Assuming this as a feature request, changing the status to Untriaged and adding the appropriate labels.
Thank You.

Hrm...Doesn't look like Windows has any docs on using fastopen.  We may have to wait for a published API.

fyi: I've pinged Edge Dev on Twitter: https://twitter.com/kurtextrem/status/763788930961567744 let's get this going :)

And an answer: https://twitter.com/toddreifsteck/status/763789371333980160
(copy:) Docs still updating. This help? set/getsockopt(sock, IPPROTO_TCP, TCP_FASTOPEN

Don't think that's enough.  Looks to me like TCP_FASTOPEN is not defined on the latest Windows SDK.  Also not clear if once that's done, a socket should be treated as a normal TCP socket or not (With some APIs, have to connect using the UDP/datagram sendto method when sending the first chunk of data).

Don't think there's any real rush to implement fast open.  It's nice to reduce connection time, but it does break a number of things, as implemented in Chrome:  Happy eyeballs, fallback to other addresses on connection failure (As they'll now look like send failures), connection timeouts (For the non-SSL case, at least), etc.

 Issue 638279  has been merged into this issue.

TCP_FASTOPEN is indeed defined in the latest Windows SDK.

Any news? :)

No news yet. This is on the list of things to be done, but is not high enough on the priority list yet. As mmenke points out, it breaks Happy Eyeballs and other things in Chrome so we can't do a broad deployment yet with the current API. (I'm talking to the Windows folks on fixing their API.)

Thank you for the update!

What the heck is going? For god's sake its 2017 and this is still not finished.
No Microsoft will not change the API to accommodate you.

Firefox has this implemented already.

So how about TCP fast open status? Will chrome deal with it? Still cant work at windows 1703

I think they've updated the SDK and also documented it.

Ref: https://msdn.microsoft.com/en-us/library/windows/desktop/ms738596(v=vs.85).aspx

This has to be a joke. A total complete joke.

What's the status on this?

Owner status is: Last visit > 30 days ago. Has he left the Chromium project?

kurtextrem: As said earlier, this is on the list of things to be done, but is not high enough on the priority list yet.

"TLS 1.3, will allow developers to eliminate that delay in most cases while still encrypting content. That means delivering better performance and security in Microsoft Edge, using modern encryption on top of the continually improved TCP stack. TLS 1.3 is working through the standardization track now, and the IETF expects to publish it this summer. But even without TLS 1.3, we can combine TCP Fast Open and the TLS False Start option, and reduce the delay from 3-RTT (round trip time) to 1-RTT. Even reducing your page load time by an average of 50 milliseconds will contribute to a better browsing experience. (src: https://blogs.windows.com/msedgedev/2016/06/15/building-a-faster-and-more-secure-web-with-tcp-fast-open-tls-false-start-and-tls-1-3/). 

This is kind of stuff that is low hanging fruit. I'm shocked Microsoft realized it and Google seemingly doesn't care

Hi I'm the PM for Windows transport at Microsoft and this just came to my attention.

TCP Fast Open is supported in Windows as well as documented on MSDN.  We also have a fallback algorithm to mitigate middle box issues and we have not experienced any interference with Happy Eyeballs.

It is a simple API call to activate TFO and we would be happy to work with your dev team.  Also I noticed that the text in chrome://flags says that TFO is not supported on Windows.  This is no longer true.

Looking forward to hearing from you.
...Daniel

We're removing FastOpen on Linux (Which we never enabled by default anyways), and have no plans to add it elsewhere.  With TLS 1.2 session restore, the session resumption packet potentially contains data that the server will process, resulting in mutation of server state before Chrome knows if the data was accepted, increasing the chance of retries.

QUIC (HTTP/3?) is also on the horizon, which may make this even less useful.

And also the big thing - it leaks data between profiles, including (but not limited to) incognito profiles, since it relies on global system state.