You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
...and the updates to the samples should be easily automated.
Ideally, we'd continue recognizing the old locations for a bit after we add the new ones. However, might not be entirely trivial for @NullMarked, which the Checker Framework wants on its classpath so that it can use a Class object for it. (I think I once got the impression we couldn't trivially switch from Class to String, but I'm not sure.) To keep it on the classpath, maybe we should arrange for JSpecify 0.3 to contain a deprecated version of the old one (yuck)? Or maybe we can just put the old @NullMarked in the checker jar itself: We already include the checker on the compile classpath (i.e., not just the processorpath) because that's how the Checker Framework likes to have at least its annotations. (Googlers can find some discussion in internal bug 187113128.) But wait, did I ever try using getDeclAnnotations (plural) instead of `getDeclAnnotation (singular)? If that works, that would be simpler.
The text was updated successfully, but these errors were encountered:
It might be worth thinking about whether the eisop enhancements to "defaulting" annotations could help us here if we did jspecify/checker-framework#29. It's been a while since I thought about what reason(s) we have for our custom defaulting code.
That is, if we change the package name in jspecify/jspecify#260, then we need to update the annotation class names that we hardcode in the checker.
The main changes should be simple, perhaps two lines...
https://github.com/jspecify/nullness-checker-for-checker-framework/blob/52cef8b0c169e613d7a42d85b4b80d94dd533088/src/main/java/com/google/jspecify/nullness/NullSpecAnnotatedTypeFactory.java#L116
https://github.com/jspecify/nullness-checker-for-checker-framework/blob/52cef8b0c169e613d7a42d85b4b80d94dd533088/src/main/java/com/google/jspecify/nullness/NullSpecAnnotatedTypeFactory.java#L1947
...and the updates to the samples should be easily automated.
Ideally, we'd continue recognizing the old locations for a bit after we add the new ones. However, might not be entirely trivial for
@NullMarked
, which the Checker Framework wants on its classpath so that it can use aClass
object for it. (I think I once got the impression we couldn't trivially switch fromClass
toString
, but I'm not sure.) To keep it on the classpath, maybe we should arrange for JSpecify 0.3 to contain a deprecated version of the old one (yuck)? Or maybe we can just put the old@NullMarked
in the checker jar itself: We already include the checker on the compile classpath (i.e., not just the processorpath) because that's how the Checker Framework likes to have at least its annotations. (Googlers can find some discussion in internal bug 187113128.) But wait, did I ever try usinggetDeclAnnotations
(plural) instead of `getDeclAnnotation (singular)? If that works, that would be simpler.The text was updated successfully, but these errors were encountered: