https://bz.apache.org/bugzilla/show_bug.cgi?id=65539
Stefan Bodewig <[email protected]> changed: What |Removed |Added ---------------------------------------------------------------------------- Target Milestone|--- |1.10.12 --- Comment #9 from Stefan Bodewig <[email protected]> --- Both of your points are correct. Actually there might be differences between setting the property to extJavac or the classname - and the former should work "better" even for 1.10.6 and earlier. For example Javac's isForkedJavac method returns true for "extJavac" but false for the fully qualified classname unless the fork attribute has been set to true explicitly (which looks like a bug as well). To me it looks as if your setup with setting build.compiler to the fully qualified classname has only worked by accident in earlier releases. 1.10.7 has reversed some decisions from "is this not JDK 1.1" to "is this JDK 1.2 or later". In 1.10.6 setting build.compiler to a fully qualified classname results in "this is not JDK 1.1" but also "this is not JDK 1.2 or later". It is not even JDK 1.1 by the existing logic and so in 1.10.6 it has taken the "not 1.1" branches that have now become "1.2 or later" branches and thus are no longer chosen. We'll fix things so that using the explicit classnames for the built-in compiler adapters work the same as if their respective compiler short names have been chosen, but in general using the full classname has never been necessary and didn't really do what the initial authors thought it would do. -- You are receiving this mail because: You are the assignee for the bug.
