RFR 8010280: jvm.cfg needs updating for non-server builds
mike.duigou at oracle.com
Tue Apr 16 17:25:44 UTC 2013
On Apr 15 2013, at 15:42 , David Holmes wrote:
> On 16/04/2013 2:25 AM, Mike Duigou wrote:
>> What's the difference between removing an entry completely and retaining it with "ERROR"?
> Just the nature of the error message:
> > java -green
> Error: green VM not supported
> > java -blue
> Unrecognized option: -blue
> Error: Could not create the Java Virtual Machine.
> Error: A fatal exception has occurred. Program will exit.
> I wasn't touching any of the legacy stuff - though if this needs to go to CCC I would suggest removing all the legacy entries.
>> Additionally I don't like that aliases have differing definitions and some confusing ones like "-server ALIASED_TO -client". Is this necessary or just historically convenient?
> I don't like aliases period! Historically (and this is very recent history) it was necessary to deal with the test suites being applied to a JDK with, eg, only client VM. Every test that specified -server would fail if the alias didn't exist (and as I stated we're moving away from that ie the tests don't set -client or -server but the complete test suite run does, and it knows what VM is under test.
> Personally I'd probably choose WARN for any VM not present.
> The problem is that the "right" thing depends on who is building what, and how they plan to use it. All I can do is define a not-unreasonable default policy. I also have a time constraint as I need to get this in before the 23rd to meet an internal deadline.
More information about the core-libs-dev