I am beginning to think that using CHKs for SplitFile checksums isn't a good 
idea.  

If we use something that is widely deployed like MD5, then SplitFile client 
authors can check the checksum themselves using their local system library, 
without having to do an FCP request.

I think insertion/retrieval client authors will be annoyed if we make them 
send their entire file *over the wire* just to get a checksum CHK.  Think 
ISOs.....

Would there be any objections to using MD5? Is there a cannonical GPL'd java 
implementation? 

If we go this route I will add an MD5 FCP command.

Also, I want to add an optional Client field to InfoPart for the insertion 
client name.  This will make bug hunting somewhat easier.  Any objections?


--gj

_______________________________________________
devl mailing list
devl at freenetproject.org
http://hawk.freenetproject.org/cgi-bin/mailman/listinfo/devl

Reply via email to