Support Board

It seems the sierra websocket is holding up. I think its good to release.

your timing of the message is interesting. I was just looking into see why it is we were having a problem previously.

I have confirmed, that when using the binary price packets, you are sending out corrupted websocket data at times which is causing the lost connection. I cannot see any other possibility. I have looked at this quite closely. Nothing changes on our side, at the websocket level when you are using the JSON format, the websocket processing remains identical. (Note: and as we said above, there could not possibly be an issue at the network I/O or encryption/decryption level/layer)

The only variance is that there is less data flowing but the only possible reason that could be an issue is that if there is a frame over 65K, there is a different block of code to determine the frame size, but it essentially works identically. But that is not what is happening. There are no frames ever coming in over that size, when we are getting corrupt data.

And with the subsequent processing of the binary price packet, that is a safe read-only processing . So it is not as though there is any potential for data corruption using the binary price packets causing this issue. If there is any potential for data corruption, it would be more likely with JSON but clearly that is not happening because now we are processing enormous amounts more of that using the compact JSON format. (Note: We already knew there is no corruption occurring with the JSON data processing either. This code is reliable and has never been a problem for years with various usage of JSON internally and also with external services)

We will force an update, over the weekend.