RFR(XS): 8001424: G1: Rename certain G1-specific flags

Bengt Rutisson bengt.rutisson at oracle.com
Fri Dec 21 20:48:03 PST 2012


Hi John,

On 12/21/12 7:46 PM, John Cuthbertson wrote:
> Hi Bengt,
>
> Great. The old names have been removed (they won't be accepted in 
> either hs24 or hs25). This makes the changes a lot smaller. Your 
> statement below was part of teh reason why I wanted them to stay 
> experimental.

Great! Ship it!

Bengt

>
> JohnC
>
>
> On 12/21/2012 5:47 AM, Bengt Rutisson wrote:
>>
>> Hi John,
>>
>> On 12/20/12 7:04 PM, John Cuthbertson wrote:
>>> Hi Ramki, Bengt,
>>>
>>> Thanks for the reviews. I kept the old names because the perf team 
>>> would like these backported to hs24 (7u12) and the old names have 
>>> been published in several presentation decks - including one from 
>>> Monica and Charlie at JavaOne. Does it still make sense to just 
>>> accept the new names? The change would be much smaller if so.
>>
>> Personally I would still not think we should keep the old names. 
>> After all they are experimental flags.
>>
>> If we should keep the old names I think it would be enough to that in 
>> hs24/7u12. I would prefer that we don't have the old names in JDK8.
>>
>> Thanks,
>> Bengt
>>
>>>
>>> Thanks,
>>>
>>> JohnC
>>>
>>> On 12/20/2012 1:19 AM, Srinivas Ramakrishna wrote:
>>>> New names look good. I agree with Bengt that for renames of exptal 
>>>> flags in a major release bothering supporting old names is not 
>>>> worthwhile; best to
>>>> make a clean break with the old names.
>>>>
>>>> reviewed
>>>> -- ramki
>>>>
>>>> On Wed, Dec 19, 2012 at 10:30 PM, Bengt Rutisson 
>>>> <bengt.rutisson at oracle.com <mailto:bengt.rutisson at oracle.com>> wrote:
>>>>
>>>>
>>>>     Hi again John,
>>>>
>>>>     I realized that I was a bit too fast with my comment about
>>>>     using ObsoleteFlag. Your code is aliasing the old names for the
>>>>     new ones which is something the obsolete flag management does
>>>>     not do.
>>>>
>>>>     But on the other hand, do we really want to do this? These are
>>>>     all experimental flags and we are pushing this change to a
>>>>     major release, JDK8. Personally I don't think it is worth
>>>>     supporting the old names.
>>>>
>>>>     Bengt
>>>>
>>>>
>>>>     On 12/20/12 5:45 AM, Bengt Rutisson wrote:
>>>>>
>>>>>     Hi John,
>>>>>
>>>>>     This looks good. But I think that instead of your change in
>>>>>     arguments.cpp you could make use of the obsolete_jvm_flags
>>>>>     list that exist in the same file. I think that is intended for
>>>>>     exactly this purpose. Accepting a removed flag name for a
>>>>>     little while. The nice thing about it it that you specify how
>>>>>     long you will accept the old name.
>>>>>
>>>>>     static ObsoleteFlag obsolete_jvm_flags[] = {
>>>>>       { "UseTrainGC", JDK_Version::jdk(5), JDK_Version::jdk(7) },
>>>>>
>>>>>     If you use this you also have to remove the old flag names
>>>>>     from globals.hpp.
>>>>>
>>>>>     Bengt
>>>>>
>>>>>
>>>>>     On 12/20/12 1:56 AM, John Cuthbertson wrote:
>>>>>>     Hi Everyone,
>>>>>>
>>>>>>     Some flag name changes suggested by the JVM performance team
>>>>>>     based upon feedback they have received. The webrev can found
>>>>>>     at: http://cr.openjdk.java.net/~johnc/8001424/webrev.0/
>>>>>>     <http://cr.openjdk.java.net/%7Ejohnc/8001424/webrev.0/>
>>>>>>
>>>>>>     Basically the changes are those listed in the webrev:
>>>>>>
>>>>>>     G1DefaultMinNewGenPercent is being replaced by G1NewSizePercent
>>>>>>     G1DefaultMaxNewGenPercent is being replaced by
>>>>>>     G1MaxNewSizePercent
>>>>>>     G1OldCSetRegionLiveThresholdPercent is being replaced by
>>>>>>     G1MixedGCLiveThresholdPercent
>>>>>>
>>>>>>     Thanks,
>>>>>>
>>>>>>     JohnC
>>>>>
>>>>
>>>>
>>>
>>
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.openjdk.java.net/pipermail/hotspot-gc-dev/attachments/20121222/6fba919c/attachment.html 


More information about the hotspot-gc-dev mailing list