On 08/21/2017 08:06 PM, Paul Sandoz wrote: >> On 21 Aug 2017, at 07:48, Claes Redestad <claes.redes...@oracle.com> wrote: >> a trivial test[1] invoking the StringConcatFactory.makeConcatWithConstants >> fails >> when providing an Integer as a constant, which appears to be due to failure >> to >> coerce boxed types to the corresponding primitive types when looking up >> various >> methods in StringConcatHelper: >> >> Webrev: http://cr.openjdk.java.net/~redestad/8186500/jdk.00/ >> Bug: https://bugs.openjdk.java.net/browse/JDK-8186500 >> >> Simply using Wrapper.asPrimitiveType coerces boxed types to their primitive >> counterpart, and is a (semantical) no-op for other types, e.g., String. > > Looks good. Perhaps a token test would be useful (maybe hard to test all code > paths here without some combinator test).
Please add tests. We have to at least assert that passing "null" works fine there -- I am not sure it does right now. Since we are talking non-String constants, passing java/lang/Object is a good test too. -Aleksey