 |
This is probably unrelated to the TOK system and more a case of the client trying to parse garbage data as a packet. The current backpressure implementation in the server netcode unfortunately can truncate packets if the client isn't receiving fast enough (fort and other low FPS situations usually trigger this since the client receive code doesn't appear to be async) and this result in a desync of the bytestream. The client then starts interpreting the next bytes as best it can, which is why you sometimes see random survey popups, huge amounts of gold or XP / renown being received, etc. It's entirely possible one of those interpretations looks like a TOK update and triggers this failure. If the TOK packet endianness was wrong then any time a TOK is unlocked would be broken or show garbage, and this isn't the case. |
 |
That wasn't the case in this scenario as the ringbuffer from the crash dump wasn't showing garbled in the next frames, or would be insanely unlikely. On another note its entirely possible to increase the clientside netcode ringbuffer size so the buffer overruns can't occur like you're describing. You can do this easily in the LAA client. |
- Anonymous
