[security-dev 00472]: Re: hg: jdk7/tl/jdk: 6496274: jar seems to use more CPU than it should

[Xueming Shen]

Hi Mark,

No, you did not miss anything.  The "original patch proposal " had not 
been sent to
any of  the community mailing lists, the "reviewed-by" is for the final 
code review,
which was not sent to the mailing list as well, I don't think we send 
each/every code
review request out, especially when the change itself is not that "big".

Maybe this one is worth a blog after it finally makes into the 
binaries/snapshot. And
I will try to backport it into previous releases after baked in the 7 
for a while.

Thanks,
Sherman

Mark Wielaard wrote:
> Hi,
>
> On Thu, 2008-12-18 at 06:57 +0000, xueming.shen at sun.com wrote:
>   
>> Changeset: 57dc40ece164
>> Author:    sherman
>> Date:      2008-12-17 22:50 -0800
>> URL:       http://hg.openjdk.java.net/jdk7/tl/jdk/rev/57dc40ece164
>>
>> 6496274: jar seems to use more CPU than it should
>> Summary: boost jar creating performance especially for the large jar file
>> Reviewed-by: martin
>>
>> ! src/share/classes/sun/tools/jar/Main.java
>>     
>
> Heay, this is a pretty cool patch.
> In IcedTea we explicitly added an configure option to work around the
> slowness (especially on platforms that only have the zero interpreter
> available):
>   --with-alt-jar  specify the location of an alternate jar binary to
>                   use for building
> So you can build against fastjar (a plain C GPLed jar implementation) to
> work around the same issue. Fixing the java based jar implementation
> directly is way cooler though (we really should backport this after some
> testing).
>
> Is there a way to raise awareness of stuff like this that people are
> working on? If I hadn't seen the mercurial commit email I might have
> completely missed it. I assume there is some way to get alerts from the
> bug tracker where this was first reported. Could should bug reports be
> send to the mailinglist somehow? And I see this actually has a
> "Reviewed-by" tag, but I never saw the original patch proposal, nor the
> review on any mailinglist (I might have missed it though, I am only
> subscribed to half of the mailinglists). If there was a code patch
> review mailinglist for patches like these I would certainly subscribe.
>
> Thanks,
>
> Mark
>
>