Skip to main content

A pure-Python implementation of the HTTP/2 priority tree

Project description

Priority is a pure-Python implementation of the priority logic for HTTP/2, set out in RFC 7540 Section 5.3 (Stream Priority). This logic allows for clients to express a preference for how the server allocates its (limited) resources to the many outstanding HTTP requests that may be running over a single HTTP/2 connection.

While priority information in HTTP/2 is only a suggestion, rather than an enforceable constraint, where possible servers should respect the priority requests of their clients.

Using Priority

Priority has a simple API. Streams are inserted into the tree: when they are inserted, they may optionally have a weight, depend on another stream, or become an exclusive dependent of another stream.

>>> p = priority.PriorityTree()
>>> p.insert_stream(stream_id=1)
>>> p.insert_stream(stream_id=3)
>>> p.insert_stream(stream_id=5, depends_on=1)
>>> p.insert_stream(stream_id=7, weight=32)
>>> p.insert_stream(stream_id=9, depends_on=7, weight=8)
>>> p.insert_stream(stream_id=11, depends_on=7, exclusive=True)

Once streams are inserted, the stream priorities can be requested. This allows the server to make decisions about how to allocate resources.

Querying The Tree: Manual Walk

Priority supports a number of methods of querying the priority tree. One way is to manually walk the tree, starting from the level of highest priority.

>>> tree = p.priorities()
>>> tree
Priorities<total_weight=64, streams=[Stream<id=1, weight=16>, Stream<id=3, weight=16>, Stream<id=7, weight=32]>,

Currently, these streams are all the highest priority: they do not depend on another stream, and so should be served first if resources are available. Of these three streams, stream 7 should receive 1/2 of the resources available, with streams 1 and 3 receiving 1/4 of the resources available each.

If for any reason a stream is blocked or cannot be served (e.g. because you are waiting on I/O), you can allocate the resources that were intended for that stream to its dependents. To do this, index into the Priorities by stream ID. For example, if you were unable to serve stream 7:

>>> dependents = tree[7]
>>> dependents
Priorities<total_weight=16, streams=[Stream<id=11, weight=16>]>
>>> second_level_dependents = dependents[11]
>>> second_level_dependents
Priorities<total_weight=8, streams=[Stream<id=9, weight=8>]>

Here, if stream 7 could not be served, all its resources should be diverted to serving stream 11. If that were also unable to be served, we should instead divert all the resources to stream 11.

Querying The Tree: Automated Resolution

An alternative approach is to inform Priority of what streams are blocked (or, alternatively, which are unblocked), in which case Priority will return a list of the streams that should be satisfied, and their weights.

>>> priorities = p.priorities(blocked=[1, 5, 7])
>>> priorities
Priorities<total_weight=48, streams=[Stream<id=3, weight=16>, Stream<id=11, weight=32>]>
>>> p.priorities(blocked=[1, 5, 7]) == p.priorities(unblocked=[3, 9, 11])

Querying The Tree: Gate

The Priority ‘gate’ is an alternative approach of accessing HTTP/2 priorities based on this presentation. It works by providing an iterator that acts as a ‘gate’. Each value popped off the iterator is the next stream that should be acted on. The PriorityTree can be mutated on the fly to add, remove, block, and unblock streams, which affects what the next value from the iterator will be.

This works well when combined with some kind of signaling mechanism to block and unblock threads of execution. This would allow you to do something like this:

>>> for stream_id in p.gate():
...     unblock(stream_id)  # Sends a single DATA frame.

In this circumstance, each time a stream is unblocked it should send a single DATA frame and then go back to waiting to be unblocked. This way we ensure that the data from streams is correctly multiplexed.

You can also block and unblock streams in the iterator, like so:

For example:

>>> for stream_id in p.gate()
...    now_blocked = unblock(stream_id)
...    if now_blocked:
...        p.blocked(stream_id)
...    unblocked = all_unblocked_streams()
...    for unblocked_stream_id in unblocked:
...        p.unblock(unblocked_stream_id)


Priority is made available under the MIT License. For more details, see the LICENSE file in the repository.


Priority is maintained by Cory Benfield, with contributions from others. For more details about the contributors, please see CONTRIBUTORS.rst in the repository.

Project details

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page