W3C home > Mailing lists > Public > public-credentials@w3.org > March 2021

Re: RDF Dataset Canonicalization - Formal Proof

From: Manu Sporny <msporny@digitalbazaar.com>
Date: Sat, 27 Mar 2021 15:40:23 -0400
To: public-credentials@w3.org
Message-ID: <9a4b62fb-4f14-5ed5-3e61-f50945a19e1f@digitalbazaar.com>
On 3/27/21 3:01 PM, Alan Karp wrote:
> Yeah.  I'm still trying to figure out what I'm going to be when I grow up.

*lol*, aren't we all! :P

> One issue you didn't mention about our paper is that a set hash is weaker 
> against collision attacks. I thought that might be the reason you couldn't
>  use that approach.

Well, yes... I wanted to say something about that, but could also see how you
could *maybe* mitigate that using large enough hashes and/or Section 6.2.2 --
making the combining function be multiplication mod some suitably-large prime
number. This was the part of the paper that interested me the most, Alan... I
could see how that would work IF we didn't have to depend on a pre-determined
set of node labels.

There are performance improvements that we know are probably still locked up
in the algorithm, but we needed to ship something (nine years ago) and we
really haven't seen a case where performance was an issue.

> In case you're interested, we wrote a follow-up, 
> https://www.hpl.hp.com/techreports/2004/HPL-2004-95.pdf

Would you mind it if we unceremoniously lift applicable parts of "Section 5:
Application for Graph Digests" from that document for the use cases document?
Any reason you didn't include digital signatures in the applications section?

-- manu

-- 
Manu Sporny - https://www.linkedin.com/in/manusporny/
Founder/CEO - Digital Bazaar, Inc.
blog: Veres One Decentralized Identifier Blockchain Launches
https://tinyurl.com/veres-one-launches
Received on Saturday, 27 March 2021 19:40:56 UTC

This archive was generated by hypermail 2.4.0 : Saturday, 27 March 2021 19:40:57 UTC