On Thu, Nov 13, 2008 at 8:23 PM, Tim Ellison <[EMAIL PROTECTED]> wrote: > Diab Abuaiadh wrote: >> Hi Xiaofeng, >> >> I left IBM at 2006. I thought that IBM gave J9 to the open source community. >> If I have J9 or (sun jvm...I think it is called sov) I will be able to >> implement >> the new algorithms in short time....Maybe one month or less. >> >> I am not familiar with DRLVM...I am afraid it may take time to implement >> the new algorithms....the other things is if I implement the new algorithms >> on J9, I will be able to compare the result with the previous algorithm in >> J9. > > If you write a new algorithm on DRLVM you can still compare it to the > existing J9 algorithm :-) or the DRLVM algorithms. > > We are aiming to improve the DRLVM in this project, as mentioned > elsewhere, J9 is not open source so no opportunity for you to contribute > directly there. > >> Question: what algrithms you use to implement the garbage collection in >> DRLVM? >> what compaction algorithm you use? > > Did you look at the website already? There are a number of documents > that will be of interest to you, including a How-To for new GC > algorithms [1] > > I suggest you check your ability to contribute to the project, then > describe your ideas and work with Xiao-Feng and everyone on this list to > implement it. I'm sure you'll get some useful feedback.
Right. I am very happy to help the new algorithm be implemented in DRLVM, assuming there is no legal obstacle. Diab, I realized I am quite familiar with the algorithm you published in OOPSLA2004. :) DRLVM has a couple of GC implementations. All of them share the same code base (or same GC infrastructure). You can write a new GC from the scratch by only using the GC/VM APIs defined in Harmony, or you can develop your algorithm in the same code base as the existing GCs, then you can leverage the infrastructures. Thanks, xiaofeng > [1] http://harmony.apache.org/subcomponents/drlvm/gc-howto.html > > Regards, > Tim > -- http://xiao-feng.blogspot.com
