-
Notifications
You must be signed in to change notification settings - Fork 964
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
{core/,swarm/}: Dial with handler and return handler on error and closed #2191
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Very nice idea. I think it is a good direction.
@@ -65,6 +65,11 @@ where | |||
self.substream_upgrade_protocol_override = version; | |||
self | |||
} | |||
|
|||
// TODO: Rethink this method. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I've always found the indirection between ConnectionHandler
and ProtocolsHandler
a bit annoying. I understand that it is necessary because libp2p-core
can't know about the swarm abstraction. I do wonder though on whether or not libp2p-core
couldn't provide the ProtocolsHandler
abstraction right away instead of us having to effectively re-define almost the same interface.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
That is an option worth exploring. Though I don't think we should do so in this pull request to separate concerns. Note that moving ProtocolsHandler
into libp2p-core
would require moving concepts like KeepAlive
into libp2p-core
as well.
This logic seems to be a leftover of libp2p#889 and unused today.
pub enum NetworkBehaviourAction< | ||
TOutEvent, | ||
THandler: IntoProtocolsHandler, | ||
TInEvent = THandlerInEvent<THandler>, | ||
> { |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Similar to THandlerInEvent
, I think we could define NetworkBehaviourAction
in terms of the NetworkBehaviour
trait. That would reduce the amount of type parameters down to one.
It would also make it easier to refer to because you can just say NetworkBehaviourAction<Self>
within the NetworkBehaviour
.
Thoughts?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Ah we would have to give up our map_XYZ
functions then.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Good idea @thomaseizinger which I started to explore. See mxinden#5
Ah we would have to give up our map_XYZ functions then.
Not as far as I can tell with additional trait parameters (each defaulting to a derivative of TBehaviour
). See PR above.
Unfortunately I don't see a way to use TBehaviour
in NetworkBehaviourAction
. Unless I do so, compilation fails.
Let's move the discussion to mxinden#5.
Reviews of all kinds are very much appreciated! (//CC @thomaseizinger in case you have time and interest.) |
swarm/src/behaviour.rs
Outdated
/// (ie. the objects returned by `new_handler`) can communicate by passing messages. Messages | ||
/// sent from the handler to the behaviour are injected with [`NetworkBehaviour::inject_event`], | ||
/// and the behaviour can send a message to the handler by making [`NetworkBehaviour::poll`] | ||
/// return [`NetworkBehaviourAction::SendEvent`]. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Did you rename it? Otherwise I think this is actually called NotifyHandler
:)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
👍 It was previously SendEvent
, I changed it to ['NetworkBehaviour::SendEvent']
. 4d0faf9 changes it to ['NetworkBehaviour::NotifyHandler']
.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Nice work, thanks for taking this on!
Left a few nits but from what I can see, this should work :)
Maybe a test or an example within libp2p-swarm
that demonstrates this usage could be useful? I am talking about the actual usecase of carrying a message within the handler and reporting back to the user in case it failed or sending it directly once the connection succeeds.
handler: THandler, | ||
}, | ||
/// The dialing attempt is rejected because the peer being dialed is the local peer. | ||
LocalPeerId { handler: THandler }, |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Maybe SelfDial
is a more descriptive name?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I want to stress the fact that the peer ids match. This would e.g. not work by matching multiaddresses. All that said, I don't think LocalPeerId
is the perfect name either.
core/src/network/event.rs
Outdated
match attempts_remaining { | ||
DialAttemptsRemaining::Some(attempts) => (*attempts).into(), | ||
DialAttemptsRemaining::None(_) => 0, | ||
} |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I think this could be written like this at which point it is questionable that we need the From
implementation?
Generally, wouldn't this be cleaner to be a regular function within an impl DialAttemptsRemaining
?
match attempts_remaining { | |
DialAttemptsRemaining::Some(attempts) => (*attempts).into(), | |
DialAttemptsRemaining::None(_) => 0, | |
} | |
attempts_remaining.map(|attempts| attempts.get()).unwrap_or(0) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I don't see a way to do the above, given that the close of map_err
would take ownership of handler, which is later on used. I don't think rustc is able to detect the early return.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I guess this comment should go somewhere else? Or is this a GitHub glitch?
@@ -747,7 +747,7 @@ mod tests { | |||
|
|||
// check that our subscriptions are sent to each of the peers | |||
// collect all the SendEvents | |||
let send_events: Vec<&NetworkBehaviourAction<Arc<GossipsubHandlerIn>, GossipsubEvent>> = gs | |||
let send_events: Vec<_> = gs |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Nice!
Yep. Good idea. I added a doc example, which thus both showcases the use-case and is continuously tested. (814ff4b) Also note that the DCUtR implementation builds on top of this patch already: #2076 |
swarm/src/lib.rs
Outdated
|
||
/// [`NetworkBehaviourAction::DialAddress`] and [`NetworkBehaviourAction::DialPeer`] require a | ||
/// handler. This handler can be used to carry state. See corresponding doc comments. | ||
#[test] | ||
fn use_handler_to_carry_state() {} |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Is that a left-over from adding a dedicated test but then resorting to a doc-test?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Ups. Thanks for the pointer. Removed in 60c7261.
Co-authored-by: Thomas Eizinger <[email protected]>
Unless there are any objections or further suggestions, I will merge here sometime this week. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM
Co-authored-by: Thomas Eizinger <[email protected]>
* protocols/gossipsub: Fix inconsistency in mesh peer tracking (libp2p#2189) Co-authored-by: Age Manning <[email protected]> * misc/metrics: Add auxiliary crate to record events as OpenMetrics (libp2p#2063) This commit adds an auxiliary crate recording protocol and Swarm events and exposing them as metrics in the OpenMetrics format. * README: Mention [email protected] * examples/: Add file sharing example (libp2p#2186) Basic file sharing application with peers either providing or locating and getting files by name. While obviously showcasing how to build a basic file sharing application, the actual goal of this example is **to show how to integrate rust-libp2p into a larger application**. Architectural properties - Clean clonable async/await interface ([`Client`]) to interact with the network layer. - Single task driving the network layer, no locks required. * examples/README: Give an overview over the many examples (libp2p#2194) * protocols/kad: Enable filtering of (provider) records (libp2p#2163) Introduce `KademliaStoreInserts` option, which allows to filter records. Co-authored-by: Max Inden <[email protected]> * swarm/src/protocols_handler: Impl ProtocolsHandler on either::Either (libp2p#2192) Implement ProtocolsHandler on either::Either representing either of two ProtocolsHandler implementations. Co-authored-by: Thomas Eizinger <[email protected]> * *: Make libp2p-core default features optional (libp2p#2181) Co-authored-by: Max Inden <[email protected]> * core/: Remove DisconnectedPeer::set_connected and Pool::add (libp2p#2195) This logic seems to be a leftover of libp2p#889 and unused today. * protocols/gossipsub: Use ed25519 in tests (libp2p#2197) With f2905c0 the secp256k1 feature is disabled by default. Instead of enabling it in the dev-dependency, simply use ed25519. * build(deps): Update minicbor requirement from 0.10 to 0.11 (libp2p#2200) Updates the requirements on [minicbor](https://gitlab.com/twittner/minicbor) to permit the latest version. - [Release notes](https://gitlab.com/twittner/minicbor/tags) - [Changelog](https://gitlab.com/twittner/minicbor/blob/master/CHANGELOG.md) - [Commits](https://gitlab.com/twittner/minicbor/compare/minicbor-v0.10.0...minicbor-v0.11.0) --- updated-dependencies: - dependency-name: minicbor dependency-type: direct:production ... Signed-off-by: dependabot[bot] <[email protected]> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> * build(deps): Update salsa20 requirement from 0.8 to 0.9 (libp2p#2206) * build(deps): Update salsa20 requirement from 0.8 to 0.9 Updates the requirements on [salsa20](https://github.com/RustCrypto/stream-ciphers) to permit the latest version. - [Release notes](https://github.com/RustCrypto/stream-ciphers/releases) - [Commits](RustCrypto/stream-ciphers@ctr-v0.8.0...salsa20-v0.9.0) --- updated-dependencies: - dependency-name: salsa20 dependency-type: direct:production ... Signed-off-by: dependabot[bot] <[email protected]> * *: Bump pnet to v0.22 Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> Co-authored-by: Max Inden <[email protected]> * *: Dial with handler and return handler on error and closed (libp2p#2191) Require `NetworkBehaviourAction::{DialPeer,DialAddress}` to contain a `ProtocolsHandler`. This allows a behaviour to attach custom state to its handler. The behaviour would no longer need to track this state separately during connection establishment, thus reducing state required in a behaviour. E.g. in the case of `libp2p-kad` the behaviour can include a `GetRecord` request in its handler, or e.g. in the case of `libp2p-request-response` the behaviour can include the first request in the handler. Return `ProtocolsHandler` on connection error and close. This allows a behaviour to extract its custom state previously included in the handler on connection failure and connection closing. E.g. in the case of `libp2p-kad` the behaviour could extract the attached `GetRecord` from the handler of the failed connection and then start another connection attempt with a new handler with the same `GetRecord` or bubble up an error to the user. Co-authored-by: Thomas Eizinger <[email protected]> * core/: Remove deprecated read/write functions (libp2p#2213) Co-authored-by: Max Inden <[email protected]> * protocols/ping: Revise naming of symbols (libp2p#2215) Co-authored-by: Max Inden <[email protected]> * protocols/rendezvous: Implement protocol (libp2p#2107) Implement the libp2p rendezvous protocol. > A lightweight mechanism for generalized peer discovery. It can be used for bootstrap purposes, real time peer discovery, application specific routing, and so on. Co-authored-by: rishflab <[email protected]> Co-authored-by: Daniel Karzel <[email protected]> * core/src/network/event.rs: Fix typo (libp2p#2218) * protocols/mdns: Do not fire all timers at the same time. (libp2p#2212) Co-authored-by: Max Inden <[email protected]> * misc/metrics/src/kad: Set query_duration lowest bucket to 0.1 sec (libp2p#2219) Probability for a Kademlia query to return in less than 100 milliseconds is low, thus increasing the lower bucket to improve accuracy within the higher ranges. * misc/metrics/src/swarm: Expose role on connections_closed (libp2p#2220) Expose whether closed connection was a Dialer or Listener. * .github/workflows/ci.yml: Use clang 11 (libp2p#2233) * protocols/rendezvous: Update prost (libp2p#2226) Co-authored-by: Max Inden <[email protected]> * *: Fix clippy warnings (libp2p#2227) * swarm-derive/: Make event_process = false the default (libp2p#2214) Co-authored-by: Max Inden <[email protected]> Co-authored-by: Max Inden <[email protected]> Co-authored-by: Age Manning <[email protected]> Co-authored-by: Ruben De Smet <[email protected]> Co-authored-by: Thomas Eizinger <[email protected]> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> Co-authored-by: rishflab <[email protected]> Co-authored-by: Daniel Karzel <[email protected]> Co-authored-by: David Craven <[email protected]>
Fix incoming dial failure handling for connection pool behaviour. Since the recent update of libp2p, [this change](libp2p/rust-libp2p#2191) introduces state to dial request that can be transmitted in `inject_dial_failure`. So there are a couple of cases where the function is called but we can ignore the error since the connection is still to be completed or is already connected.
Fix incoming dial failure handling for connection pool behaviour. Since the recent update of libp2p, [this change](libp2p/rust-libp2p#2191) introduces state to dial request that can be transmitted in `inject_dial_failure`. So there are a couple of cases where the function is called but we can ignore the error since the connection is still to be completed or is already connected.
Fix incoming dial failure handling for connection pool behaviour. Since the recent update of libp2p, [this change](libp2p/rust-libp2p#2191) introduces state to dial request that can be transmitted in `inject_dial_failure`. So there are a couple of cases where the function is called but we can ignore the error since the connection is still to be completed or is already connected. This fixes #447.
Fix incoming dial failure handling for connection pool behavior. Since the recent update of libp2p, [this change](libp2p/rust-libp2p#2191) introduces state to dial request that can be transmitted in `inject_dial_failure`. So there are a couple of cases where the function is called but we can ignore the error since the connection is still to be completed or is already connected. This fixes #447.
Fix incoming dial failure handling for connection pool behavior. Since the recent update of libp2p, [this change](libp2p/rust-libp2p#2191) introduces state to dial request that can be transmitted in `inject_dial_failure`. So there are a couple of cases where the function is called but we can ignore the error since the connection is still to be completed or is already connected. This fixes #447.
Background
As promised in #2118 (reply in thread) I wanted to trial adding something similar to
ProtocolsHandler::OutboundOpenInfo
for connections instead of substreams toNetworkBehaviourAction::{DialPeer,DialAddress}
.#2118 (comment) details the benefits. Long story short, it allows one to attach state to a dial request, thus not having to keep track of it within a
NetworkBehaviour
implementation.While trialing the above, I had the ideas for #2182 and #2183, both simplifying trait parameters, one in
libp2p-swarm
and one inlibp2p-core
. Passing an opaqueOpenInfo
type along withNetworkBehaviourAction::{DialPeer,DialAddress}
would require yet another trait parameter. To circumvent this additional complexity, I came up with the below, in other words with the change proposed in this pull request.Proposal
Require
NetworkBehaviourAction::{DialPeer,DialAddress}
to contain aProtocolsHandler
(see diff below). This allows a behaviour to attach custom state to its handler. The behaviour would no longer need to track this state separately during connection establishment, thus reducing state required in a behaviour. E.g. in the case oflibp2p-kad
the behaviour can include aGetRecord
request in its handler, or e.g. in the case oflibp2p-request-response
the behaviour can include the first request in the handler.Diff
Return
ProtocolsHandler
on connection error and close (see diff below). This allows a behaviour to extract its custom state previously included in the handler on connection failure and connection closing. E.g. in the case oflibp2p-kad
the behaviour could extract the attachedGetRecord
from the handler of the failed connection and then start another connection attempt with a new handler with the sameGetRecord
or bubble up an error to the user.Diff
What do people think? Do you see additional use-cases this would support or do you see use-cases that would not be but should be supported? Especially interested n @thomaseizinger opinion given your initial proposal on #2118.