Should namespaces be globally defined? #920
Replies: 2 comments
-
@bbockelm this is something we tried to discuss during the team meeting on 03/08/24 but quickly decided we needed your input. |
Beta Was this translation helpful? Give feedback.
-
This also ties into what should the In addition, within the client code, we changed looking for There were a few other ideas that came up when discussing this with @jhiemstrawisc but this is becoming a little confusing. I think we need to have a discussion about this with the whole team |
Beta Was this translation helpful? Give feedback.
-
As we start to work on disentangling origins and namespaces, we've run into the issue where multiple origins may export the same namespace prefix with different advertisements. Under the hood we map an origin advertisement to the list of namespace advertisements it exports, where those namespace advertisements are defined by the origin. So, for example, if
origin1
andorigin2
both export the namespace prefix/foo
, they may do so such thatorigin1
says/foo
is publicly readable andorigin2
says it isn't.This begs the question -- should these namespace advertisements be global, preventing two origins from trying to represent the same namespace prefix differently? Currently we are not equipped to handle the case where two origins have misaligned namespace ads -- we simply grab the namespace ad from whichever origin is the first we match against. Using the previous example, this means that if we match against
origin1
first, our headers will indicate the namespace is public and thus provide no token generation information. However, the client might still try to read from the namespace prefix atorigin2
if it makes use of the metalink headers.Beta Was this translation helpful? Give feedback.
All reactions