known non-optional property then the optionality shouldn't be added
|
I have what I think it the right fix in #57549. The main difference is that my PR is based on computing combined modifier flags, which I think is a better approach. It avoids resolving the members, which may cause circularities as you've observed. Also, member resolution is based on what members are known to be present, not which members might be present. |
5 participants
fixes #57487