[Standards-JIG] File sharing JEP

Pedro Melo melo at co.sapo.pt
Thu Apr 13 15:20:55 UTC 2006


On Apr 13, 2006, at 3:57 AM, Trejkaz wrote:

> On 12/04/2006, at 20:13 PM, Pedro Melo wrote:
>> Not quite true: all the bytes must come through. The order is not  
>> important as bittorrent proven already.
> True.  Although it would greatly violate the Jabber ideals of  
> keeping client development simple, if clients suddenly had to  
> implement something like BitTorrent. :-)


Well, yes, but if each UDP packet tell's you the size, offset, and ID  
of the piece of the file he is carrying, reconstructing the file is  
very simple. The boring part is keeping track of the missing IDs to  
ask for them later.

And after that, having multiple sources for your file is not a long  
strech. :)

hmms... bittorrent profile for jingle... mesa likes.

>> If later a jingle "super-fast" protocol comes along cool, you'll  
>> just use it. But not having a common basic file transfer ASAP is  
>> hurting us all, me thinks.
> But we have one already...
> Nobody has been saying "abolish stream negotiation NOW, let's move  
> to Jingle."  They have been saying "let's try and see if Jingle  
> will work for file transfers, and over time, start using it instead."

yeah, jep-0065 is starting to be generaly available. I guess you're  
right on that one.

py-msnt has it, wildfire, ejabberd and xcp. And I think peter is  
close enough to put out a cleaned up version of the external  
component :)

On the client side, exodus does it well, and I think gajim also, but  
psi doesn't do the discovery part.

It's getting beter.

best regards,
HIId: Pedro Melo
SMTP: melo at co.sapo.pt
XMPP: pedro.melo at sapo.pt

More information about the Standards mailing list