Socket.io VS SocketCluster

5. Socket.io

Socket.io: Real-time communication used by the application to provide live update of some feature and settings.

Websockets in Node.js There are various libraries that let you create a ws server:- https://www.npmjs.com/package/websocket https://github.com/websockets/ws https://socket.io/

For the socket integration I use https://socket.io/ and follow their integration guide about nextjs ( https://socket.io/how-to/use-with-nextjs)

Wasp has lots of time-saving features, including WebSocket support via Socket.IO, Authentication, Database Management, and Full-stack type-safety out-of-the box.

I don’t like to use sockets on nextjs, just looking at the integration page of socket.io (How to use with Next.js | Socket.IO) feel so weird for me, also combining the fact that the developer server is just a nightmare, I prefer to avoid everything related to websockets and nextjs at least for the moment.

Repository: Socket.io

When developing web applications, you might encounter connectivity issues between your client and server when using Socket.io on localhost.

I never fully understood the need for back end message queues TBH. You can just poll the database or data store every few seconds and process tasks in batches... IMO, the 'real time' aspect was only ever useful for front end use cases for performance reasons since short polling every second with HTTP (with all its headers/overheads) is prohibitively expensive. Also, HTTP long polling introduces some architectural complexity which is not worth it (e.g. sticky sessions are required when you have multiple app servers).

Unfortunately, real-time messaging complexity entirely to the back end has been the norm for a very long time. My experience is that, in general, it makes the architecture way more difficult to manage. I've been promoting end-to-end pub/sub as an alternative for over a decade (see https://socketcluster.io/) but it mostly fell on deaf ears. Most devs just don't realize how much complexity is added by micromanaging pub/sub channels on the back end and figuring out which message belongs to what user instead of letting end-users decide what channels to subscribe to directly.

I think part of the problem was the separation between front end and back end developer responsibilities. Back end developers like to ignore the front end as much as possible; when it comes to architecture, their thinking rarely extends beyond the API endpoint; gains which can be made from better integrating the back end with the front end is 'not their job'. From the perspective of front-end developers, they see anything performance-related as 'not their job' either... There aren't enough full stack developers with the required insights to push for integration efficiency/simplicity.

Recently, I added an AI-generated soundtrack to my open source project's home page https://socketcluster.io/

It seems unconventional at first but I distinctly remember about a decade ago when Adobe Flash was still broadly supported, many Flash websites had soundtracks. I think the reason why regular HTML websites didn't have them was because it was difficult to implement and internet was much slower so they had to be streamed in a special way.

If redis doesn't satisfy your requirements or you're unable to make it work using adaptor, SocketCluster is a great package for this https://socketcluster.io/

Interesting how this feature set is pretty much exactly the same as offered by SocketCluster https://socketcluster.io/

This resonates with me 100%. Every bit of unhappiness I've felt in my career so far has been caused by a bad manager. The main issue for me has been the final point you mentioned about "Imposed artificial limitations" - I cannot tell you the number of times that I've been forced to use an inefficient tool or do something in a sub-optimal or downright incorrect way (knowing that it would have to be re-written later) by a bad manager... In some companies, it was a daily occurrence; that's why I never stayed at a single company for longer than 2 years. It's almost impossible to find a company that lets me implement things correctly.

Thankfully, nobody could constrain me in my open source work. I (with the help of community members) built:

- SocketCluster (https://socketcluster.io/): A distributed pub/sub framework.

- Capitalisk (https://capitalisk.com/): A lightweight quantum-resistant blockchain which is less than 5K lines of code.

- LDEX (https://ldex.trading/): A deterministic decentralized exchange (DEX) which can work with many different blockchain protocols. It's less than 4K lines of code in total and only has 3 small third-party dependencies (including sub-dependencies).

The two best lightweight solutions I have worked with are SocketCluster and NATS. SocketCluster is used to build out a custom backend that scales very well. NATS would be used as a central pub/sub system. If you need queueing, you'll need to add additional services like Redis.

In the homepage for SocketCluster it is mentioned that it is a Highly scalable pub/sub and RPC framework optimized for async/await. However in the documentation I have not found anything related to RPC. Am I interpreting anything wrong here?

At the time of writing, SocketCluster has almost 6k stars on GitHub and 7k downloads on npm weekly,

SocketCluster is a framework which allows you to use the WebSocket protocol the transmit between its backend API and client library. The client can be used both on a backend (E.g. a Raspberry pi) or frontend application.