Closed jamezp closed 2 years ago
NB: Apparently localhost
is covered by https://datatracker.ietf.org/doc/html/rfc6761 hence a mandatory Internet Standard.
The host I've got no strong argument for. I can see where using localhost is fine, but I've definitely seen issues in some environments before. Maybe not a big deal here though.
The HTTP I respectively disagree with. If it was meant to be "HTTP" then the protocol should be represented by a type not a String. Requiring a specific string in a specific case is bad practice IMO.
Regarding "HTTP" the explicit uppercase is enforced by the commonly accepted API already. This is not the schema string, this is the protocol name. If e. g. a product supports HTTP/3 then the schema still is lowercase "http" but to enforce it, the protocol name needs to be "HTTP/3" vs "HTTP". Maybe it is more clear now?
Please note that the time to change the API is over. It is discussed and accepted. The TCK only enforces the API as-is.
I can't say I fully agree, but it's not really that big of a deal so I'll just close this.
…TP in all uppercase.
Again if this needs an issue file please let me know.
For the
HOST
change I don't think it's generally safe to assumelocalhost
will work on all environments, e.g. CI.