[art] Is CT single-use origins or not? (Re: On BCP 190)

[Adam Roach <[e-mail]>]

On 7/24/19 09:46, Melinda Shore wrote:
> In a chat with Adam earlier today he said that there has, in fact,
> been squatting on namespace elements like robots.txt and favicon.ico.
> That's extremely helpful and the sort of thing that might support
> the case for requiring BCP 190.  If we have (or have not) actually seen
> the same problem in application namespaces or on single-use servers,
> that would be be helpful to know, as well.


While I think we need to have a broader conversation on BCP 190 that 
covers this in depth, there is one critically important bit of 
information that has not been made clear about the way that CT is being 
deployed, and I don't think there is any hope of quickly processing the 
CT document until it is clearly answered:



     Is the mechanism defined in the "Certificate Transparency
     Version 2.0" document constrained to run on an origin on
     which no other services are present?



The answer to this question *radically* effects the conversation we need 
to be having to get the document published as quickly as possible. We 
can try to finish the wide-ranging conversation about BCP 190 in general 
without an answer to this question, but unless we want to block the 
TRANS document while we wait for that to happen, we're going to need to 
understand this fundamental property as soon as possible.

/a