Notices where this attachment appears
-
@benis ASCII is not more semantic than binary. I agree that a larger namespace provides more extensibility than a smaller namespace, but I don't think multihash will run out of that byte any time soon, and I don't think the management of the semantics of that byte will have a huge ideological clash. If it does, fine, replace it with something better.
Maybe do like microsoft did and replaced their fourCC with a GUID. By the time we need 512-bit hashes I guess a 128-bit identifier isn't quite as insane, but even at that point it still seems quite overkill. And "urn:sha512" is already 80 bits.
I'm a big fan of the http://wiki.c2.com/?PowerOfPlainText , but it only makes systems more understandable for humans, not machines. And while I do think some people go on a djb-like crusade against plaintext in protocol, I don't think it's entirely unwarranted when you're planning to shove terabytes around. And I'm also a big fan of http://wiki.c2.com/?YouAintGonnaNeedIt .