RFR: JDK-8165314: Javac server process left running if build fails on Windows

Bradford Wetmore bradford.wetmore at oracle.com
Fri Sep 2 20:18:51 UTC 2016


P.S.  It fixed the problem on my Windows build.  Thanks!

Brad


On 9/2/2016 1:13 PM, Bradford Wetmore wrote:
>> Looks good to me.
>
> Ditto.  Always learn something new about gmake reading the build code.
>
> Thanks for adding the ReportBuildTimes.  That's a nice thing to have.
>
> Brad
>
>
>
> On 9/2/2016 8:02 AM, Tim Bell wrote:
>> On 09/02/16 05:16, Erik:
>>
>>
>>> The symptom of this problem is that if the build fails, quickly fixing
>>> the issue and rerunning the build will just fail again very early with
>>> an error message like this:
>>>
>>>    make[1]: *** [/cygdrive/d/java/ws/dev/make/Init.gmk:290: main]
>>> Error 1
>>>    make[1]: Leaving directory '/cygdrive/d/java/ws/dev'
>>>    make: *** [/cygdrive/d/java/ws/dev/make/Init.gmk:185: java.base-only]
>>> Error 2
>>>
>>> The fix is rather simple. There is a cleanup call to stop the sjavac
>>> server background process in Init.gmk, but it's only run on successful
>>> builds and not in the "on-failure" routine. Adding the call to the
>>> on-failure routine fixes the issue. While at it I also added the build
>>> time report since I have missed not having it there on failures many
>>> times.
>>>
>>> Bug: https://bugs.openjdk.java.net/browse/JDK-8165314
>>>
>>> Webrev: http://cr.openjdk.java.net/~erikj/8165314/webrev.01/
>>
>> Looks good to me.
>>
>> /Tim
>>
>>



More information about the build-dev mailing list