Remove `NotifyHandler::All` thus removing the requirement for events
send from a `NetworkBehaviour` to a `ProtocolsHandler` to be `Clone`. An
implementor of `NetworkBehaviour` can still notify all
`ProtocolHandler`s for a given peer by emitting one `NotifyHandler`
event per connection to that peer.
A user of libp2p-request-response is guaranteed to receive an additional event
after receiving a request via `RequestResponseEvent::Message`.
After receiving the request:
- If the user responds in time and the connection is still alive, the
user can expect a `ResponseSent`.
- If the user drops the response channel, the user can expect an
`InboundFailure::ResponseOmission`.
- If the user does not respond in time, the user can expect an
`InboundFailure::Timeout`.
Thus far the user did not receive an event when the connection to the
peer closes. With this commit:
- If the connection to the peer closes before the users calls
`send_response` or after the user calls `send_response` but before the
response can be send on the network, the user can expect an
`InboundFailure::ConnectionClosed`.
* [mdns] Split response packets.
Prevent MDNS response packets becoming too large by creating
multi-packet responses. Also skip addresses that don't fit
into a TXT record or contain invalid characters.
* Update protocols/mdns/src/dns.rs
Co-authored-by: Max Inden <mail@max-inden.de>
* Refactor response packet construction.
* Update mdns changelog.
Co-authored-by: Max Inden <mail@max-inden.de>
* core: Add rudimentary benchmark for PeerId::from_bytes and clone
* .github/workflow: Include benchmarks
To ensure changes through pull requests won't make benchmarks fail to
compile or run, run them as part of CI.
* Refine error reporting for inbound request handling.
At the moment one can neither get confirmation when a
response has been sent on the underlying transport, nor
is one aware of response omissions. The latter was
originally intended as a feature for support of
one-way protocols, which seems like a bad idea in
hindsight. The lack of notification for sent
responses may prohibit implementation of some
request-response protocols that need to ensure
a happens-before relation between sending a
response and a subsequent request, besides uses
for collecting statistics.
Even with these changes, there is no active notification
for failed inbound requests as a result of connections
unexpectedly closing, as is the case for outbound requests.
Instead, for pending inbound requests this scenario
can be identified if necessary by the absense of both
`InboundFailure` and `ResponseSent` events for a particular
previously received request. Interest in this situation is
not expected to be common and would otherwise require
explicitly tracking all inbound requests in the `RequestResponse`
behaviour, which would be a pity. `RequestResponse::send_response`
now also synchronously returns an error if the inbound upgrade
handling the request has been aborted, due to timeout or
closing of the connection, giving more options for graceful
error handling for inbound requests.
As an aside, the `Throttled` wrapper now no longer emits
inbound or outbound error events occurring in the context
of sending credit requests or responses. This is in addition
to not emitting `ResponseSent` events for ACK responses of
credit grants.
* Update protocols/request-response/src/lib.rs
Co-authored-by: Max Inden <mail@max-inden.de>
* Address some minor clippy warnings. (#1868)
* Track pending credit request IDs.
In order to avoid emitting events relating to credit grants or acks
on the public API. The public API should only emit events relating
to the actual requests and responses sent by client code.
* Small cleanup
* Cleanup
* Update versions and changelogs.
* Unreleased
Co-authored-by: Max Inden <mail@max-inden.de>
* [multistream-select] Listener conformity for failed negotiations.
When `V1Lazy` is used and the listener does not support the
optimistic (and singular) proposal of the dialer, it currently
happens that dialer and listener get a different outcome of
the negotiation. The dialer eventually detects the failed
negotiation as soon as it tries to read from the stream, but
the listener either encounters an invalid message or unexpected
premature EOF, depending on the payload that the dialer sent
prematurely after its protocol proposal. In these cases the
listener must be lenient and fail the negotiation "normally",
i.e. not with a protocol violation or an I/O error.
* Update misc/multistream-select/src/tests.rs
Co-authored-by: Max Inden <mail@max-inden.de>
* Refine error handling.
Only be lenient with garbage or sudden EOF when reading
just after having sent a protocol rejection.
* Update misc/multistream-select/src/listener_select.rs
Co-authored-by: Max Inden <mail@max-inden.de>
Co-authored-by: Max Inden <mail@max-inden.de>
Commit 335e55e6 removed the `ConnectionInfo` trait in favor of
`PeerId`s. Commit 1bd013c8 removed `ExpandedSwarm::connection_info` as
it would only return the `PeerId` that the caller is already aware of.
One could use `ExpandedSwarm::connection_info` not only to retrieve the
`ConnectionInfo` for a given peer, but also to check whether the
underlying `Network` has a connection to the peer.
This commit exposes the `is_connected` method on `Network` via
`ExpandedSwarm` to check whether the `Network` has an established
connection to a given peer.
* Make the lazy variant interoperable.
The remaining optimisation for `V1Lazy` for a listener
in the negotiation, whereby the listener delays flushing
of the multistream version header, is hereby removed.
The remaining effect of `V1Lazy` is only on the side of
the dialer, which delays flushing of its singular
protocol proposal in order to send it together with
the first application data (or an attempt is made to
read from the negotiated stream, which similarly
triggers a flush of the protocol proposal). This
permits `V1Lazy` dialers to be interoperable with
`V1` listeners. The remaining theoretical pitfall whereby
application data gets misinterpreted as another protocol
proposal by a listener remains, however unlikely.
`V1` remains the default, but we may eventually risk
just making this lazy dialer flush a part of the default
`V1` implementation, removing the dedicated `V1Lazy`
version identifier.
* Update CHANGELOG
* Separate versions from mere header lines.
Every multistream-select version maps to a specific header line,
but there may be different variants of the same multistream-select
version using the same header line, i.e. the same wire protocol.
* Cleanup
* Update misc/multistream-select/CHANGELOG.md
* Refactor and extend configurable connection limits.
To better track different connection counts, permit configurable
limits for these counts and make these available for
inspection efficiently, introduce dedicated connection counters
via a `ConnectionCounters` structure that is exposed on the
API via the `NetworkInfo`. All connection or connection
states that are counted in this way can also have effective
configurable limits.
* Cleanup
* Add missing file.
* Refine naming and config API.
* Update core/CHANGELOG.md
Co-authored-by: Max Inden <mail@max-inden.de>
* Update core/CHANGELOG.md
Co-authored-by: Max Inden <mail@max-inden.de>
Co-authored-by: Max Inden <mail@max-inden.de>
* Add "infinite" scores for external addresses.
Extend address scores with an infinite cardinal, permitting
addresses to be retained "forever" or until explicitly removed.
Expose (external) address scores on the API.
* Update swarm/src/registry.rs
Co-authored-by: Pierre Krieger <pierre.krieger1708@gmail.com>
* Fix compilation.
* Update CHANGELOG
Co-authored-by: Pierre Krieger <pierre.krieger1708@gmail.com>
* Fix ls response encoding/decoding.
Thereby remove the now unnecessary arbitrary protocol name
length limit. Since it an 'ls' response is always terminated
with a dedicated newline (and thus ends with two newlines),
an 'ls' response with a single protocol can be disambiguated
from a single protocol response by this additional newline.
* More commentary
* Update versions and changelogs.
* Resolve remaining conflict.
* Permit empty ls responses, as before.
* feat: upgrade to multihash 0.13
`multihash` changes a lot internally, it is using stack allocation instead
of heap allocation. This leads to a few limitations in regards on how
`Multihash` can be used.
Therefore `PeerId` is now using a `Bytes` internally so that only minimal
changes are needed.
* Update versions and changelogs.
Co-authored-by: Roman Borschel <romanb@users.noreply.github.com>
Co-authored-by: Roman S. Borschel <roman@parity.io>
* multiaddr: feature gate url dependency
Make it possible to exclude url and its dependencies
from the dependency tree, by feature gating the
from_url* functions in the public API.
* Document feature gating of multiaddr::from_url*
* [mplex] Benchmark different split_send_size configurations.
With both TCP and memory transports. As a result, change the
default `split_send_size` to 8KiB.
* Cleanup
* [mplex] Update CHANGELOG.
* Add throughput information.
Without a limit, one can send malformed input such that seq.size_hint() returns
a very large value and crashes the program even if the following data is pretty
small.
* Retain remaining read buffer.
When the plaintext protocol handshake finishes and
the `Framed` I/O is discarded in favour of the underlying
I/O stream, the remaining read buffer of `Framed` must
be retained, as it may have buffered data beyond the
end of the handshake.
* Update versions and changelogs.
