[flud-devel] Short-term plans

[Chinmay Kulkarni]

Hi.

New to this mail group.

Why aren't LT codes being considered? Also, if I haven't forgotten; RS codes
cannot generate packets independently of the packets generated so far. This
could be a problem if a host goes offline, and we simply have to create one
more packet to maintain the desired redundancy.

Is the variance high enough that we can't cover it using our redundancy
criteria?

On 9/28/07, Alen Peacock <alenlpeacock at gmail.com> wrote:
>
> On 9/27/07, Alen Peacock <alenlpeacock at gmail.com> wrote:
> >
> > A1. fix erasure coding
>
>   I've long maintained that LDPC is a good choice for flud, based on
> some initial experiments that I did a long time ago, confirming that
> the *average* number of blocks needed for decode could be made to be
> reasonable, even for a small number of blocks.  And while that is
> still true, especially when there are a very large number of blocks,
> I've done some more recent measurements that seem to show the
> *variance* in the number of blocks might be unacceptable (that will
> come as no surprise to many).  As part of fixing this, I may have to
> recant my previous biases against RS encoding, which *can* provide us
> with hard limits on the number of blocks needed.
>
> Alen
>
> _______________________________________________
> flud-devel mailing list
> flud-devel at flud.org
> http://flud.org/mailman/listinfo/flud-devel_flud.org
>



-- 
Yours,
Chinmay Kulkarni,
www.celbits.org
+91 99835025822

Think BIG!
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://flud.org/pipermail/flud-devel_flud.org/attachments/20070928/4368d72f/attachment.html