Normally, a DBR increments daily to change a key which changes the content on 
a page daily.  What if that remains the same, but a diff file is posted, say 
hourly, that shows the difference in the content from the last full daily 
insert?  The daily DBR would be reliable because it is needed for the hourly 
diff patch to be placed, which would remove the unreliability of short-time 
DBRs.
This technique could also be layered into smaller time chunks.  The hourly 
diff would be based off the daily DBR, and a 1-minute diff could be based off 
the hourly diff.  If enough people are interested in the same content at the 
same time, that data would be most reliable.  If the most recent one fails, 
the request would have already requested the diff or data of the next level 
up.  This would allow sites to update frequently and not risk unreliability.  
It would also make freenet quicker at getting less recent data when the most 
recent data is inaccessible.
Since diff isn't the most efficient algorithm, another one could be used 
instead, such as the one in rsync (which could be optimized for this purpose 
since rsync runs in very limited conditions).  This could decrease the size 
of the "diff" files and possibly make the reconstruction faster.
So how does this sound?

-Scott Young

_______________________________________________
freenet-tech mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/tech

Reply via email to