Hello Sergey,
Just concerning: does the fix also cover the situation when, e.g., we
have two annotated getters (regular + indexed) or setters (should their
priority be specified somehow in such a case)? A separate issue exists
for this case (JDK-8132732), so just wondering if it shouldn't be
trailed to the fix (as it seems at a 1st glance that the matter is very
similar).
Thanks,
Alexander
P.S. at least the test fixes look good for me (not a reviewer)
On 5/23/2016 12:31 PM, Sergey Bylokhov wrote:
Hello.
Please review the fix for jdk9.
The bug describe the situation when for the same property both
read/write methods are annotated. Our code assumes that the annotation
exists only for one of these methods, and we produce the unspecified
mix of both annotations(some of the data from reader and other from
writer).
As a fix I changed the order of initialization, first we will try the
reader and if necessary the writer (before it was expected that the
data from the read will overwrite the data from write, but it fails
because of optimization). The specification is clarified and ccc will
be filed after the technical review. The tests are updated also.
Bug: https://bugs.openjdk.java.net/browse/JDK-8132973
Webrev can be found at:
http://cr.openjdk.java.net/~serb/8132973/webrev.00
