Re: [quicwg/base-drafts] encoding of CONNECTION_CLOSE reason phrases (#1990)

[Martin Thomson <[e-mail]>]

The problem here is that tagging isn't that useful unless you also couple it with negotiation.  We can annotate the string with a language tag, but that doesn't make the text more comprehensible if the chosen language isn't understood.  If the endpoint is able to generate translated strings, then it doesn't know which one to use.  It's possible that we could signal the preferred language of endpoints in transport parameters, but that leaves clients exposing their language preferences in the clear.  

Then we need to consider that the language preferences of the endpoint might not even be relevant.  This information is largely only useful during debugging, so it is the language preference of the stack developer that is likely to matter.  That said, if the messages were passed to the stack developer, that is likely to represent a risk to the privacy of the endpoints or those using those endpoints.  And there is a risk of misinterpretation if the language preference of the person using the endpoint differs from that of the stack developer.

In other words, this isn't a trivial problem.  We could just tag the message and leave it to online translation tools and the like to sort things out, I guess.  But @mikkelfj's conclusion is the one we took in HTTP/2, though we chose to say nothing in that case.

-- 
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/quicwg/base-drafts/issues/1990#issuecomment-437716951