Mailing-List: contact dev-help@harmony.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@harmony.apache.org
Received-SPF: pass (athena.apache.org: domain of xiaofeng.li@gmail.com
 designates 209.85.142.187 as permitted sender)
DomainKey-Signature: a=rsa-sha1; c=nofws;
        d=gmail.com; s=beta;
        h=message-id:date:from:to:subject:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references;
        b=OJ97JJ02VxKVss9KiXjPbws/rXGJw5k6VgvWJ1ZjWnVQbrlutym+EOTpRykYvsCo0v+6Pin+Ch1fzBXAhCYHOEA35ePV+JKIEOhgHm99dxCUpjuUM0k5TWUmgYPF7d4TB5qg7+xWjkeFIDksH5mpYVtLlw0dDz4jbsISerLfg4U=
Message-ID: <9623c9a50804021710i238e26c1x220bd70b58740496@mail.gmail.com>
Date: Thu, 3 Apr 2008 08:10:13 +0800
From: "Xiao-Feng Li" <xiaofeng.li@gmail.com>
To: dev@harmony.apache.org
Subject: Re: Write a graphical front-end for Harmony memory management
In-Reply-To: <2c9597b90804020424o260bd1c4ie87ed524fa52d472@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline
References: <930503ab0803140408u2b8bc49aq8ddbad9c88464f6f@mail.gmail.com>
	 <9623c9a50803160553l559354f3q8b7bc4e214b929d2@mail.gmail.com>
	 <47DD8781.4020808@gmail.com>
	 <9623c9a50803161819l1e0de514i1337daf7b55df358@mail.gmail.com>
	 <47DE23C6.3040604@Stolsvik.com>
	 <9623c9a50803170243o45326380q1765f71093ead622@mail.gmail.com>
	 <c30d70960803290449j9db3ab1k5e9c4c2f700ae062@mail.gmail.com>
	 <2c9597b90803302234x72c8a0b8r2508ae93c224deae@mail.gmail.com>
	 <9623c9a50804020323p7daa2c97jbac7b30c0cf42db9@mail.gmail.com>
	 <2c9597b90804020424o260bd1c4ie87ed524fa52d472@mail.gmail.com>

On Wed, Apr 2, 2008 at 7:24 PM, Alexei Zakharov
<alexei.zakharov@gmail.com> wrote:
> Hi Xiao-Feng,
>
>  AFAIK it's just a general-purpose JMX console with SWT front end. I.e.
>  it can manage everything that exposes JMX-compliant interface. IIRC I
>  successfully tested it with Bea Weblogic J2EE server and Bea JRockit
>  JVM. BTW, to be able to manage JVM internal resources you'll need to
>  enable its management agent for local  monitoring: "-Xmanagement" on
>  JRockit and "-Dcom.sun.management.jmxremote" on Sun. I'm not sure if
>  we have something like it in Harmony. These agents just provide the
>  access to mbeans from "java.lang.management" package. So its
>  management interface is limited to the functionality provided by
>  java.lang.management. Sun also has management console called jconsole
>  - see [1]. JRockit has JRockit Management Console.
>
>  [1] http://java.sun.com/developer/technicalArticles/J2SE/jconsole.html

Ok, thanks. Looks like it's not directly applicable to the project.
java.lang.management is not so specific to GC internals. It has a few
interfaces for memory management, which are too high-level to expose
GC internals. On the other hand, TPTP could be a good reference.

Thanks,
xiaofeng

>  Regards,
>  Alexei
>
>  2008/4/2, Xiao-Feng Li <xiaofeng.li@gmail.com>:
>
>
> > On Mon, Mar 31, 2008 at 1:34 PM, Alexei Zakharov
>  >  <alexei.zakharov@gmail.com> wrote:
>  >  > BTW I just like pay everybody's attention on HARMONY-1105 [1]. It i=
s a
>  >  >  Java management console implemented as an eclipse plugin; donated =
to
>  >  >  Harmony on Aug 2006. May be it can be (re)used somehow for memory
>  >  >  management purposes too.
>  >  >
>  >  >  [1] http://issues.apache.org/jira/browse/HARMONY-1105
>  >
>  >
>  > Alexei, thanks for the suggestion. I need more understandings about th=
e console:
>  >  1. Which component does it work with in Harmony? I'd like to see chec=
k
>  >  the features in details.
>  >  2. Can it work with GC in low-level details? For example, can it
>  >  support to show an object is moved from one location to another (as
>  >  what we saw in Windows defragmentor)?
>  >
>  >  Thanks,
>  >  xiaofeng
>  >
>  >
>  >
>  >  >  Thanks,
>  >  >  Alexei
>  >  >
>  >  >  2008/3/29, Alexei Fedotov <alexei.fedotov@gmail.com>:
>  >  >
>  >  >
>  >  > > Let me share few ideas:
>  >  >  >
>  >  >  >
>  >  >  >  > Step 1. to make Harmony work with GCspy based on its current =
codebase;
>  >  >  >
>  >  >  >
>  >  >  > This is a good work breakdown, but I want to see more in the
>  >  >  >  application. I believe the goal of the project should be more v=
isible
>  >  >  >  and somehow reflect the usefulness. I would set the goal to imp=
rove
>  >  >  >  memory or performance profile of some application on Harmony us=
ing
>  >  >  >  GCspy, or in simpler words, enable GCspy to a degree when it wo=
uld be
>  >  >  >  able to do something useful.
>  >  >  >
>  >  >  >
>  >  >  >  > Step 2. to modify GCspy for our monitoring/tuning needs as an=
 Eclipse plugin.
>  >  >  >
>  >  >  >
>  >  >  > May be one should think of aligning GCspy and TPTP instead. This=
 is
>  >  >  >  more about supplementing TPTP with missed and useful things fro=
m GCspy
>  >  >  >  rather than GCspy porting. While it is useful sometimes and
>  >  >  >  unavoidable during hard times, for these two projects I do not =
see
>  >  >  >  enough arguments why they should compete instead of being merge=
d into
>  >  >  >  something more useful.
>  >  >  >
>  >  >  >  I agree with Xiao Feng that each of these tasks might be too am=
bitious
>  >  >  >  for a student project.
>  >  >  >
>  >  >  >
>  >  >  >  On Mon, Mar 17, 2008 at 12:43 PM, Xiao-Feng Li <xiaofeng.li@gma=
il.com> wrote:
>  >  >  >  >
>  >  >  >  > On Mon, Mar 17, 2008 at 3:54 PM, Endre St=F8lsvik <Endre@stol=
svik.com> wrote:
>  >  >  >  >  > Xiao-Feng Li wrote:
>  >  >  >  >  >
>  >  >  >  >  >  > For this GSoC project, it's probably enough to have the=
 bullet 1
>  >  >  >  >  >  > achieved, and GCspy is a very good reference. Well the =
wish for bullet
>  >  >  >  >  >  > 1 is to make it self-contained within Harmony as a plug=
in of Eclipse
>  >  >  >  >  >  > (i.e., least dependent on external software but Eclipse=
 related).
>  >  >  >  >  >
>  >  >  >  >  >  Why wouldn't it be *much* better to integrate the existin=
g GCSpy project
>  >  >  >  >  >  (which apparently requires a "server" integration in that=
 JVM to be done
>  >  >  >  >  >  - KVM has e.g. done it), and if it lacks something, code =
it into GCSpy
>  >  >  >  >  >  proper code? Then all will benefit..
>  >  >  >  >
>  >  >  >  >  Endre, thanks for the suggestion. I actually thought the sam=
e thing, I
>  >  >  >  >  am just not sure if a GSoC project can be that ambitious, si=
nce in my
>  >  >  >  >  understanding, GCspy is quite well-designed to accommodate v=
arious
>  >  >  >  >  runtime systems, and it's keep improving. Also I don't know =
GCspy's
>  >  >  >  >  maintenance model.
>  >  >  >  >
>  >  >  >  >  Probably we can partition the tasks into two steps:
>  >  >  >  >  Step 1. to make Harmony work with GCspy based on its current=
 codebase;
>  >  >  >  >  Step 2. to modify GCspy for our monitoring/tuning needs as a=
n Eclipse plugin.
>  >  >  >  >
>  >  >  >  >  How do you think about them? Thanks.
>  >  >  >  >
>  >  >  >  >  I don't know how Ian wants to roll out his RVM visualization=
 proposal.
>  >  >  >  >  Ian, do you plan to enhance GCspy with TuningFork idea or to=
 migrate
>  >  >  >  >  RVM to work with the TuningFork framework?
>  >  >  >  >
>  >  >  >  >  Thanks,
>  >  >  >  >  xiaofeng
>  >  >  >  >
>  >  >  >  >  >  Endre.
>  >  >  >  >
>  >  >  >  >
>  >  >  >  > >
>  >  >  >  >
>  >  >  >  >
>  >  >  >  >
>  >  >  >  >  --
>  >  >  >  >  http://xiao-feng.blogspot.com
>  >  >  >  >
>  >  >  >
>  >  >  >
>  >  >  >
>  >  >  >
>  >  >  > --
>  >  >  >  With best regards,
>  >  >  >
>  >  >  > Alexei
>  >  >  >
>  >  >
>  >
>  >
>  >
>  >
>  > --
>  >
>  > http://xiao-feng.blogspot.com
>  >
>


--=20
http://xiao-feng.blogspot.com