"match may not be exhaustive" from synthetic code #12602
Comments
|
Great, thank you! |
|
Actually, the fix is incorrect and the warning is correct: because |
liufengyun
added a commit
to dotty-staging/dotty
that referenced
this issue
May 25, 2021
liufengyun
added a commit
to dotty-staging/dotty
that referenced
this issue
May 25, 2021
|
... right. But then the synthetic Why is |
Indeed, the typeclass derivation code should not be generated in the first place. We miss a condition somewhere. |
liufengyun
added a commit
to dotty-staging/dotty
that referenced
this issue
May 26, 2021
… abstract class
EnzeXing
added a commit
to EnzeXing/dotty
that referenced
this issue
Jun 11, 2021
Add tests Add tests Add tests Update Desugar.scala Update InteractiveDriver.scala Update closureLeak.scala Update default-this.scala Update enum.scala Update leak-this-inner.scala Update leak-this.scala Update leak-warm.scala Update local-warm.scala updated test cases updated-default-this updated tests Add tests for cycle structure Add desugared version for enum.scala Update Desugar.scala Update InteractiveDriver.scala Update closureLeak.scala Update default-this.scala Update enum.scala Update leak-this-inner.scala Update leak-this.scala Update leak-warm.scala Update local-warm.scala Comment Typo Fix in Concat Correctly check experimental features - When inlining: emit "experimental" error only when the feature in question is actually annotated as experimental - Compiler-wide: treat non-bootstrapped compiler as experimental-friendly. The motivation is to compile the standard library with experimental features enabled at all times. This change isn't supposed to introduce any experimental features leaks into stable builds since we do not publish non-bootstrapped compiler. Fix scala#12128: Add test Don't force info to get signature in TreeChecker When we are reading a new definition from tasty and setting `info` of a symbol in `readNewDef`: sym.info = methodType(paramss, resType) The line above will reach `TreeChecker.transformSym`: assert(symd.signature == initial.signature At the time, the `symd` is not yet completed. Getting signature of the denotation will force the symbol and trigger completion of the info, which reaches the `PositionTree` in `readIndexedDef. Fix regression test for scala#12128 To properly reproduce scala#12128, the regression test has some atypical classpath requirements in addition to separate compilation. Snippet compiler PoC Reporter fix. Message levels Fix snippet wrapper. Integrate snippet checker with Wiki and Markdown renderers Further implementation of snippet compiler. Add package name and self type to pass context Add arguments to control snippet compiler. Start work on handling hide directives in code snippets Hide irrelevant code snippet fragments in scaladoc Apply review suggestions from PR scala#11822 Add returning real line of error in source file of snippet for snippet scaladoc compiler Add failing flag Add debug flag. Add addition operation for snippet compiler arg Refactor snippet compiler args Move snippet compiler logic to parsing phase. Refactor debug flag Display snippet compilation messages in scaladoc Apply review changes Run snippet compiler on static sites Further enhancements of snippet contexts Filter non-scala snippets. Report errors in parsing override flags Fix positions in static sites. Enchance snippet wrapping CRs Add and update tests for snippet compiler. Bugfixes found during testing Wrap compiler messages into Try block to avoid lazy exceptions Update branch Fix classpath problems. Add support for scala.js. Add bootclasspath to generateDocumentation task Temporarily turn off snippet checking for scala3 library Change requests from reviews Scala2Unpickler: Don't trust the owner field of tparams When a type parameter of a class external to the current pickle is referenced, Scala 2 will sometimes pickle it as if it was a type parameter of the current class. When we unpickle such a type parameter on our side, we enter it in the class, this leads the compiler to believe the class has more type parameters than it actually has. This doesn't happen in Scala 2 itself because it never enters unpickled type parameters in the class that owns them (it must be recreating them in some way, but I don't know how). It would be interesting to dig deeper to understand exactly how Scala 2 handles type parameter unpickling so we could emulate it given that this is not the first time we run into trouble here (cf scala#12120), but for now we fix the problem with the following heuristic: once we've loaded all the type parameters of a class (which we do eagerly via `ClassUnpickler#init()`), we simply stop from being entered any subsequent type parameter we see for this class. Time will tell if this is good enough. While I was touching that code, I also made ClassCompleter#completerTypeParams idempotent as one would expect it to be. Fixes scala#12641. Preserve hard unions in more situations There's multiple places where we take apart a union, apply some transformation to its parts, then either return the original union if nothing changed or recombine the transformed parts into a new union. The problem is that to check if nothing changed we use referential equality when `=:=` would be more correct, so we sometimes end up returning a union equivalent to the original one except the new one is a soft union. I've seen this lead to subtle type inference differences both when experimenting with changes to constraint solving and when trying to replace our Uniques hashmaps by weak hashmaps. We could fix this by replacing these `eq` calls by `=:=` but instead this commit takes the approach of preserving the softness of a union even when its components are modified, see the test case. Fix implicit ValueOf synthesis Foxes scala#12583 Fix hole in scan for outer references Type tests of singletons need to be considered as well. Fix typo in syntax.md The supertype symbol was ':>' and it was changed to '>:' . Always generate a partial function from a lambda `scalac` no longer complains, neither should `dotc`. I verified that the output of `i4241.scala` is the same for both `scalac` and `dotc`. Fixes scala#12661 drop unused code Fix test name Extract the function type and SAM in union type (scala#11760) Fixes scala#11694. forbid given patterns in val definitions. Suggest that they use an alias given instead. Fix scala#12544: Add test Add neg test Harden type comparer in presence of kind errors It should not crash when an override has the wrong kind. Fixes scala#12664 Fix scala#12546: Handle enum or object in provablyDisjoint A enum or object is provably disjoint with a class A if the module class does not derive from A. Better support type-heavy pattern matches Avoid type cast Fix unreachable code Fix CI Handle enums that have multiple parents Fix scala#12559: Try decomposable types in isSubspace Fix scala#12602: Only generate sum mirrors for sealed trait or sealed abstract class Fix scala#12681: Approximate `ThisType` in child instantiation Fix master: Remove test We will add the test again with a new PR fix scala#12634 - port sbt/zinc#979 Add sealedDescendants method to SymDenotation - recursive children of a symbol Add Matchable to the parents of Null in explicit nulls Add icons for all members. Add icons for search results. Re-enable the Scala.js test JSOptionalTest212FunParamInference. The issue scala#11694, which prevented this test from compiling, was fixed in scala#11760. Bump compiler version to the next one – 3.0.2-RC1 fix scala#12401: mangle ctor names in ExtractAPI Avoid assertion failure when forcing LazyRef while printing Fixes scala#12650 The fix that matters for scala#12650 is the one in Checking. The others generally increase robustness of printing diagnostics when triggering a LazyRef recursion. Drop flaky test Fixes scala#12693 Deskolemize PatternTypeConstrainer Before this commit, constrainSimplePatternType did a bunch of manipulations and then ran a subtype check. It wrapped one of the types in a SkolemType and replaces type arguments of the other one with wildcards. The point of those manipulations was to make TypeComparer take a specific path and compare appropriate subcomponents of constrainSimplePatternType's arguments. This turned out to be a problem overall. To improve the situation, we instead manually compare the appropriate subcomponents of constrainSimplePatternType - the changes can be seen as inlining pieces of code from TypeComparer. Also, this commit adds restoring the GADT constraint to TypeComparer - this turned out to be more important now that constrainSimplePatternType now does multiple isSubType checks. Fix failing test Add explanatory comments Adjust comments Insert conversions also on selections wrapped in type applications In i12708.scala, the problematic function was a selection `qual.m[tvs]` that was already applied to type variables. In that case we need to backtrack, forget the type variables and try to insert a conversion or extension method on `qual`. Fixes scala#12708 Fix tests Add 3.0.1-RC1 blog article Remove experimental macro feature mention Co-authored-by: Nicolas Stucki <nicolas.stucki@gmail.com> macros.md: fix parentheses/braces in inlined Macros.assert() example The parantheses and braces where swapped. Also align the style with the follow up example to make it easier to spot the differences. Add regression test for scala#12723 Deleted unnecessary explanations As a result of replacing 'Dotty' with 'Scala 3' in the following commit, the explanation 'which is scheduled to become Scala 3' is no longer needed. scala@ccb76da Improve display of references for ambiguous implicit errors If the first attempt yields equal strings, print the full path to the implicits instead. Fixes scala#12591 Reclassify test Add check file Improvements to search UX Pressing "S" while not focused on an input/textarea will now open the search. Pressing "F" while not focused on an input/textarea will focus the member filter input. Pressing Escape while focused on the search/filter will clear the input and close/unfocus it Search result links are now treated as block elements so you can click anywhere inside the rectangle rather than specifically on the result text. also allow '/' to trigger the search Add browsable versions extension for scaladoc Change local storage to session storage, fix bugs with undefined js variable Chagne try catching of undefined scalajs property to js.typeOf Add test for infix operator usage from Scala 2 libraries Fixes scala#7304 Preserve tvar instantiation in implicit error messages make default colors less harsh, add dark/light theme swapper make dottydoc anchors transparent instead of white when not hovered infer dark theme preference from user's system add some 'missing' semicolons to some JS set min-width on the theme switch so it doesn't get squished Make dark-mode footer actually dark. Improve contrast/accessibility. Used the "poor man's dark mode" approach to deal with the images in the footer in dark mode, i.e. invert and hue rotate (without the hue rotate, the red scala3doc logo becomes blue) Also fix an issue where if you click "back to top" and then refresh the page, the "container" element gains the "expand" class, which causes the main signature (i.e. `class List[A] ...`) to become 'inline' instead of a block, and it also triggers the 6.5em left-margin that's intended for method signatures when they get expanded. Also consolidated the `footer` styles, since they were spread all over. Check CLA for user who opened pull request, not workflow initiator Add regression test Closes scala#12754 Don't allow wildcard types in constraints Fixes scala#12677 Only drop wildcards for necesessary comparisons Wildcards in constraints essentially mean that the constrained variable is a sub/supertype of _every_ instance of the bound. For useNecesaryEither, this is wrong. Approximate the bound not to use wildcards instead. drop all kinds of WildcardType under useNecessaryEither better bounds for wildcards in resultTypeApprox Avoid wildcard types in constraints Ther semantics is fishy. Instead, use avoidance in necessarySubType and in TypeVarsMiss context mode, and use fresh type variables elsehwere. Cleanup addOneBound logic Verify that constraints do not contain wildcard types Cleanups Streamline approximateWildcards logic Test assertion Don't let wildcards enter constraints when adding type lambdas Fix typo Turn off "no wildcards in constraint" check Address review comments macros.md: Fix htmlized variant of first source code example Confusingly, the last part of the first source code example would appear as def showExpr(expr: Expr[Boolean])(using Quotes): Expr[String] = '{ " " } in the htmlized version of the markdown documentation source. Which reads as follows: def showExpr(expr: Expr[Boolean])(using Quotes): Expr[String] = '{ "<some source code>" } // Better implementation later in this document This seems to be cause by the usage of angle brackets. Also note that the code comment does not appear in the htmlized output. Fix typo in 3.0.1-RC1 release notes Fix scala#9588: Enable Scala.js test ObjectTest. The issue preventing that test from compiling in the past was fixed, although we don't know when or by what. Scala.js: Mangle test names when generating JUnit bootstrappers. This fixes the last remaining JUnit-related tests, which we therefore enable. Copy old site versions for Scala 3 instead of Dotty 0.X Support both 0.X and 3.X lines For extension label, avoid scala style Fix escaping illegal tokens in snippet renderer in scaladoc Use sbt makeScalaInstance Add definition Add doc for env Fix compilation Refactor trace Support Value.source for friendly error message Disallow non-hot arguments to non-constructors Handle access of parameters in constructors Handle exception No exceptions Fix Env.isHot Add missing checks for arguments Handle new expression with only cold args Fix soundness test Reorganize tests Add test Add test for secondary constructor Refactor promotion Add test More logging Fix bug with promotion of this The previous code does not work, because private fields are not included. Refactor code: introduce ArgInfo instead of Value.source Handle O.this outside of it's definition Fix owner of values with static flag With the flag <static>, the enclosing class will be skipped. See the documentation `SymDenotation.enclosingClass` Add documentation about caching Add test Add another neg test for promotion of ThisRef Handle outers of trait as the same in concrete semantics For traits, its outers will be proxy methods of the class that extends the trait. As the prefix is stable and is a valid value before any super constructor calls. Therefore, we may think the outers for traits are immediately set following the class parameters. Also, when trying promotion of warm values, we never try warm values whose fields are not fully filled -- which corresponds to promote ThisRef with non-initialized fields, and errors will be reported when the class is checked separately. Address review (thanks @michelou) Rename isFullyInitialized to isFullyFilled to avoid confusion Restrict function arguments to be hot This aligns with the design of restricting method arguments to be hot. We only allow non-hot to constructors. Add documentation Address review Simplify logic for checking local classes inside secondary constructors Address review: simplify logic Abstract parameters of non-local constructor parameters as cold This is a simple fix which should have no impact on expressiveness and usability, as local classes inside secondary constructors are extremely rare. Update docs Fix scala#11973: Add regression test community build: Update munit Add Libretto to community build. re-organize tests
EnzeXing
added a commit
to EnzeXing/dotty
that referenced
this issue
Jun 11, 2021
Add tests Add tests Add tests Update Desugar.scala Update InteractiveDriver.scala Update closureLeak.scala Update default-this.scala Update enum.scala Update leak-this-inner.scala Update leak-this.scala Update leak-warm.scala Update local-warm.scala updated test cases updated-default-this updated tests Add tests for cycle structure Add desugared version for enum.scala Update Desugar.scala Update InteractiveDriver.scala Update closureLeak.scala Update default-this.scala Update enum.scala Update leak-this-inner.scala Update leak-this.scala Update leak-warm.scala Update local-warm.scala Comment Typo Fix in Concat Correctly check experimental features - When inlining: emit "experimental" error only when the feature in question is actually annotated as experimental - Compiler-wide: treat non-bootstrapped compiler as experimental-friendly. The motivation is to compile the standard library with experimental features enabled at all times. This change isn't supposed to introduce any experimental features leaks into stable builds since we do not publish non-bootstrapped compiler. Fix scala#12128: Add test Don't force info to get signature in TreeChecker When we are reading a new definition from tasty and setting `info` of a symbol in `readNewDef`: sym.info = methodType(paramss, resType) The line above will reach `TreeChecker.transformSym`: assert(symd.signature == initial.signature At the time, the `symd` is not yet completed. Getting signature of the denotation will force the symbol and trigger completion of the info, which reaches the `PositionTree` in `readIndexedDef. Fix regression test for scala#12128 To properly reproduce scala#12128, the regression test has some atypical classpath requirements in addition to separate compilation. Snippet compiler PoC Reporter fix. Message levels Fix snippet wrapper. Integrate snippet checker with Wiki and Markdown renderers Further implementation of snippet compiler. Add package name and self type to pass context Add arguments to control snippet compiler. Start work on handling hide directives in code snippets Hide irrelevant code snippet fragments in scaladoc Apply review suggestions from PR scala#11822 Add returning real line of error in source file of snippet for snippet scaladoc compiler Add failing flag Add debug flag. Add addition operation for snippet compiler arg Refactor snippet compiler args Move snippet compiler logic to parsing phase. Refactor debug flag Display snippet compilation messages in scaladoc Apply review changes Run snippet compiler on static sites Further enhancements of snippet contexts Filter non-scala snippets. Report errors in parsing override flags Fix positions in static sites. Enchance snippet wrapping CRs Add and update tests for snippet compiler. Bugfixes found during testing Wrap compiler messages into Try block to avoid lazy exceptions Update branch Fix classpath problems. Add support for scala.js. Add bootclasspath to generateDocumentation task Temporarily turn off snippet checking for scala3 library Change requests from reviews Scala2Unpickler: Don't trust the owner field of tparams When a type parameter of a class external to the current pickle is referenced, Scala 2 will sometimes pickle it as if it was a type parameter of the current class. When we unpickle such a type parameter on our side, we enter it in the class, this leads the compiler to believe the class has more type parameters than it actually has. This doesn't happen in Scala 2 itself because it never enters unpickled type parameters in the class that owns them (it must be recreating them in some way, but I don't know how). It would be interesting to dig deeper to understand exactly how Scala 2 handles type parameter unpickling so we could emulate it given that this is not the first time we run into trouble here (cf scala#12120), but for now we fix the problem with the following heuristic: once we've loaded all the type parameters of a class (which we do eagerly via `ClassUnpickler#init()`), we simply stop from being entered any subsequent type parameter we see for this class. Time will tell if this is good enough. While I was touching that code, I also made ClassCompleter#completerTypeParams idempotent as one would expect it to be. Fixes scala#12641. Preserve hard unions in more situations There's multiple places where we take apart a union, apply some transformation to its parts, then either return the original union if nothing changed or recombine the transformed parts into a new union. The problem is that to check if nothing changed we use referential equality when `=:=` would be more correct, so we sometimes end up returning a union equivalent to the original one except the new one is a soft union. I've seen this lead to subtle type inference differences both when experimenting with changes to constraint solving and when trying to replace our Uniques hashmaps by weak hashmaps. We could fix this by replacing these `eq` calls by `=:=` but instead this commit takes the approach of preserving the softness of a union even when its components are modified, see the test case. Fix implicit ValueOf synthesis Foxes scala#12583 Fix hole in scan for outer references Type tests of singletons need to be considered as well. Fix typo in syntax.md The supertype symbol was ':>' and it was changed to '>:' . Always generate a partial function from a lambda `scalac` no longer complains, neither should `dotc`. I verified that the output of `i4241.scala` is the same for both `scalac` and `dotc`. Fixes scala#12661 drop unused code Fix test name Extract the function type and SAM in union type (scala#11760) Fixes scala#11694. forbid given patterns in val definitions. Suggest that they use an alias given instead. Fix scala#12544: Add test Add neg test Harden type comparer in presence of kind errors It should not crash when an override has the wrong kind. Fixes scala#12664 Fix scala#12546: Handle enum or object in provablyDisjoint A enum or object is provably disjoint with a class A if the module class does not derive from A. Better support type-heavy pattern matches Avoid type cast Fix unreachable code Fix CI Handle enums that have multiple parents Fix scala#12559: Try decomposable types in isSubspace Fix scala#12602: Only generate sum mirrors for sealed trait or sealed abstract class Fix scala#12681: Approximate `ThisType` in child instantiation Fix master: Remove test We will add the test again with a new PR fix scala#12634 - port sbt/zinc#979 Add sealedDescendants method to SymDenotation - recursive children of a symbol Add Matchable to the parents of Null in explicit nulls Add icons for all members. Add icons for search results. Re-enable the Scala.js test JSOptionalTest212FunParamInference. The issue scala#11694, which prevented this test from compiling, was fixed in scala#11760. Bump compiler version to the next one – 3.0.2-RC1 fix scala#12401: mangle ctor names in ExtractAPI Avoid assertion failure when forcing LazyRef while printing Fixes scala#12650 The fix that matters for scala#12650 is the one in Checking. The others generally increase robustness of printing diagnostics when triggering a LazyRef recursion. Drop flaky test Fixes scala#12693 Deskolemize PatternTypeConstrainer Before this commit, constrainSimplePatternType did a bunch of manipulations and then ran a subtype check. It wrapped one of the types in a SkolemType and replaces type arguments of the other one with wildcards. The point of those manipulations was to make TypeComparer take a specific path and compare appropriate subcomponents of constrainSimplePatternType's arguments. This turned out to be a problem overall. To improve the situation, we instead manually compare the appropriate subcomponents of constrainSimplePatternType - the changes can be seen as inlining pieces of code from TypeComparer. Also, this commit adds restoring the GADT constraint to TypeComparer - this turned out to be more important now that constrainSimplePatternType now does multiple isSubType checks. Fix failing test Add explanatory comments Adjust comments Insert conversions also on selections wrapped in type applications In i12708.scala, the problematic function was a selection `qual.m[tvs]` that was already applied to type variables. In that case we need to backtrack, forget the type variables and try to insert a conversion or extension method on `qual`. Fixes scala#12708 Fix tests Add 3.0.1-RC1 blog article Remove experimental macro feature mention Co-authored-by: Nicolas Stucki <nicolas.stucki@gmail.com> macros.md: fix parentheses/braces in inlined Macros.assert() example The parantheses and braces where swapped. Also align the style with the follow up example to make it easier to spot the differences. Add regression test for scala#12723 Deleted unnecessary explanations As a result of replacing 'Dotty' with 'Scala 3' in the following commit, the explanation 'which is scheduled to become Scala 3' is no longer needed. scala@ccb76da Improve display of references for ambiguous implicit errors If the first attempt yields equal strings, print the full path to the implicits instead. Fixes scala#12591 Reclassify test Add check file Improvements to search UX Pressing "S" while not focused on an input/textarea will now open the search. Pressing "F" while not focused on an input/textarea will focus the member filter input. Pressing Escape while focused on the search/filter will clear the input and close/unfocus it Search result links are now treated as block elements so you can click anywhere inside the rectangle rather than specifically on the result text. also allow '/' to trigger the search Add browsable versions extension for scaladoc Change local storage to session storage, fix bugs with undefined js variable Chagne try catching of undefined scalajs property to js.typeOf Add test for infix operator usage from Scala 2 libraries Fixes scala#7304 Preserve tvar instantiation in implicit error messages make default colors less harsh, add dark/light theme swapper make dottydoc anchors transparent instead of white when not hovered infer dark theme preference from user's system add some 'missing' semicolons to some JS set min-width on the theme switch so it doesn't get squished Make dark-mode footer actually dark. Improve contrast/accessibility. Used the "poor man's dark mode" approach to deal with the images in the footer in dark mode, i.e. invert and hue rotate (without the hue rotate, the red scala3doc logo becomes blue) Also fix an issue where if you click "back to top" and then refresh the page, the "container" element gains the "expand" class, which causes the main signature (i.e. `class List[A] ...`) to become 'inline' instead of a block, and it also triggers the 6.5em left-margin that's intended for method signatures when they get expanded. Also consolidated the `footer` styles, since they were spread all over. Check CLA for user who opened pull request, not workflow initiator Add regression test Closes scala#12754 Don't allow wildcard types in constraints Fixes scala#12677 Only drop wildcards for necesessary comparisons Wildcards in constraints essentially mean that the constrained variable is a sub/supertype of _every_ instance of the bound. For useNecesaryEither, this is wrong. Approximate the bound not to use wildcards instead. drop all kinds of WildcardType under useNecessaryEither better bounds for wildcards in resultTypeApprox Avoid wildcard types in constraints Ther semantics is fishy. Instead, use avoidance in necessarySubType and in TypeVarsMiss context mode, and use fresh type variables elsehwere. Cleanup addOneBound logic Verify that constraints do not contain wildcard types Cleanups Streamline approximateWildcards logic Test assertion Don't let wildcards enter constraints when adding type lambdas Fix typo Turn off "no wildcards in constraint" check Address review comments macros.md: Fix htmlized variant of first source code example Confusingly, the last part of the first source code example would appear as def showExpr(expr: Expr[Boolean])(using Quotes): Expr[String] = '{ " " } in the htmlized version of the markdown documentation source. Which reads as follows: def showExpr(expr: Expr[Boolean])(using Quotes): Expr[String] = '{ "<some source code>" } // Better implementation later in this document This seems to be cause by the usage of angle brackets. Also note that the code comment does not appear in the htmlized output. Fix typo in 3.0.1-RC1 release notes Fix scala#9588: Enable Scala.js test ObjectTest. The issue preventing that test from compiling in the past was fixed, although we don't know when or by what. Scala.js: Mangle test names when generating JUnit bootstrappers. This fixes the last remaining JUnit-related tests, which we therefore enable. Copy old site versions for Scala 3 instead of Dotty 0.X Support both 0.X and 3.X lines For extension label, avoid scala style Fix escaping illegal tokens in snippet renderer in scaladoc Use sbt makeScalaInstance Add definition Add doc for env Fix compilation Refactor trace Support Value.source for friendly error message Disallow non-hot arguments to non-constructors Handle access of parameters in constructors Handle exception No exceptions Fix Env.isHot Add missing checks for arguments Handle new expression with only cold args Fix soundness test Reorganize tests Add test Add test for secondary constructor Refactor promotion Add test More logging Fix bug with promotion of this The previous code does not work, because private fields are not included. Refactor code: introduce ArgInfo instead of Value.source Handle O.this outside of it's definition Fix owner of values with static flag With the flag <static>, the enclosing class will be skipped. See the documentation `SymDenotation.enclosingClass` Add documentation about caching Add test Add another neg test for promotion of ThisRef Handle outers of trait as the same in concrete semantics For traits, its outers will be proxy methods of the class that extends the trait. As the prefix is stable and is a valid value before any super constructor calls. Therefore, we may think the outers for traits are immediately set following the class parameters. Also, when trying promotion of warm values, we never try warm values whose fields are not fully filled -- which corresponds to promote ThisRef with non-initialized fields, and errors will be reported when the class is checked separately. Address review (thanks @michelou) Rename isFullyInitialized to isFullyFilled to avoid confusion Restrict function arguments to be hot This aligns with the design of restricting method arguments to be hot. We only allow non-hot to constructors. Add documentation Address review Simplify logic for checking local classes inside secondary constructors Address review: simplify logic Abstract parameters of non-local constructor parameters as cold This is a simple fix which should have no impact on expressiveness and usability, as local classes inside secondary constructors are extremely rare. Update docs Fix scala#11973: Add regression test community build: Update munit Add Libretto to community build. re-organize tests
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
gives
The text was updated successfully, but these errors were encountered: