-
Notifications
You must be signed in to change notification settings - Fork 2k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
implement bundling changes suggested in #4062 #4096
base: 16.x.x
Are you sure you want to change the base?
Conversation
Generated files look like this: index.js.d.mtsexport {
version,
versionInfo,
graphql,
graphqlSync,
resolveObjMapThunk,
resolveReadonlyArrayThunk,
GraphQLSchema,
GraphQLDirective,
GraphQLScalarType,
GraphQLObjectType,
GraphQLInterfaceType,
GraphQLUnionType,
GraphQLEnumType,
GraphQLInputObjectType,
GraphQLList,
GraphQLNonNull,
specifiedScalarTypes,
GraphQLInt,
GraphQLFloat,
GraphQLString,
GraphQLBoolean,
GraphQLID,
GRAPHQL_MAX_INT,
GRAPHQL_MIN_INT,
specifiedDirectives,
GraphQLIncludeDirective,
GraphQLSkipDirective,
GraphQLDeprecatedDirective,
GraphQLSpecifiedByDirective,
TypeKind,
DEFAULT_DEPRECATION_REASON,
introspectionTypes,
__Schema,
__Directive,
__DirectiveLocation,
__Type,
__Field,
__InputValue,
__EnumValue,
__TypeKind,
SchemaMetaFieldDef,
TypeMetaFieldDef,
TypeNameMetaFieldDef,
isSchema,
isDirective,
isType,
isScalarType,
isObjectType,
isInterfaceType,
isUnionType,
isEnumType,
isInputObjectType,
isListType,
isNonNullType,
isInputType,
isOutputType,
isLeafType,
isCompositeType,
isAbstractType,
isWrappingType,
isNullableType,
isNamedType,
isRequiredArgument,
isRequiredInputField,
isSpecifiedScalarType,
isIntrospectionType,
isSpecifiedDirective,
assertSchema,
assertDirective,
assertType,
assertScalarType,
assertObjectType,
assertInterfaceType,
assertUnionType,
assertEnumType,
assertInputObjectType,
assertListType,
assertNonNullType,
assertInputType,
assertOutputType,
assertLeafType,
assertCompositeType,
assertAbstractType,
assertWrappingType,
assertNullableType,
assertNamedType,
getNullableType,
getNamedType,
validateSchema,
assertValidSchema,
assertName,
assertEnumValueName,
Token,
Source,
Location,
OperationTypeNode,
getLocation,
printLocation,
printSourceLocation,
Lexer,
TokenKind,
parse,
parseValue,
parseConstValue,
parseType,
print,
visit,
visitInParallel,
getVisitFn,
getEnterLeaveForKind,
BREAK,
Kind,
DirectiveLocation,
isDefinitionNode,
isExecutableDefinitionNode,
isSelectionNode,
isValueNode,
isConstValueNode,
isTypeNode,
isTypeSystemDefinitionNode,
isTypeDefinitionNode,
isTypeSystemExtensionNode,
isTypeExtensionNode,
execute,
executeSync,
defaultFieldResolver,
defaultTypeResolver,
responsePathAsArray,
getArgumentValues,
getVariableValues,
getDirectiveValues,
subscribe,
createSourceEventStream,
validate,
ValidationContext,
specifiedRules,
ExecutableDefinitionsRule,
FieldsOnCorrectTypeRule,
FragmentsOnCompositeTypesRule,
KnownArgumentNamesRule,
KnownDirectivesRule,
KnownFragmentNamesRule,
KnownTypeNamesRule,
LoneAnonymousOperationRule,
NoFragmentCyclesRule,
NoUndefinedVariablesRule,
NoUnusedFragmentsRule,
NoUnusedVariablesRule,
OverlappingFieldsCanBeMergedRule,
PossibleFragmentSpreadsRule,
ProvidedRequiredArgumentsRule,
ScalarLeafsRule,
SingleFieldSubscriptionsRule,
UniqueArgumentNamesRule,
UniqueDirectivesPerLocationRule,
UniqueFragmentNamesRule,
UniqueInputFieldNamesRule,
UniqueOperationNamesRule,
UniqueVariableNamesRule,
ValuesOfCorrectTypeRule,
VariablesAreInputTypesRule,
VariablesInAllowedPositionRule,
LoneSchemaDefinitionRule,
UniqueOperationTypesRule,
UniqueTypeNamesRule,
UniqueEnumValueNamesRule,
UniqueFieldDefinitionNamesRule,
UniqueArgumentDefinitionNamesRule,
UniqueDirectiveNamesRule,
PossibleTypeExtensionsRule,
NoDeprecatedCustomRule,
NoSchemaIntrospectionCustomRule,
GraphQLError,
syntaxError,
locatedError,
printError,
formatError,
getIntrospectionQuery,
getOperationAST,
getOperationRootType,
introspectionFromSchema,
buildClientSchema,
buildASTSchema,
buildSchema,
extendSchema,
lexicographicSortSchema,
printSchema,
printType,
printIntrospectionSchema,
typeFromAST,
valueFromAST,
valueFromASTUntyped,
astFromValue,
TypeInfo,
visitWithTypeInfo,
coerceInputValue,
concatAST,
separateOperations,
stripIgnoredCharacters,
isEqualType,
isTypeSubTypeOf,
doTypesOverlap,
assertValidName,
isValidNameError,
BreakingChangeType,
DangerousChangeType,
findBreakingChanges,
findDangerousChanges,
} from './index.js';
export type {
GraphQLArgs,
GraphQLType,
GraphQLInputType,
GraphQLOutputType,
GraphQLLeafType,
GraphQLCompositeType,
GraphQLAbstractType,
GraphQLWrappingType,
GraphQLNullableType,
GraphQLNamedType,
GraphQLNamedInputType,
GraphQLNamedOutputType,
ThunkReadonlyArray,
ThunkObjMap,
GraphQLSchemaConfig,
GraphQLSchemaExtensions,
GraphQLDirectiveConfig,
GraphQLDirectiveExtensions,
GraphQLArgument,
GraphQLArgumentConfig,
GraphQLArgumentExtensions,
GraphQLEnumTypeConfig,
GraphQLEnumTypeExtensions,
GraphQLEnumValue,
GraphQLEnumValueConfig,
GraphQLEnumValueConfigMap,
GraphQLEnumValueExtensions,
GraphQLField,
GraphQLFieldConfig,
GraphQLFieldConfigArgumentMap,
GraphQLFieldConfigMap,
GraphQLFieldExtensions,
GraphQLFieldMap,
GraphQLFieldResolver,
GraphQLInputField,
GraphQLInputFieldConfig,
GraphQLInputFieldConfigMap,
GraphQLInputFieldExtensions,
GraphQLInputFieldMap,
GraphQLInputObjectTypeConfig,
GraphQLInputObjectTypeExtensions,
GraphQLInterfaceTypeConfig,
GraphQLInterfaceTypeExtensions,
GraphQLIsTypeOfFn,
GraphQLObjectTypeConfig,
GraphQLObjectTypeExtensions,
GraphQLResolveInfo,
ResponsePath,
GraphQLScalarTypeConfig,
GraphQLScalarTypeExtensions,
GraphQLTypeResolver,
GraphQLUnionTypeConfig,
GraphQLUnionTypeExtensions,
GraphQLScalarSerializer,
GraphQLScalarValueParser,
GraphQLScalarLiteralParser,
ParseOptions,
SourceLocation,
TokenKindEnum,
KindEnum,
DirectiveLocationEnum,
ASTVisitor,
ASTVisitFn,
ASTVisitorKeyMap,
ASTNode,
ASTKindToNode,
NameNode,
DocumentNode,
DefinitionNode,
ExecutableDefinitionNode,
OperationDefinitionNode,
VariableDefinitionNode,
VariableNode,
SelectionSetNode,
SelectionNode,
FieldNode,
ArgumentNode,
ConstArgumentNode,
FragmentSpreadNode,
InlineFragmentNode,
FragmentDefinitionNode,
ValueNode,
ConstValueNode,
IntValueNode,
FloatValueNode,
StringValueNode,
BooleanValueNode,
NullValueNode,
EnumValueNode,
ListValueNode,
ConstListValueNode,
ObjectValueNode,
ConstObjectValueNode,
ObjectFieldNode,
ConstObjectFieldNode,
DirectiveNode,
ConstDirectiveNode,
TypeNode,
NamedTypeNode,
ListTypeNode,
NonNullTypeNode,
TypeSystemDefinitionNode,
SchemaDefinitionNode,
OperationTypeDefinitionNode,
TypeDefinitionNode,
ScalarTypeDefinitionNode,
ObjectTypeDefinitionNode,
FieldDefinitionNode,
InputValueDefinitionNode,
InterfaceTypeDefinitionNode,
UnionTypeDefinitionNode,
EnumTypeDefinitionNode,
EnumValueDefinitionNode,
InputObjectTypeDefinitionNode,
DirectiveDefinitionNode,
TypeSystemExtensionNode,
SchemaExtensionNode,
TypeExtensionNode,
ScalarTypeExtensionNode,
ObjectTypeExtensionNode,
InterfaceTypeExtensionNode,
UnionTypeExtensionNode,
EnumTypeExtensionNode,
InputObjectTypeExtensionNode,
ExecutionArgs,
ExecutionResult,
FormattedExecutionResult,
SubscriptionArgs,
ValidationRule,
GraphQLErrorOptions,
GraphQLFormattedError,
GraphQLErrorExtensions,
IntrospectionOptions,
IntrospectionQuery,
IntrospectionSchema,
IntrospectionType,
IntrospectionInputType,
IntrospectionOutputType,
IntrospectionScalarType,
IntrospectionObjectType,
IntrospectionInterfaceType,
IntrospectionUnionType,
IntrospectionEnumType,
IntrospectionInputObjectType,
IntrospectionTypeRef,
IntrospectionInputTypeRef,
IntrospectionOutputTypeRef,
IntrospectionNamedTypeRef,
IntrospectionListTypeRef,
IntrospectionNonNullTypeRef,
IntrospectionField,
IntrospectionInputValue,
IntrospectionEnumValue,
IntrospectionDirective,
BuildSchemaOptions,
BreakingChange,
DangerousChange,
TypedQueryDocumentNode,
} from './index.js'; index.js.mjsexport {
version,
versionInfo,
graphql,
graphqlSync,
resolveObjMapThunk,
resolveReadonlyArrayThunk,
GraphQLSchema,
GraphQLDirective,
GraphQLScalarType,
GraphQLObjectType,
GraphQLInterfaceType,
GraphQLUnionType,
GraphQLEnumType,
GraphQLInputObjectType,
GraphQLList,
GraphQLNonNull,
specifiedScalarTypes,
GraphQLInt,
GraphQLFloat,
GraphQLString,
GraphQLBoolean,
GraphQLID,
GRAPHQL_MAX_INT,
GRAPHQL_MIN_INT,
specifiedDirectives,
GraphQLIncludeDirective,
GraphQLSkipDirective,
GraphQLDeprecatedDirective,
GraphQLSpecifiedByDirective,
TypeKind,
DEFAULT_DEPRECATION_REASON,
introspectionTypes,
__Schema,
__Directive,
__DirectiveLocation,
__Type,
__Field,
__InputValue,
__EnumValue,
__TypeKind,
SchemaMetaFieldDef,
TypeMetaFieldDef,
TypeNameMetaFieldDef,
isSchema,
isDirective,
isType,
isScalarType,
isObjectType,
isInterfaceType,
isUnionType,
isEnumType,
isInputObjectType,
isListType,
isNonNullType,
isInputType,
isOutputType,
isLeafType,
isCompositeType,
isAbstractType,
isWrappingType,
isNullableType,
isNamedType,
isRequiredArgument,
isRequiredInputField,
isSpecifiedScalarType,
isIntrospectionType,
isSpecifiedDirective,
assertSchema,
assertDirective,
assertType,
assertScalarType,
assertObjectType,
assertInterfaceType,
assertUnionType,
assertEnumType,
assertInputObjectType,
assertListType,
assertNonNullType,
assertInputType,
assertOutputType,
assertLeafType,
assertCompositeType,
assertAbstractType,
assertWrappingType,
assertNullableType,
assertNamedType,
getNullableType,
getNamedType,
validateSchema,
assertValidSchema,
assertName,
assertEnumValueName,
Token,
Source,
Location,
OperationTypeNode,
getLocation,
printLocation,
printSourceLocation,
Lexer,
TokenKind,
parse,
parseValue,
parseConstValue,
parseType,
print,
visit,
visitInParallel,
getVisitFn,
getEnterLeaveForKind,
BREAK,
Kind,
DirectiveLocation,
isDefinitionNode,
isExecutableDefinitionNode,
isSelectionNode,
isValueNode,
isConstValueNode,
isTypeNode,
isTypeSystemDefinitionNode,
isTypeDefinitionNode,
isTypeSystemExtensionNode,
isTypeExtensionNode,
execute,
executeSync,
defaultFieldResolver,
defaultTypeResolver,
responsePathAsArray,
getArgumentValues,
getVariableValues,
getDirectiveValues,
subscribe,
createSourceEventStream,
validate,
ValidationContext,
specifiedRules,
ExecutableDefinitionsRule,
FieldsOnCorrectTypeRule,
FragmentsOnCompositeTypesRule,
KnownArgumentNamesRule,
KnownDirectivesRule,
KnownFragmentNamesRule,
KnownTypeNamesRule,
LoneAnonymousOperationRule,
NoFragmentCyclesRule,
NoUndefinedVariablesRule,
NoUnusedFragmentsRule,
NoUnusedVariablesRule,
OverlappingFieldsCanBeMergedRule,
PossibleFragmentSpreadsRule,
ProvidedRequiredArgumentsRule,
ScalarLeafsRule,
SingleFieldSubscriptionsRule,
UniqueArgumentNamesRule,
UniqueDirectivesPerLocationRule,
UniqueFragmentNamesRule,
UniqueInputFieldNamesRule,
UniqueOperationNamesRule,
UniqueVariableNamesRule,
ValuesOfCorrectTypeRule,
VariablesAreInputTypesRule,
VariablesInAllowedPositionRule,
LoneSchemaDefinitionRule,
UniqueOperationTypesRule,
UniqueTypeNamesRule,
UniqueEnumValueNamesRule,
UniqueFieldDefinitionNamesRule,
UniqueArgumentDefinitionNamesRule,
UniqueDirectiveNamesRule,
PossibleTypeExtensionsRule,
NoDeprecatedCustomRule,
NoSchemaIntrospectionCustomRule,
GraphQLError,
syntaxError,
locatedError,
printError,
formatError,
getIntrospectionQuery,
getOperationAST,
getOperationRootType,
introspectionFromSchema,
buildClientSchema,
buildASTSchema,
buildSchema,
extendSchema,
lexicographicSortSchema,
printSchema,
printType,
printIntrospectionSchema,
typeFromAST,
valueFromAST,
valueFromASTUntyped,
astFromValue,
TypeInfo,
visitWithTypeInfo,
coerceInputValue,
concatAST,
separateOperations,
stripIgnoredCharacters,
isEqualType,
isTypeSubTypeOf,
doTypesOverlap,
assertValidName,
isValidNameError,
BreakingChangeType,
DangerousChangeType,
findBreakingChanges,
findDangerousChanges,
} from './index.js'; |
I just noticed that the folders are declared as official entry points: * This also includes utility functions for operating on GraphQL types and
* GraphQL documents to facilitate building tools.
*
* You may also import from each sub-directory directly. For example, the
* following two import statements are equivalent:
*
* ```ts
* import { parse } from 'graphql';
* import { parse } from 'graphql/language';
* ``` |
This still ended up as a dual module hazard in "import": "./jsutils/instanceOf.js.d.mts",
"default": "./jsutils/instanceOf.d.ts"
},
- "module": "./jsutils/instanceOf.mjs",
"import": "./jsutils/instanceOf.js.mjs",
+ "module": "./jsutils/instanceOf.mjs",
"default": "./jsutils/instanceOf.js"
}, The question is just if that has any unintended side effects... Asking Andarist on the side :) |
Just documenting this little nugget of knowledge before I never find it again:
|
Moving So instead, I've added a import { defineConfig } from 'vite';
export default defineConfig(async ({ mode }) => {
return {
resolve: mode === 'test' ? { conditions: ["dual-module-hazard-workaround"] } : undefined,
};
}); This would allow vitest users to avoid the dual package hazard without breaking anything else. |
@@ -6,11 +6,18 @@ | |||
}, | |||
"dependencies": { | |||
"graphql": "file:../graphql.tgz", | |||
"typescript-4.1": "npm:[email protected]", |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
These changes break TS 4.1, but that would probably be okay for a new major of graphql
.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
what exactly causes them to break TS 4.1 out of curiosity?
do we have to remove 4.3 through 4.8?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I was going with the DefinitelyTyped support window for TypeScript versions here.
No other reason to drop them than that not even Microsoft themself want to maintain types for those old versions any more - it seemed reasonable as I had to touch this already, and we won't be able to drop any TS versions between majors, so if we even want to remotely consider using modern features in the future, this would be the time.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
As for 4.1, it doesn't resolve the exports
, while all other versions do.
Tons of
basic-test.ts:1:38 - error TS7016: Could not find a declaration file for module 'graphql/execution'. '/private/var/folders/ms/w25rss0n4vq830bvptjjkqvr0000gp/T/graphql-js-integrationTmp/ts/node_modules/graphql/execution/index.js' implicitly has an 'any' type.
Try `npm i --save-dev @types/graphql` if it exists or add a new declaration (.d.ts) file containing `declare module 'graphql/execution';`
1 import type { ExecutionResult } from 'graphql/execution';
~~~~~~~~~~~~~~~~~~~
basic-test.ts:3:29 - error TS7016: Could not find a declaration file for module 'graphql'. '/private/var/folders/ms/w25rss0n4vq830bvptjjkqvr0000gp/T/graphql-js-integrationTmp/ts/node_modules/graphql/index.js' implicitly has an 'any' type.
Try `npm i --save-dev @types/graphql` if it exists or add a new declaration (.d.ts) file containing `declare module 'graphql';`
3 import { graphqlSync } from 'graphql';
~~~~~~~~~
basic-test.ts:4:65 - error TS7016: Could not find a declaration file for module 'graphql/type'. '/private/var/folders/ms/w25rss0n4vq830bvptjjkqvr0000gp/T/graphql-js-integrationTmp/ts/node_modules/graphql/type/index.js' implicitly has an 'any' type.
Try `npm i --save-dev @types/graphql` if it exists or add a new declaration (.d.ts) file containing `declare module 'graphql/type';`
4 import { GraphQLString, GraphQLSchema, GraphQLObjectType } from 'graphql/type';
~~~~~~~~~~~~~~
etc.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Ah. Then why don’t we remove support in a separate PR so it goes in as a separate breaking change
integrationTests/ts/package.json
Outdated
"typescript-5.0": "npm:[email protected]", | ||
"typescript-5.1": "npm:[email protected]", | ||
"typescript-5.2": "npm:[email protected]", | ||
"typescript-5.3": "npm:[email protected]", | ||
"typescript-5.4": "npm:[email protected]" |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
How many versions should be supported here?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I've chosen the same range as DefinitelyTyped for now: https://github.com/DefinitelyTyped/DefinitelyTyped?tab=readme-ov-file#support-window (>= 4.8)
We might choose to enforce more modern versions, though.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
this has been updated in main, maybe we can rebase this PR against main?
14dc95a
to
44516d2
Compare
'execution/execute.ts', | ||
'jsutils/instanceOf.ts', | ||
'language/parser.ts', | ||
'language/ast.ts', |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
These have been chosen by gut feeling out of the exports mentioned in #4074
We need to generally decide on a way forward here:
- expose those
.js
file? - instead, export these exports from the parent "nested" entrypoint?
- export them from the main entrypoint?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Labelling a function with export
but burying it so users would have to deep import it was a way to allow entryway into our private API without issuing any semver guarantee => a convenience to power users, so to speak.
With explicit exports, with option 1 and 2, we kind of lose that distinction, and it's much more difficult to garden off these bits as a private API that happens to be exposed.
So I vote that we examine each in term, and make the somewhat difficult decision of whether to support semver on them, and in that case go with option C, export from main entrypoint, or no longer expose them at all.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yeah, I feel similarly.
I would suggest that I remove all of these exports and the annotations I made in this PR, and we open a new PR to expose these after deciding which we want to expose?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Should the base for these changes be 16.x.x
=> are we trying to get this into v16 (current base branch: 16.x.x) or v17 (proposed base branch: main)?
@@ -6,11 +6,18 @@ | |||
}, | |||
"dependencies": { | |||
"graphql": "file:../graphql.tgz", | |||
"typescript-4.1": "npm:[email protected]", |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
what exactly causes them to break TS 4.1 out of curiosity?
do we have to remove 4.3 through 4.8?
'execution/execute.ts', | ||
'jsutils/instanceOf.ts', | ||
'language/parser.ts', | ||
'language/ast.ts', |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Labelling a function with export
but burying it so users would have to deep import it was a way to allow entryway into our private API without issuing any semver guarantee => a convenience to power users, so to speak.
With explicit exports, with option 1 and 2, we kind of lose that distinction, and it's much more difficult to garden off these bits as a private API that happens to be exposed.
So I vote that we examine each in term, and make the somewhat difficult decision of whether to support semver on them, and in that case go with option C, export from main entrypoint, or no longer expose them at all.
This was targeting the 16.x branch because at the time of discussing this in the WG, it wasn't clear when a release of the current v17 alphas as stable would be possible. If this has changed by now I can also target the v17 branch. Notes here: https://github.com/graphql/graphql-js-wg/blob/main/notes/2024/2024-04.md |
My feeling is that this has changed but perhaps @JoviDeCroock and @benjie might want to weigh in directly. |
This is a breaking change so, when accepted, it should go in the next major release. Ideally whether incremental delivery goes out in the next major release or not should be a separate question - I don't think we need to tie these two features together - but since it sounds from Yaacov like I'm a little hesitant because incremental delivery has had a big impact on the codebase, even when turned off, and that warrants careful scrutiny. |
@phryneas just thinking about this more, just want to check if I understand things correctly, the dual packages when using vitest stem from what exactly? I am guessing it is because vitest (1) uses the vite dev server (which uses esbuild vs. rollup vs both?) and picks up by default the build from the "module" condition and gets the pure esm package, while (2) vitest itself as a node test runner uses the import/default conditions? And so by adding this specific condition, we are telling esbuild via vite that for every other package, you can go ahead and continue to use the module condition, but with graphql-js you get a dual package hazard, so don't? If I understand things correctly, vitest will still give you dual package hazards for every other affected library besides graphql-js? Within esbuild itself, if you set any custom condition, or even an empty list, esbuild will not add the "module" condition. I am not sure if that carries through when vite calls whatever it calls in its pipeline? But if it does, then when we tell them to set our custom condition, then they never get module respected for any package? https://esbuild.github.io/api/#platform That would be odd.
Another workaround for vitest users is the suggestion here for modifying how vite resolves graphql-js, which will have a parallel for other affected libraries, but definitely not affect them? vitest-dev/vitest#4233 (comment) I may have most of the above wrong -- would love more information. Overall, I would love to avoid putting this dual-package-hazard condition in here, as it seems pretty confusing, the rest of the setup is designed to avoid the dual-package-hazard => this seems like more of an honest-to-goodness tricky bug within vitest and its dependencies. |
To supplement: there is a meta-issue documenting some of the problems with vitests's approach and some of the root causes and workarounds: vitest-dev/vitest#5486 I am leaning toward not trying to fix this with a workaround that switches vitest from using esm to our esm=>cjs wrapper, but just using the workarounds discussed there. |
@phryneas => do you have any interest in rebasing this on the v17 branch in |
I won't be able to get to this within the next 2-3 weeks or so - after that, I could give it a try. |
This is now ready for review.
Current AreTheTypesWrong output:
Running
publint
also is successful.