As noted in #8, 200.html should only be served in response to a request where text/html is in the Accept header.
At first (as described in #8), I was thinking about using a directive to restrict MIME negotiation. Then, I was thinking about making this something specified as part of the URL, with a hash-fragment in the identifier for handling exotic rules (an idea that would be pretty spooky, in terms of magic-y behavior).
Ultimately, after a little thinking about it, the answer seems pretty obvious: just make content negotiation a part of route matching (at least for 200 responses) - so, if the file that would be served isn't acceptable to the user agent, the route doesn't work.
Indeed, this idea of making content negotiation a part of routing is pretty cool - this might even get supported further with a method like negotiated 200, where multiple file paths can be specified and the winner is served according to content negotiation. (This wouldn't be an extension to the current 200, because I still consider the fourth field onward to be reserved for normal methods.)
As noted in #8,
200.html
should only be served in response to a request wheretext/html
is in theAccept
header.At first (as described in #8), I was thinking about using a directive to restrict MIME negotiation. Then, I was thinking about making this something specified as part of the URL, with a hash-fragment in the identifier for handling exotic rules (an idea that would be pretty spooky, in terms of magic-y behavior).
Ultimately, after a little thinking about it, the answer seems pretty obvious: just make content negotiation a part of route matching (at least for
200
responses) - so, if the file that would be served isn't acceptable to the user agent, the route doesn't work.Indeed, this idea of making content negotiation a part of routing is pretty cool - this might even get supported further with a method like
negotiated 200
, where multiple file paths can be specified and the winner is served according to content negotiation. (This wouldn't be an extension to the current200
, because I still consider the fourth field onward to be reserved for normal methods.)