Add section on extensibility. Fixes #58.

[msporny]

---

Preview | Diff

[ChristopherA]

ACK, though I’d prefer a different example rather than favorite food as personal information should not be put into a DID Object. Instead something like a preferred contact address or something relevant to to a legacy identity (PGP fingerprint?). We don’t want to encourage DID Objects to become FOAF-like bags of personal information but instead have it point to the bags :-)

[msporny]

ACK, though I’d prefer a different example rather than favorite food as personal information should not be put into a DID Object.

Hmm, so one of the first things I'm going to do is port my LinkedIn profile to a public DID Document... I think there are valid use cases.

Then again... I also did this: http://manu.sporny.org/2011/public-domain-genome/

...so maybe I'm not a good statistical sample. :)

Your point stands, we need a better example.

Instead something like a preferred contact address or something relevant to to a legacy identity (PGP fingerprint?).

... I'll try to brainstorm something that's not super geeky... maybe a public photo stream?

We don’t want to encourage DID Objects to become FOAF-like bags of personal information but instead have it point to the bags :-)

I agree, with a few caveats. We really need a "best practices" guide.

[msporny]

@ChristopherA wrote:

I’d prefer a different example

Done in https://github.com/w3c-ccg/did-spec/pull/78/commits/60782e3a3e757937d0bbd8dcba238ce7a243ed6b

[ChristopherA]

@msporny commented:

Hmm, so one of the first things I'm going to do is port my LinkedIn profile to a public DID Document... I think there are valid use cases.

In fact, with BTCR the common pattern is to point to a static file or IPFS object that is bag of self-signed verifiable credentials (which in fact may be appended to the DiD Object). But inside the BTCR DID Object only points points to it.

Longer term the DID Object pointer may include a clue to describe how to decrypt the file, such as using a mutual DH key. I really like the miniLock format design, which I’d like to adapt to work with DID keys. This way you could only decrypt my Verifiable Credentials if you were in my web-of-trust set.

[msporny]

Hey @rxgrant - please review PR #78 on the DID spec... it's in response to issue #58 that you raised at RWoT 6.

[msporny]

Merging, we can modify the section at a later date if @rxgrant wants further changes.