On 28/12/06, Anthony wikilegal@inbox.org wrote:
On 12/28/06, Gregory Maxwell gmaxwell@gmail.com wrote:
On 12/28/06, Anthony wikilegal@inbox.org wrote:
You shouldn't buy another 300 servers in the first place.
What a great idea. When will you have your implementation of "distributed mediawiki" completed?
It's already completed. I just need you to transfer over the domain names.
whuh??? Project page?
Your post actually started me thinking on this. A Freenet (for encryption and to avoid a [[:en:trusted client]] problem) or BitTorrent (for convenient distribution with a program lots of people have) method of distributing Wikipedia. A peer-to-peer network with WMF as the only body supplying content files. The downsides that spring to mind are (a) doesn't work through any old web browser (this is a big one) (b) trusted client problem (how to ensure rogue clients don't redistribute corrupted content in our name).
- d.
David Gerard wrote:
Your post actually started me thinking on this. A Freenet (for encryption and to avoid a [[:en:trusted client]] problem) or BitTorrent (for convenient distribution with a program lots of people have) method of distributing Wikipedia. A peer-to-peer network with WMF as the only body supplying content files.
Erik Garrison (CC'd) looked into this in detail (and I think even worked on a prototype) for a class project. He'd be the person to talk to about details, downsides, and so forth.
On 12/29/06, David Gerard dgerard@gmail.com wrote:
is a big one) (b) trusted client problem (how to ensure rogue clients don't redistribute corrupted content in our name).
I bet this one has been solved a zillion times over. A CRC check determines whether it's corrupted or not. How do you force the user to CRC check it? Perhaps a binary decompressor/decrypter only available from wikimedia.org?
In any case, en dumps over bittorrent sounds good. Especially if they were readable without mediawiki.
Steve
wikitech-l@lists.wikimedia.org