added testnet attestation

[cloutier]

This is a pretty trivial change overall, but it seems people building apps on top of OTS are reimplementing this all the time in slightly incompatible ways, so it would better to have this part of the protocol.

I have not found anywhere the methodology to determine the tags for the attestation, so I used the mainnet one + 1.

Pinging @RCasatta because he seems be using testnet OTS as well: https://github.com/RCasatta/opentimestamps-server/tree/testnet

[petertodd]

Strong NACK

The point of testnet and regtest is to test things, not do useful work. The best test is one that uses the exact same code paths; adding a specific testnet attestation means the codepaths aren't identical, and thus it's not as good of a test.

[cloutier]

I've been thinking and I think you are right.

The motivation was to make sure a our timestamping product would never get corrupted by a misconfiguration in our OTS server, but it would be better that we don't even treat our ots server as trusted and verify everything. It would remove a whole bag of potential bugs.

[petertodd]

Thanks!

Added an issue for this: #21

You're not the first person to change their mind about this. Heck, I changed my mind about this awhile back. :)

[cloutier]

Would it make sense to have a reserved tag for WIP that I would add to the documentation? Let's say 0xFFFFFFFFFFFFFFFFF

And for the sake of documentation, how would one choose a tag for a chain? I tried to reverse engineer 0x0588960d73d71901 with no success.

[petertodd]

The tags are just 64-bit random numbers, picked randomly. They mean absolutely nothing. 64-bits is big enough that it'd take 4 billion of them to have a 50% chance of collision (birthday paradox).

So for WIP's, picking a non-random number is reasonable while you're still testing. But equally, just picking your own is fine too.