harmony-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Senaka Fernando" <senaka...@gmail.com>
Subject Re: Comparison of Harmony GC_Gen vs. Parrot
Date Mon, 07 Apr 2008 04:55:53 GMT
Hi Xiao-Feng,

Thanks for these questions. I believe that they'd be really helpful in
understanding VM <-> GC assumptions on the Parrot end.

Will work on answering these, and perhaps a comparison with Harmony.

Regards,
Senaka

On Mon, Apr 7, 2008 at 5:11 AM, Xiao-Feng Li <xiaofeng.li@gmail.com> wrote:

> Senaka, thanks for the page. I think the most important things are
> related to the VM <-> GC protocol. Some questions that may help you:
> 1. How Parrot layout/encode an object/array? fields, size, object
> header info, etc.
> 2. How Parrot layout/arrange the heap? free list? pages?
> 3. What's the process of an object creation? When and how?
> 4. How is a collection process triggered?
> 5. How does Parrot GC trace live objects and collect them?
>
> Some of the questions above might be GC internals, so it's more
> desirable if you can understand the Parrot VM's assumptions on GC.
> I.e., does it assume the heap is laid out in certain way, does it
> assume the objects are encoded in certain way, does it assume the
> roots are enumerated in certain way, etc.? Depending on your progress,
> more details might be needed later on.
>
> For this project, you might have to understand the Parrot current
> status for the above questions. It helps you and us to identify the
> key issues to resolve, and the main efforts to be focused on. For GC
> porting over different VMs, it's not like porting an application over
> different OSes, because of the implicit assumptions between VM and GC.
> I would expect some redesign work required for GC porting, hence you
> have to understand the Parrot design in certain depth.
>
> Thanks,
> xiaofeng
>
> On Mon, Apr 7, 2008 at 4:26 AM, Alexei Fedotov <alexei.fedotov@gmail.com>
> wrote:
> > Good job, Senaka.
> >
> >  The general perception was that internal and external GC interfaces
> >  were mixed, which maked this document less usable for harmony-gc-5
> >  project than it could be. For example, sweeping, marking and
> >  reclaiming are internal interfaces while allocation, stack enumeration
> >  (please take a look at vm_enumerate_root_set_all_threads) and gc
> >  invocation are external interfaces. We should pay more attention to
> >  external interfaces for harmony-gc-5 project.
> >
> >  Thanks.
> >
> >
> >
> >
> >
> >
> >
> >  On Sun, Apr 6, 2008 at 8:35 PM, Senaka Fernando <senakafdo@gmail.com>
> wrote:
> >  >
> >  > Hi all,
> >  >
> >  >  I have almost finished comparing the two interfaces of Harmony and
> Parrot.
> >  >  However, I'm not 100% sure on whether I got everything right, but I
> believe
> >  >  that most of it is. Therefore, it would be really great if you could
> review
> >  >  the wiki page and let me know whether it is correct and precise.
> >  >
> >  >  Sections marked as TBD are not yet finalized. And, I have omitted
> some
> >  >  instances of where Harmony supports something and Parrot doesn't for
> >  >  simplicity.
> >  >
> >  >  The wiki page is found at [1]
> >  >
> >  >  [1]
> http://wiki.apache.org/harmony/gc_comparison/gc_gen_harmony_vs_parrot
> >  >
> >  >  Regards,
> >  >  Senaka
> >  >
> >
> >
> >
> >  --
> >  With best regards,
> >  Alexei
> >
>
>
>
> --
> http://xiao-feng.blogspot.com
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message