Re: [quicwg/base-drafts] 5-tuple routing and SPA (#3608)

[Martin Thomson <[e-mail]>]

A truth table might help here.

... | migration supported (no signal) | disable_active_migration set | 
---|---|---|
no preferred address | normal (migrate as you like);  | don't migrate |
preferred address | normal (migrate at your leisure) | don't migrate |

Right now, our definition of disable_active_migration is a simple "don't migrate".  But the argument here says that the server that provides a preferred address is *probably* capable of handling migrations on the basis that it has to support one.  We might insist on the source address not change, but the fact is that some NATs (those with address-dependent mappings) will change the source address.

Changing to a definition where disable_active_migration is instead "don't move to a new address unless you use a preferred address" is a little more complicated, but doable.

Is it worth the extra complexity this late in the process?  Is that complexity justified when it would be trivially possible to define another extension that said "I support migration on my preferred address"?

-- 
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/3608#issuecomment-623267502