Regarding ESNI RFC (https://tools.ietf.org/html/draft-ietf-tls-esni-02) it copes with many possible security vulnerabilities. However I'm interested how is it supposed to prevent confidential information leakage (SNI) in case of HTTP Proxy is used (or forced to be used)?
HTTP CONNECT request sends domain's FQDN in plaintext before any connection is tunneled.
I know proxy is another layer not so related to ESNI itself but I think it should be considered as a possible weakness.
Am I missing something here?
(btw please someone add new tags "esni", "sni" and "tls-1.3" to this question - don't have reputation for that, thanks)