Allow empty if statements if they're captured by a function #720
Conversation
Johnnei
changed the title
Codecov ReportBase: 87.19% // Head: 87.20% // Increases project coverage by
Additional details and impacted files@@ Coverage Diff @@
## master #720 +/- ##
=======================================
Coverage 87.19% 87.20%
=======================================
Files 141 141
Lines 1523 1524 +1
Branches 221 226 +5
=======================================
+ Hits 1328 1329 +1
Misses 195 195
Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here. ☔ View full report at Codecov. |
| |import scala.concurrent.ExecutionContext.Implicits.global | ||
| |object Test { | ||
| | Future("test").map(value => { | ||
| | if (value.contains("foo")) { |
There was a problem hiding this comment.
Don't have my IDE here, but I think the original authors' intent of scapegoat was to write:
Future("test").toOption match {
case Some("foo") => ()
case _ => throw new IllegalStateException("Bar!")
}
Does it make sense?
There was a problem hiding this comment.
Not entirely sure I follow, I based the sample on a further simplified version of the one in the issue:
def checkFoo(things: Future[Seq[String]]): Future[Unit] = {
things.map(ts => if (ts.contains("foo")) () else throw new Exception("No foo!"))
}The comment with the pattern matching is a valid workaround for this warning (although I'd prefer the transformWith solution which doesn't use flow control mechanisms), but does it really solve the problem scapegoat highlights? Instead of an empty if branch you now have an empty case branch. While empty case branches are maybe a bit more common to see, I wouldn't really see it as much different.
So I tend to agree with the the original issue and consider it a false-positive for cases where the empty branch (if or case in that regard) are being captured by the return type.
There was a problem hiding this comment.
does it really solve the problem scapegoat highlights
I think the intended thinking was that empty branches in ifs are not OK, but empty case is fine.
Nevertheless, I don't feel strongly about that.
There was a problem hiding this comment.
if @sksamuel speaks up soon, I'd let him decide.
Otherwise I am fine with merging that.
There was a problem hiding this comment.
I have no strong opinion either so we can merge.
|
Thanks for your contribution, @Johnnei! |
|
2.1.1 has been released. |
Fixes #229
While I don't fully agree with the presented sample, I do agree with the spirit. So a more generic solution is to ignore the If blocks entirely if they're the body of a Function (ie, they affect the return types).