Request for review and approval: JDK-8177770: Need more precise control on build system logging
Magnus Ihse Bursie
magnus.ihse.bursie at oracle.com
Thu Mar 30 07:23:16 UTC 2017
On 2017-03-30 09:13, David Holmes wrote:
> On 30/03/2017 5:10 PM, Magnus Ihse Bursie wrote:
>> On 2017-03-30 03:04, David Holmes wrote:
>>> Hi Alexey,
>>> Can't really review the details of this but wanted to ensure that this
>>> gets tested on all platforms before being pushed. Internally use JPRT
>>> and "-testset hotspot", but you should also reach out to the AIX-port
>>> folk to ensure it works as expected there too.
>> What are your worries here? There are no platform-dependent changes,
>> apart from the actual logging script which might not work properly if
>> your local copy of time/flock is non-compliant. (We try to test for
>> this, at least for "time".) Unless you're running with LOG=profile, this
>> patch affects nothing.
> Just want to make sure it doesn't cause any unexpected failures on
> less mainstream OS.
>> If the profiling does not work properly on e.g. AIX, then some of the
>> AIX folks would need to supply an additional patch to update
> Sure. But it would be nice to not break things first. Some
> contributors continually pull OpenJDK changes and run them through
> their own build and test systems.
Actually, I seriously doubt many people outside me and Erik have used
the LOG=trace (now LOG=profile) option. It is highly intrusive, and not
something that can be enabled on normal build systems.
>>> On 30/03/2017 2:21 AM, alexey semenyuk wrote:
>>>> Please review and approve enhancement for JDK10.
>>>> Bug: https://bugs.openjdk.java.net/browse/JDK-8177770
>>>> Webrev: http://cr.openjdk.java.net/~asemenyuk/8177770/webrev.00/
More information about the build-dev