Topic/modifier lockdown #3
Closed
Conversation
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
Notable omissions are ValDef and DefDef, which are
subclassed by the empty variety.
Final classes are easier for the reader to reason about
and enable additional static analysis by the compiler.
As an example of the latter,
scala> trait T
defined trait T
scala> Tuple1(0) match { case _: T => true; case _ => false }
res0: Boolean = false
scala> final case class T1[A](a: A)
defined class T1
scala> T1(0) match { case _: T => true; case _ => false }
<console>:11: error: scrutinee is incompatible with pattern type;
found : T
required: T1[Int]
T1(0) match { case _: T => true; case _ => false }
This commit exposes a number of inexhaustive matches:
[warn] /Users/jason/code/dotty/src/dotty/tools/dotc/core/Trees.scala:716: match may not be exhaustive.
[warn] It would fail on the following inputs: Function(_, _), GenericApply(), ModuleDef(_, _, _), TypedSplice()
[warn] def transform(tree: Tree[T], c: C): Tree[T] = tree match {
[warn] ^
[warn] /Users/jason/code/dotty/src/dotty/tools/dotc/core/Trees.scala:870: match may not be exhaustive.
[warn] It would fail on the following inputs: Function(_, _), GenericApply(), ModuleDef(_, _, _), TypedSplice()
[warn] def foldOver(x: T, tree: Tree[U]): T = tree match {
[warn] ^
[warn] two warnings found
[info] Compiling 52 Scala sources to /Users/jason/code/dotty/target/scala-2.10/classes...
[warn] /Users/jason/code/dotty/src/dotty/tools/dotc/core/PluggableTransformers.scala:80: match may not be exhaustive.
[warn] It would fail on the following inputs: Alternative(_), AndTypeTree(_, _), Annotated(_, _), AppliedTypeTree(_, _), Assign(_, _), Bind(_, _), Block(_, _), CaseDef(_, _, _), ClassDef(_, _, _, _), DefDef(_, _, _, _, _, _), EmptyTree(), EmptyValDef(), Function(_, _), GenericApply(), If(_, _, _), Import(_, _), Literal(_), Match(_, _), ModuleDef(_, _, _), NamedArg(_, _), New(_), OrTypeTree(_, _), PackageDef(_, _), Pair(_, _), RefineTypeTree(_, _), Return(_, _), SelectFromTypeTree(_, _), SeqLiteral(_, _), Shared(_), SingletonTypeTree(_), Super(_, _), Template(_, _, _), This(_), Throw(_), Tree(), Try(_, _, _), TypeBoundsTree(_, _), TypeDef(_, _, _), TypeTree(_), Typed(_, _), TypedSplice(), UnApply(_, _)
[warn] protected def postProcess(tree: Tree[T], old: Tree[T], c: Context, plugins: Plugins): Tree[T] = tree match {
[warn] ^
[warn] /Users/jason/code/dotty/src/dotty/tools/dotc/core/Trees.scala:716: match may not be exhaustive.
[warn] It would fail on the following inputs: Function(_, _), GenericApply(), ModuleDef(_, _, _), TypedSplice()
[warn] def transform(tree: Tree[T], c: C): Tree[T] = tree match {
[warn] ^
[warn] /Users/jason/code/dotty/src/dotty/tools/dotc/core/Trees.scala:870: match may not be exhaustive.
[warn] It would fail on the following inputs: Function(_, _), GenericApply(), ModuleDef(_, _, _), TypedSplice()
[warn] def foldOver(x: T, tree: Tree[U]): T = tree match {
[warn] ^
[warn] /Users/jason/code/dotty/src/dotty/tools/dotc/core/TypedTrees.scala:359: match may not be exhaustive.
[warn] It would fail on the following inputs: EmptyTree(), GenericApply(), Try(_, _, _)
[warn] def checkType(tree: tpd.Tree)(implicit ctx: Context): Unit = tree match {
[warn]
Perhaps controversially seals the hierarchy and as a consequence moves two types from the unpickler into Types.scala.
- Introduce a marker trait for Tree[Nothing] nodes, and use this in TreeAccumulator. - seal GenericApply - Make the match demo plugin transformer @unchecked - Add tree checking for Try - Match on EmptyTree() rather than tpd.EmptyTree
- info was calling itself, rather than the field _info - tryComplete must interrogate and set the Locked bit through the raw _flags.
|
I guess this won't merge cleanly any more. Also my checking for |
|
@DarkDimius I'll close this one to as it will be well and truly out of date. But maybe someone will be inspired to hold the torch of privacy/finality/sealedness in dotty for me? You will appreciate it as the code base grows larger. |
|
Oh, and the inliner likes finality, too! |
DarkDimius
referenced
this pull request
in dotty-linker/dotty
May 9, 2016
odersky
added a commit
that referenced
this pull request
Jul 18, 2019
KacperFKorban
referenced
this pull request
in KacperFKorban/dotty
Mar 25, 2021
…debug Add returning real line of error in source file of snippet for snippet scaladoc compiler
odersky
added a commit
that referenced
this pull request
Dec 23, 2023
`given ... with` or `given ... = new { ... }` kinds of definitions now follow
the old rules. This allows recursive `given...with` definitions as they are
found in protoQuill.
We still have the old check in a later phase against directly recursive methods.
Of the three loops in the original i15474 we now detect #2 and #3 with new new
restrictions. #1 slips through since it is a loop involving a `given...with` instance
of `Conversion`, but is caught later with the recursive method check.
Previously tests #1 and #3 were detected with the recursive methods check and #2 slipped
through altogether.
The new rules are enough for defining simple givens with `=` without fear of looping.
Kordyjan
pushed a commit
that referenced
this pull request
Jan 18, 2024
`given ... with` or `given ... = new { ... }` kinds of definitions now follow
the old rules. This allows recursive `given...with` definitions as they are
found in protoQuill.
We still have the old check in a later phase against directly recursive methods.
Of the three loops in the original i15474 we now detect #2 and #3 with new new
restrictions. #1 slips through since it is a loop involving a `given...with` instance
of `Conversion`, but is caught later with the recursive method check.
Previously tests #1 and #3 were detected with the recursive methods check and #2 slipped
through altogether.
The new rules are enough for defining simple givens with `=` without fear of looping.
WojciechMazur
added a commit
to WojciechMazur/dotty
that referenced
this pull request
Mar 19, 2025
Backport "Add better error reporting for inlined non-immutable paths" to LTS
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
I propose that we make a more concerted effort to keep things final/sealed; this PR does this across the board. It's easy enough to open things up later on, until then we can profit from better pattern match analysis, notification of non-sensical equality tests, and, last but not least, less to think about when reasoning about the code.
As a bonus, the exercise uncovered two compiler bugs: SI-7171 and SI-7172.
The parts that warrant closer review:
UntypedTreeMarkerin 4598e8c.Try, and ignorance ofUntypedTrees inTreeTransformerandTreeAccumulator.lastDenotand_typeParams.Type, and therefore move two type nodes from the unpickler to Types.scalaReview by @odersky, if you like we can chat about this on Tuesday.