On 16/07/13 23:52, maurizio cimadamore wrote:
FYI,
those changes seem to trigger a spurious failure when running
langtools regression tests on a Windows machine [1]. We have been
working to isolate the problem and put together a fix which I am
testing as we speak. If the experiment is successful we can decide
what's the best course of action (i.e. direct push to master) in order
to minimize the number of teams that will be affected by this issue.
Quick update - the experimental nightly build I ran over the
lambda-repository with the fix for 8013638 gave good results on Windows
machine; it appears the issue is correctly addressed by the fix. Please
advice as to what I should do next to have this fix into master.
Thanks
Maurizio
Maurizio
[1] - https://jbs.oracle.com/bugs/browse/JDK-8013638
On 16-Jul-13 8:42 PM, Alejandro E Murillo wrote:
Later today I plan to push the hs25-b41 snapshot into jdk8 master [1]
<http://hg.openjdk.java.net/jdk8/jdk8/>,
along with the fix for 8016281
<http://bugs.sun.com/view_bug.do?bug_id=8016281>. This push includes
lambda-related changes that will
create lambda-related incompatibilities with class files compiled
with earlier
JDK 8 versions. See [2]
<http://mail.openjdk.java.net/pipermail/lambda-dev/2013-July/010449.html>
for more details.
Once the changes have been integrated, the hotspot, langtools and jdk
repos must match - i.e., either all must have these changes or not
have any of them.
A JDK with mismatched repos may build, but some lambda-related
functionality
might not work properly (again, see [2]
<http://mail.openjdk.java.net/pipermail/lambda-dev/2013-July/010449.html>
for details of the symptoms).
After we integrate, If you do pull, make sure to pull all: hotspot,
langtools and jdk,
and to build all
[1] http://hg.openjdk.java.net/jdk8/jdk8/
[2]
http://mail.openjdk.java.net/pipermail/lambda-dev/2013-July/010449.html
Thanks
