About Engine IO V3 Clients compability

[tenmo2003]

Hi, this repository seems to be a descendant of https://github.com/mrniko/netty-socketio. I want to ask about a specific commit 5e56faa. This commit removed the Engine IO's packet type byte strip in PacketDecoder, causing the payload received in user-end has an additional byte (e.g. '4') in front. Is there any reason behind this because it can easily be fixed with EIO version check stripping? I don't really want to have to manually check and strip this byte

[sanjomo]

@NeatGuyCoding can you please review this one?

[sanjomo]

@tenmo2003 is not this also done in the parent project? what exactly the problem you are facing due to this? can you add more details with minimum reproducible example ?

[sanjomo]

@tenmo2003 Hi! Are you currently using any Engine.IO v3 clients? If so, could you share the project or a minimal example?

From what I've checked, most of the v3 compatibility has already been removed from the parent project, and only v4 appears to be working. Given that, do we really want to keep partial v3 support?

My preference is to support only Engine.IO v4 going forward. That keeps the implementation simpler and avoids carrying legacy compatibility that's no longer maintained.

[tenmo2003]

I have some projects running with BestHTTP 1.12.5, the server's library upgrade can be seamless if it supports both EngineIO version.

I don't know why the v3 compatibility was removed from the parent project as it states 1.x - 4.x version of [Socket.IO-client](https://github.com/socketio/socket.io) is supported in the README.

And as I've read through the specifications and changes between the EIO versions, I don't see many changes so I think that keeping the backward compatibility is not too much of a burden. Nonetheless, I appreciate and respect the open source maintenance, if you guys find it may cause tech debt, I can just keep my own fork. Thanks

[sanjomo]

@tenmo2003 I checked some basic operations using Postman. The CONNECT request works, and ping/pong is functioning correctly. However, message send/receive isn't working as expected, even though it was working fine with v4.

If we're going to support v3, we should make sure all functionality works correctly before committing to it. I'll create integration tests for all v3 operations using the JavaScript client to verify everything. In the meantime, feel free to fork the branch or open a PR here if you'd like to contribute.

[tenmo2003]

That's weird, I tested with my unity client and it worked fine. I also conducted another test with Hoppscotch client on Websocket and the server received the message successfully.

[sanjomo]

@tenmo2003 Yes, I've gone through the v3 docs as well. The changes are relatively minor. The message issue turned out to be caused by an error in our test script, and it's working correctly now.

There are still a few areas that require more rigorous testing. We're looking beyond a straightforward fix and validating different scenarios, such as mixed v3/v4 client environments and other compatibility cases. We'll update you once we're confident that everything has been thoroughly tested and we're satisfied with the fix.