JEP 173: Remove Rarely-Used Combinations of Garbage Collectors

Bengt Rutisson bengt.rutisson at oracle.com
Thu Dec 6 13:31:00 PST 2012


On 12/6/12 9:52 PM, Srinivas Ramakrishna wrote:
>
>
> On Thu, Dec 6, 2012 at 5:11 AM, Bengt Rutisson 
> <bengt.rutisson at oracle.com <mailto:bengt.rutisson at oracle.com>> wrote:
>
>
>     I've changed the title of the JEP now. I'm not sure you think it
>     is better, but it is at least what I had promised Jon. The new
>     title is:
>
>     "Remove GC Combinations that are not Used Enough to Warrant Future
>     Support"
>
>
> Or perhaps just simply "Retire some GC combinations", and say what you 
> are saying above but not in the title.

Thanks, Ramki! It turns out that I hadn't forgotten to update the title. 
Mark had just changed it when he published the JEP since he thought the 
title above was too long. So I was just sitting here trying to come up 
with a shorter and better title.

I just updated the JEP title with your suggestion. :)

>
>     I am thinking there may be "emdebbed" jvm's running in
>     core/power-starved environments
>>     that might still find it useful. I agree that for desktop and
>>     server jvm's iCMS has probably outlived its useful life.
>
>     I thought "embedded" were building with a very limited set of GC
>     support and explicitly excluding all GCs except the serial collector.
>
>
> That's true of the "embedded" offering (thanks for the reminder). I 
> was thinking more generically of the Jav SE offering being
> used in what might be considered "embedded" settings (such as routers, 
> network devices, switches, printers and so forth).
> Not sure if there is such a market, or I am thinking of a phantom 
> (insert appropriate Shakespeare quote here).
>> I agree though that from the standpoint of server environments, less 
>> is probably better, given the complexity of maintaining
>> the various combinations.
>
>     Yes, and there is also the issue of being able to test and verify
>     the quality of the combinations. At the moment we have about 9
>     combinations but we only have resources to continuously test  3-4
>     of them. I feel very uncomfortable recommending customers to run
>     untested GC combinations in production.
>
>
> Completely agree! I think this is a welcome step to keep the platform 
> evolving with the times and shedding vestigial organs, as it were.

Thanks, I genuinely believe it is a step in the right direction.

Bengt

> -- ramki
>
>     Thanks for the feedback!
>     Bengt
>
>
>>
>>     -- ramki
>>
>>     On Wed, Dec 5, 2012 at 2:10 PM, <mark.reinhold at oracle.com
>>     <mailto:mark.reinhold at oracle.com>> wrote:
>>
>>         Posted: http://openjdk.java.net/jeps/173
>>
>>         - Mark
>>
>>
>
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.openjdk.java.net/pipermail/hotspot-gc-dev/attachments/20121206/909c2389/attachment.html 


More information about the hotspot-gc-dev mailing list