hobohacker

* In order to get multiplexing and other features introduced with HTTP/2, you need to change protocol framing. However, this means that the protocol is no longer backwards compatible. There are many ways to roll out non-backwards-compatible changes, but for TCP, they were deemed unacceptable. For example, you could negotiate the protocol change via a TCP extension. However, TCP extensions are known not to be widely deployable (https://www.imperialviolet.org/binary/ecntest.pdf) over the public internet. You could use a specific port, but that doesn't traverse enough middleboxes on the public internet (http://www.ietf.org/mail-archive/web/tls/current/msg05593.ht...). Yada yada.

* More importantly, TCP is generally implemented in the OS. That means, updating a protocol requires OS updates. That means we need both server and client OS updates in order to speak the new protocol. If you look at the very sad numbers on Windows versions uptake, Android version uptake, etc, you'll understand why many people don't want to wait for all OSes to update in order to take advantage of new protocol features.

To quote Roberto (http://www.ietf.org/mail-archive/web/tsvwg/current/msg12184....) in his response to IETF transport protocol folks asking why application protocol folks are hacking around the transport:

""" I understand the points you make, and am sympathetic. I feel the same way when I see people abusing HTTP to provide async notifications, etc. The fact is, however, if it isn't deployed, no matter how nice it would theoretically be when it is, it isn't useful and people WILL work around the problem. That is true regardless of whether the problem is at the application layer (e.g. HTTP), or at the transport layer (e.g. TCP), or elsewhere. The primary motivation is to get things working, and making things that lack redundancy, or are elegant comes in at a distant second.

Deployed is the most important feature, thus things which quickly move protocols and protocol changes from theoretical to deployed are by far the most important things. The longer this takes, the more likely that the work-around becomes standard practice, at which point we've all "lost" the game. -=R """

Basically, this is another instance of Linus's quote (https://lkml.org/lkml/2009/3/25/632): "Theory and practice sometimes clash. And when that happens, theory loses. Every single time." Theoretically, it'd be far more elegant to fix this in the transport layer. But in practice, it doesn't work. Except, maybe if you implement on top of UDP (e.g. QUIC), since that would be in user-space and firewalls don't filter out UDP as much.

For more extensive analysis of HTTP/2's deployability options, amongst other considerations, check out my post on HTTP/2 considerations: https://insouciant.org/tech/http-slash-2-considerations-and-....

For more discussion on this topic, please refer to these discussion threads: https://groups.google.com/forum/#!topic/spdy-dev/_uvxTJkeCP0 https://groups.google.com/forum/#!topic/spdy-dev/XDudMZSq3e4

And note that your reference is slightly misleading (I don't think you intended this). It's true, the vast majority of hosts supporting SPDY are running nginx. In practice though, the vast majority of these hosts are Cloudflare or WordPress.com hosted sites. Both run newer versions of nginx with SPDY/3.1 support.

It's true, HTTPS is full of tradeoffs. You've identified some of them.

What do you see in HTTP/2 that "codifies this workaround"? That wasn't immediately obvious to me. Recall that HTTP/2 is basically just multiplexing with prioritized streams. There's no requirement on TLS in the spec, although all current browser deployments (of SPDY) require TLS.

[1]: http://en.wikipedia.org/wiki/Content_delivery_network

This specific proposal is interesting because it specifically is related to opportunistic encryption proposals, in particular, the one that allows sending http:// URIs over an unauthenticated TLS connection: http://tools.ietf.org/html/draft-nottingham-httpbis-alt-svc-.... The problem here for proxies is, if you mix http and https (authenticated) traffic on the same TLS connection, the proxy cannot tell if it can safely MITM the connection. The proxy vendor would like to know if it can do so, probably for network management / caching / content modification reasons. Of course, the point of the opportunistic encryption proposal is to increase security (although its actual effective impact is controversial: https://insouciant.org/tech/http-slash-2-considerations-and-...). But if you believe in opportunistic encryption's security purposes, then it doesn't seem to really make sense to make the MITM'able traffic identifiable so proxies on the network path can successfully MITM them without detection.

If someone can install a root cert onto your computer then you are already owned - there is no end to the other things they can do too. Call it a virus, call it an enterprise, but call it a day - you're owned and there is no in-charter policy this working group can enact to change the security level of that user for good or for bad..

The good news is not everyone is already owned and SSL helps those people today.

Now, as far as "SHOULD NOT", that's a reasonable thought for people not aware of what popular user agents currently do. The thing is, the majority of major browsers rewrite POST to GET on a 301. Here's my browser's code for it: https://code.google.com/p/chromium/codesearch#chromium/src/n.... Here's Firefox's code for it: http://mxr.mozilla.org/mozilla-central/source/netwerk/protoc.... To my knowledge, all browsers implement this behavior. We basically copied IE's behavior, because, IE did it and websites expected all user agents to do what IE did. Story of the web, sound familiar? :P

So, as you can see, Julian was merely acknowledging the pragmatic reality of the situation when he updated the httpbis specs to reflect this behavior: http://trac.tools.ietf.org/wg/httpbis/trac/changeset/1428. And this reasoning behind it is covered in the introduction to the relevant section in the httpbis docs: http://tools.ietf.org/html/draft-ietf-httpbis-p2-semantics-2....

"SHOULD NOT" implies that our implementations are behaving badly. Now, it's true, our implementations may not be behaving ideally from a spec cleanliness point of view, but interop trumps spec cleanliness, at least from the perspective of anyone who actually deploys real software on the internet. So it's probably best for the spec to acknowledge this and officially allow this. Specs that don't mirror reality are...probably not just useless, but actively harmful.

I hope that explains things, cheers.

- HTTP2 != httpbis. Both work is being done by the same working group "httpbis". http://datatracker.ietf.org/wg/httpbis/charter/ covers this. httpbis (http://stackoverflow.com/questions/9105639/httpbis-what-does...) was originally chartered to revise HTTP/1.1 (RFC2616) The working group will refine RFC2616 to: * Incorporate errata and updates (e.g., references, IANA registries, ABNF) * Fix editorial problems which have led to misunderstandings of the specification * Clarify conformance requirements * Remove known ambiguities where they affect interoperability * Clarify existing methods of extensibility * Remove or deprecate those features that are not widely implemented and also unduly affect interoperability * Where necessary, add implementation advice * Document the security properties of HTTP and its associated mechanisms (e.g., Basic and Digest authentication, cookies, TLS) for common applications

As for the HTTP/2 work, here's a snippet from the charter on that: The Working Group will produce a specification of a new expression of HTTP's current semantics in ordered, bi-directional streams. As with HTTP/1.x, the primary target transport is TCP, but it should be possible to use other transports.

- He seems to think the httpbis folks gratuitously redefined 301. It should be noted that RFC2616 (which, by definition, predates the httpbis work since httpbis is defined to revise RFC2616) had already noted the issue with 301 (http://tools.ietf.org/html/rfc2616#section-10.3.2): Note: When automatically redirecting a POST request after receiving a 301 status code, some existing HTTP/1.0 user agents will erroneously change it into a GET request.

- It's unclear to me whether or not the author acknowledges the existence of buggy implementations as noted in section 10.3.2. It's an open question as to what to do in the presence of buggy implementations. From a server standpoint, if the client is buggy, and you don't want to break the client (willingness to break clients probably depends on how many of the server's users use that client), then you will attempt to work around it, irrespective of what the standard says. Therefore, it's simply pragmatic to ignore the spec if it doesn't mirror reality, and pragmatic spec editors may update the spec to acknowledge this difference.

- As far as current status of the various 308 usages, Julian (author of the 308 draft) is lobbying major user agents to adopt this, and has written up a status update on the Chromium bug tracker: https://code.google.com/p/chromium/issues/detail?id=109012#c....