Skip to content
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

chore(lightpush)!: move protocol implementation opinions to @waku/sdk #1887

Merged
merged 11 commits into from
Mar 11, 2024

Conversation

danisharora099
Copy link
Collaborator

@danisharora099 danisharora099 commented Mar 5, 2024

Problem

#1886

Solution

Moves away the abstraction of using multiple peers for lightpush by:

  • creating a new class LightPushSDK (and renaming protocol implementation to LightPushCore)
  • LightPushCore remains accessible from within LightPushSDK (it is an abstraction)
  • similarly introduces BaseProtocolSDK that houses information such as number of peers for the protocol
    • common between all SDK-type protocol classes (as BaseProtocol is common between all protocol implementation classes)

Notes

@danisharora099 danisharora099 changed the title chore!(lightpush): move protocol opinions to @waku/sdk chore(lightpush)!: move protocol opinions to @waku/sdk Mar 5, 2024
@danisharora099 danisharora099 changed the title chore(lightpush)!: move protocol opinions to @waku/sdk chore(lightpush)!: move protocol implementation opinions to @waku/sdk Mar 5, 2024
@danisharora099 danisharora099 force-pushed the chore/refactor-protocol-structure branch from cd79658 to 8dae746 Compare March 5, 2024 13:05
Copy link

github-actions bot commented Mar 5, 2024

size-limit report 📦

Path Size Loading time (3g) Running time (snapdragon) Total time
Waku node 185.23 KB (-0.01% 🔽) 3.8 s (-0.01% 🔽) 1.1 s (+37.38% 🔺) 4.8 s
Waku Simple Light Node 185.28 KB (+0.07% 🔺) 3.8 s (+0.07% 🔺) 1.1 s (-26.32% 🔽) 4.8 s
ECIES encryption 22.89 KB (0%) 458 ms (0%) 480 ms (+91.22% 🔺) 938 ms
Symmetric encryption 22.34 KB (0%) 447 ms (0%) 173 ms (-59.75% 🔽) 619 ms
DNS discovery 69.95 KB (0%) 1.4 s (0%) 716 ms (-0.35% 🔽) 2.2 s
Privacy preserving protocols 39.96 KB (+0.29% 🔺) 800 ms (+0.29% 🔺) 612 ms (+62.16% 🔺) 1.5 s
Light protocols 0 B (-100% 🔽) 0 ms (-100% 🔽) 0 ms (-100% 🔽) 0 ms
History retrieval protocols 19.34 KB (-0.04% 🔽) 387 ms (-0.04% 🔽) 353 ms (+46.42% 🔺) 740 ms
Deterministic Message Hashing 4.96 KB (0%) 100 ms (0%) 31 ms (-48.69% 🔽) 130 ms
Waku Filter 20.11 KB (+100% 🔺) 403 ms (+100% 🔺) 298 ms (+100% 🔺) 700 ms
Waku LightPush 115.49 KB (+100% 🔺) 2.4 s (+100% 🔺) 937 ms (+100% 🔺) 3.3 s

/**
* A class with predefined helpers, to be used as a base to implement Waku
* Protocols.
*/
export class BaseProtocol implements IBaseProtocol {
export class BaseProtocol implements IBaseProtocolCore {
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Why these rename? I think we can keep previous names, no need to add core as it is already clear from the package name.

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I refactor these names as we now have two types of BaseProtocol:

  • one on the protocol implementation level (in @waku/core)
  • other on the SDK level (in @waku/sdk)

I personally incline towards having more verbose & readable names.

@@ -340,6 +340,8 @@ class Subscription {
}
}

const DEFAULT_NUM_PEERS = 3;
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I know it is not a part of the PR but how is it enforced? It seems we do have this as default

const DEFAULT_NODE_REQUIREMENTS = {

Copy link
Collaborator Author

@danisharora099 danisharora099 Mar 8, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

These variables are for two separate uses:

  • DEFAULT_NODE_REQUIREMENTS: the default configuration of number of peers for each protocol that we want to discovery using DNS discovery
  • DEFAULT_NUM_PEERS: the number of peers to use for the protocol request (redundancy to increase decentralization)

It is used here:

/**
* Number of peers to connect to, for the usage of the protocol.
* This is used by:
* - Light Push to send messages,
* - Filter to retrieve messages.
* Defaults to 3.
*/
numPeersToUse?: number;

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

but then it seems not to be used and fall back to DEFAULT_NODE_REQUIREMENTS, am I missing something?

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I'm not sure what you mean: the above two constants are for two very different purposes as described.

numPeersToUse falls back to DEFAULT_NUM_PEERS if not provided part of the options argument

@@ -357,6 +359,9 @@ class Filter extends BaseProtocol implements IReceiver {
return subscription;
}

//TODO: Remove when FilterCore and FilterSDK are introduced
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Maybe FitlerRFC and Filter or FilterCore and Filter? I think one can be left clean or renamed entirely as it is something else than Filter in the meaning of protocol.

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I'd like to avoid this kind of names in general as it is kind of duplication when we have FilterX and FilterY or SendResult and CoreSendResult which just brings difficulty to understand what to use under which conditions.

Ideally we don't want anyone use core protocols except when people know what they are doing. In that case for end user I'd like to keep out API the same: LightPush/Filter or create new entities like Sender/Receiver. That way proposition of implementation is clear.

wdyt?

Copy link
Collaborator Author

@danisharora099 danisharora099 Mar 11, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Maybe FitlerRFC and Filter or FilterCore and Filter?

I personally incline towards a more verbose naming convention, but I'm happy to change this to naming like Filter and FilterCore if you prefer

I think one can be left clean or renamed entirely as it is something else than Filter in the meaning of protocol.

Can you elaborate? As I see it, they are both functionalities for Filter: one provides us with an implementation to interact with the protocol (core), and the other is an implementation that offers opinions by us on how you can (better) use it (sdk)

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I personally incline towards a more verbose naming convention

I think move simpler names make it easier to comprehend and comply with RFC and other implementations.
For example - FilterRFC is visible that follows RFC and Filter being exposed at the root level of node makes it go-to choice with it's API being documented.

I think one can be left clean or renamed entirely as it is something else than Filter in the meaning of protocol.

here I mean that if we have FilterRFC being implementation of RFC - then Filter with it's proxy capabilities might as well be named something else, as it is not Filter in sense of Waku terminology but convenient API we provide

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

RFC stands for Request For Comments - I'm not sure if the class called FilterRFC gives the right messaging.
We could, however, keep FilterCore as Filter and FilterSDK could remain as is.

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Since this was merged, @weboko do we need to followup here? Can be added as a separate task that can be tackled in #1886

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

let's not add any subtasks for now - as we don't know what we are optimizing for

@danisharora099 danisharora099 force-pushed the chore/refactor-protocol-structure branch from c66ad58 to 299b5d9 Compare March 8, 2024 11:08
@danisharora099 danisharora099 merged commit 8deab11 into master Mar 11, 2024
9 of 10 checks passed
@danisharora099 danisharora099 deleted the chore/refactor-protocol-structure branch March 11, 2024 13:20
@weboko weboko mentioned this pull request Mar 11, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

None yet

3 participants