RFR: JDK-8193061 Add run-test-prebuilt functionality

Magnus Ihse Bursie magnus.ihse.bursie at oracle.com
Tue Dec 5 20:14:13 UTC 2017

On 2017-12-05 21:05, Erik Joelsson wrote:
> Looks pretty good, only some minor notes:
> * RunTestsPrebuilt.gmk:72: indentation
> * 95: You could use the sequence macro to generate a large set of 
> increasing numbers here.
Unfortunately, it's not available since I can't read MakeBase.gmk yet 
(it requires a SPEC file). :-( I could have copied it, but that didn't 
seem worth the effort compared to just listing the integers.
*sound of coin landing in machine*
Oh, but wait, that's not correct anymore. I *do* have MakeBase.gmk 
included here. That also means I should replace "rm -f" with "$(RM)". :-)

> * 182: "... all others use uname -m"

New webrev with above fixes:

> /Erik
> On 2017-12-05 05:02, Magnus Ihse Bursie wrote:
>> This addition makes it possible to run tests using the normal 
>> RunTests.gmk framework, even if configure has not been run. The 
>> typical use case is for automated testing, where a test machine can 
>> download a prebuilt JDK image and test image, and the source tree, 
>> and then start executing tests as specified by RunTests.gmk in the 
>> source tree, but excersicing the prebuilt JDK and test image.
>> This mode of operation has historically been provided by 
>> test/Makefile, but that system is messy, not maintained and parallel 
>> to the run-test implementation.
>> With this patch, a new "global" target "run-test-prebuilt" is 
>> provided. To use it, you must also specify a number of variables, 
>> either on the command line or in the environment. These are the 
>> variables that run-test-prebuilt cannot guess. With these values as 
>> base, run-test-prebuilt creates a "fake" spec file, with some generic 
>> values that are likely to work in many cases, and/or values that can 
>> be detected in runtime.
>> Bug: https://bugs.openjdk.java.net/browse/JDK-8193061
>> WebRev: 
>> http://cr.openjdk.java.net/~ihse/JDK-8193061-add-run-test-prebuilt/webrev.01
>> /Magnus

More information about the build-dev mailing list