20. How does it run?
●
Volunteers
●
Reviews by mail
●
Mailing list driven
●
Test suite and autobuilds
●
Small core team
21. When is it done?
●
How long is a rope?
●
When is the last bug found?
●
When do we stop adding
functionality?
22. New Technology
●
Internet, protocols and file
transfers evolve
●
Curl has to evolve along
●
You can help!
23. Happy Eyeballs
●
Dualstack behaviors
●
RFC6555
●
Basically two connect attempts
at once
24. DANE
●
DNSBased Authentication of
Named Entities (DANE)
●
RFC6698
●
Because SSL's CA system is
broken
●
Resolver dependency
●
Should use a lib
25. SRV/URI records
●
DNS based hints to find (web)
servers
●
In use by nonHTTP protocols
●
Latency penalties
●
Tricky resolver dependencies
26. HTTP pipelining
●
Present in HTTP 1.1, RFC2616
●
Riddled with server problems
●
Circumvents latency issues
●
Coming soon!
27. HTTP2 (SPDY)
●
Discussed in IETF's httpbis WG
●
Based on SPDY/3
●
SCTP and SSH like with multiple
streams within a physical TCP
connection
●
Spindly vs spdylay
28. New HTTP auth
●
Passwords must die
●
Existing auth methods are full of
problems
●
Digest relies on MD5
●
But... browser basically don't do
HTTP auth
29. SSL to proxy
●
SSL over proxy is usually done with
a HTTP CONNECT over plain HTTP
●
Increase privacy within
organizations
●
Supported by Chrome
●
Tricky because of SSL backends
30. … and much much more!
●
It never ends
●
Transports are fundamental to
Internet
●
curl transports the Internet
31. How?
●
Companies fund features
●
Individuals fix problems
●
Join us and help out!