Doing HTTP Pipelining with libcurl
==================================

Background

Since pipelining implies that one or more requests are sent to a server before
the previous response(s) have been received, it cannot be implemented easily
into libcurl's easy interface due to its synchronous nature. We therefore only
aim on adding it for multi interface use.

Considerations

When using the multi interface, you create one easy handle for each transfer.
Bascially any number of handles can be created, added and used with the multi
interface - simultaneously. It is an interface designed to allow many
simultaneous transfers while still using a single thread.

Pipelining however, will force us to allow apps to somehow "connect" two (or
more) easy handles that are added to a multi handle. The first one sends a
request and receives a response, just as normal, while the second (and
subsequent) ones need to be attached to the first handle so that it can send
its request on the same connection and then sit and wait until its response
comes.

API

We add a new option to curl_multi_setopt() called CURLMOPT_PIPELINING that
enables "attempted pipelining" and then all easy handles used on that handle
will attempt to use an existing pipeline.

Decisions Already Made

- A pipeline is only created if a previous connection exists to the same IP
  address that the new request is being made to use.

- Pipelines are only supported for HTTP(S) as no other currently supported
  protocol has features resemembling this, but we still name this feature
  plain 'pipelining' to possibly one day support it for other protocols as
  well.

- HTTP Pipelining is for GET and HEAD requests only.

- When a pipeline is in use, we must take precautions so that when used easy
  handles (i.e those who still wait for a response) are removed from the multi
  handle, we must deal with the outstanding response nicely.

- Explicitly asking for pipelining handle X and handle Y won't be supported.
  It isn't easy for an app to do this association. The lib should probably
  still resolve the second one properly to make sure that they actually _can_
  be considered for pipelining. Also, asking for explicit pipelining on handle
  X may be tricky when handle X get a closed connection.

To Ponder About

- We need options to control max pipeline length, and probably how to behave
  if we reach that limit. As was discussed on the list, it can probably be
  made very complicated, so perhaps we can think of a way to pass all
  variables involved to a callback and let the application decide how to act
  in specific situations. Either way, these fancy options are only interesting
  to work on when everything is working and we have working apps to test with.

- Currently (before pipelining) we do not have any code or concept that lets
  multiple handles share the same physical connection. We need to carefully
  make sure that each easy handle knows exactly what they can do and when, on
  the shared connection.

- We need to keep a linked list of each handle that is part of a single pipe
  so that if it breaks, we know which handles that need to resend their
  requests. The pipe linked-lists could very well be "held" in the multi
  handle struct so that they won't "belong" to a particular easy handle that
  happens to be part of the pipeline during a certain period.