goran at krampe.se wrote:
> And how would it be referenced? :) The thing is that if you want to be
> able to use Squeak without having the map inside the image - then you
> will have to use some kind of textual or numeric id that can be remapped
> when the SMSqueakMap structure is loaded into the image. And... right,
> that is what we have today - the dev initials. :)

Yep. I was thinking of using the UUID, or some combination of dev 
initials/uuid/email address, rather than just the initials, for 
reconstructing the developer-ID object from the change-stamp.

This quickly gets us into web-of-trust territory, and...

> So unless you explain more specifically what you mean and what advantage
> it gives us, I am not really seeing it.

... you're right, I don't have it perfectly clear in my head. Ideally 
I'd love to see some formal definition of what a developer identity 
really is (a PGP key? a UUID? just the initials? ...) and once the 
equivalence between developer-IDs is rigourously defined, to work toward 
having an implementation of a developer-ID object that honoured that 
equivalence.

I'll think about it some more.

Oh, reading further, it seems you've already gotten close to where I 
want to end up! To wit:

> 1. Add an object in the system representing "me". A little more than
> just the dev initials. An object keeping at least full name, email and
> dev initials. And perhaps a public key. This object can then be used by
> changeset tools and MC etc in order to put more interesting info into
> files that are exported out of the image.
> 
> 2. When you have SM - our collective memory - in your image (typically
> what 99% of us will have) then the above described object can easily be
> mapped using the dev initials to the corresponding SMAccount and then
> the system also "knows" your "SM-public" full name, email, key.
> 
> Perhaps this is what you were describing?

Yes, that's very close. Including the public-key and/or some kind of 
UUID seems very important.

I'd love to see a system that implemented the principles described in 
these pages:

http://zooko.com/distnames.html
http://www.erights.org/elib/capability/pnml.html

More context:

http://shirky.com/writings/domain_names.html
http://www.livejournal.com/users/ciphergoth/110893.html

Regards,
   Tony