On Thu, Mar 13, 2003 at 01:59:51PM -0500, Dan Merillat wrote:
> There's a few splitfiles out there (*cough*movie central*cough*) That
> are barely retrievable because of impatience.
> 
> The first segment returns fine, but each segment after that has higher
> and higher loss rate.  In fact, this is the exact problem we had with
> splitfiles when they were retrieved linearly (first blocks were always
> found, latter blocks were gone)
> 
> The problem is REALLY aggrivated when people retry over and over... they
> heal segment 1 a lot, so it has close to 100% retrievability, but
> segment2, 3, 4 lose more and more blocks.
> 
> "obviously", we should retrieve the segments in random order.  That has
> the obvious problem that we can't send the file until the download is
> complete, anyway.

The problem with that is that if we are sending the data to the browser,
it should be possible to send it a segment at a time.
> 
> So, let's step back a second on this.  Browser timeout is killer on
> large files already, that download session going for possibly hours
> without getting a single byte.  So, why not spool the file to local
> holding and initiate the download to the browser once the whole thing
> is there.

In other words, implement the background download manager thing ian has
been suggesting (and oskar before that)? Good idea. Whether we save
direct to the final filename or save to a buffer area and then let the
user grab the file all at once is open to question; probably we should
implement some way of doing either or both.
> 
> I'd prefer this solution, actually... this way I can simply wget the
> resulting file to wherever I want it (or copy the segments from
> store/tmp) and be able to check the progress from anywhere (by knowing
> the splitfile ID URL)
> 
> --Dan



-- 
Matthew Toseland
toad at amphibian.dyndns.org/amphibian at users.sourceforge.net
Full time freenet hacker.
http://freenetproject.org/
Freenet Distribution Node (temporary) at 
http://80-192-4-36.cable.ubr09.na.blueyonder.co.uk:8889/069-bfPzGwU/
ICTHUS.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: 
<https://emu.freenetproject.org/pipermail/devl/attachments/20030313/e999676d/attachment.pgp>

Reply via email to