Heads up: Deprecating the CMS foreground mode

[Krystal Mok]

Hi guys,

I've actually seen quite a few cases where people grep'd globals.hpp, found
the flag CMSFullGCsBeforeCompaction but not knowing what it really is, and
suffer from old gen fragmentation problems when they set the flag to a
value greater than zero. It's more like a trap than a tool in these cases.
I'll be glad when this flag is gone ;-)

Thanks,
Kris


On Sat, Oct 26, 2013 at 1:57 AM, Kirk Pepperdine <kirk at kodewerk.com> wrote:

> Hi Bengt,
>
> Thank you very much for this heads up. All I can say is that I don't know
> anyone using the foreground collector, I don't know of anyone using the
> flags needed to turn on the foreground collector and I don't recall seeing
> any evidence of the foreground collector in any of the logs that I've
> looked at.
>
> So, unlike iCMS, I'm not about to add any resistance to the decision to
> deprecate.
>
> Regards,
> Kirk
>
> On 2013-10-24, at 2:48 PM, Bengt Rutisson <bengt.rutisson at oracle.com>
> wrote:
>
> >
> > Hi all,
> >
> > Just a heads up for anyone using the CMS foreground collector.
> >
> > I just sent out a review request to this list that proposes a change to
> print warning messages for the flags that enable the CMS foreground
> collector. The review request is titled "JDK-8027132: Print deprecation
> warning message for the flags controlling the CMS foreground collector" and
> it would be good if any discussion can be handled in that email thread.
> >
> > We don't know of anyone using the foreground collector, so I thought I'd
> send out an extra email out to draw some attention to the review request if
> there is anybody using it.
> >
> > Just to be clear. The change proposed now only prints warning messages
> for some flags. Everything will keep working as before. This is just to
> communicate that the foreground collector has been deprecated. Hopefully we
> will follow this up with a change to actually remove the support for the
> foreground collector for one of the future major releases.
> >
> > Thanks,
> > Bengt
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.openjdk.org/pipermail/hotspot-gc-dev/attachments/20131027/d98369e4/attachment.htm>