Fix #10178: Put the wildcard demon back in the bottle #10195
Merged
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
|
Reminiscent of scala/scala#6218 which in 2017 used the encoding with empty tree, but that was deemed too big a change, so in 2018 it just got some encapsulation at ImportSelectorApi and ImportSelector. I mention it only because of the psychological scarring. |
This was a very interesting bug. Why did `any2stringadd` appear out of the blue?
The first observation was that it was resolved by an import of an identifier with
`_` as its name. Why? Because we have a standard predefined import
```scala
import Predef.{any2stringadd => _, _}
```
This is represented as a "rename" from `anyToStringAdd` to `_`, assuming that
no real identifier could be `_`. But that assumption turned out to be wrong.
How did we end up with an identifier `_` that needed to be resolved? This was a
second bug in Desugar, method `makeIdPat`. Here we needed to create binds for parts
of a pattern. If the pattern was already a Bind, we used that one instead. But the
`Bind` in question was anonymous, using `_` as the pattern variable. So we did not
replace that by a fresh identifier, but used `_` as the identifier instead. However,
`_` is treated specially as a binder; it does not generate a symbol in the enclosing
scope. So resolving the `_` identifier did not see the enclosing bind and searched
further out instead, until it found the "renamed" import. Beautiful! It shows that
treating wildcards as some sort of identifiers is fraught with surprises.
|
A more robust representation could be to avoid using |
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Outdated suggestions cannot be applied.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.
This was a very interesting bug. Why did
any2stringaddappear out of the blue?The first observation was that it was resolved by an import of an identifier with
_as its name. Why? Because we have a standard predefined importThis is represented as a "rename" from
anyToStringAddto_, assuming thatno real identifier could be
_. But that assumption turned out to be wrong.How did we end up with an identifier
_that needed to be resolved? This was asecond bug in Desugar, method
makeIdPat. Here we needed to create binds for partsof a pattern. If the pattern was already a Bind, we used that one instead. But the
Bindin question was anonymous, using_as the pattern variable. So we did notreplace that by a fresh identifier, but used
_as the identifier instead. However,_is treated specially as a binder; it does not generate a symbol in the enclosingscope. So resolving the
_identifier did not see the enclosing bind and searchedfurther out instead, until it found the "renamed" import. Beautiful! It shows that
treating wildcards as some sort of identifiers is fraught with surprises.