clayton rollins zei:
>>Quoting Raphael Manfredi
>>
>>I expect that a lot of PARQ entries will obsolete at some point:
>>servents  that
>>have requested them will disconnect and no longer be reacheable.  Only
>>servents up 24x7 will be able to maintain their queued entries,
>>

> <<<< SNIP >>>>
> A reasonable solution, it seemed to me, would be to count the amount of
>  timeouts/connection failures for those hosts and remove all entries if
> this  exceeded the user defined amount, using that as a guideline for a
>  non-existant host. Of course, connection failures like "too many
> uploads"  wouldn't enter into this, as they at least indicate that a
> servent is there.

Well, according to the PARQ specifications a host which didn't retry
within the timeout specified in the Retry-After header should be removed
from the queue.
However, if queues can get large, this isn't a modem friendly thing. A
cable user would be able to have an uptime of 24x7 (I do) and thus
maintaining their queue position. A modem user which is only online 1 a 3
hours a day will never be able to reach an upload slot.
So I wondered if it would be a good thing to reserve a queued slot for 7
days. However, if such a slot isn't retried within the Retry-After period
it doesn't move forward into the queue, but stays at the current position.This will 
give the '1 hour online' modem users the ability to retreive an
upload slot, allthough it might take a bit longer than a cable user.To determine 
wether an (or or more) upload slot is already reserved for
client the GUID should be used I guess as IP and port might and probably
will change when a modem user reconnects (does this still 'break'
morpheus?).
I know this is not according to the PARQ specs but I do think it is
another point for 'fairness'. Does it make sense?
With regards,
Jeroen




-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com
_______________________________________________
Gtk-gnutella-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/gtk-gnutella-devel

Reply via email to