Field separators, again

[Martin Atkins]

Brad Fitzpatrick wrote:
> 
> Then they say:
> 
> openid.response_type=text
> Or:
> openid.response_type=xml
> 
> Then we leave it to consumers to decide which they want to parse.
> 

This seems to be getting a bit complicated. My vote would be just for
picking one (any one; as a perl programmer, I really don't care!) format
and making it mandatory. Any real programming language can parse any of
these in a trivial amount of code. You'd be able to parse that
key:value\n format in pure S2, even. (not that there'd be much point)

My only reservation about XML is that it's a lot "bigger" (especially
when you're putting all of the element names in an attribute for some
reason) and wastes a lot of bytes. I've long ago lost track of how
exactly these long strings are passed, but assuming we're still passing
them in a URL it's just going to get silly.

As you noted way back at the beginning, giving ID servers and consumers
choices just for the sake of it is bad, because they'll just support
whatever the big guys (LiveJournal/TypeKey) do and ignore everyone else.