Proposed (Protocol Negotiation)
Note
This is NOT the UDT protocol reworking I have been working on. This is a significant change to the internal protocol contents below the UDT level that I hope to be working on once I clear the current project. The UDT protocol rewrite is likely required before work can begin on this. In the meantime, this is a proposal being written out to elicit some discussion. Any future work I do on this will likely be here: (GitHub branch)
Currently the contents of packets are a hand-optimized list of fields, either explicitly written out or lightly-encoded using QDataStream. Similar to use of protobuf these structures cannot change without also changing the protocol incompatibly, as they are not self-describing and have no means to encode or decode anything other than the exact expected string of bytes.
My intent here is to come up with a potential replacement that can be self-describing, allow new features to be introduced with or without both sides needing to recognize them, and do so in a compact manner. The amount of data will be less dense but I’m hoping that the flexibility will allow for both a more powerful protocol as well as potentially dropping unused optional fields.
I am considering migration to a protocol negotiation to be CRITICAL to the long-term viability of Vircadia as not just a single group of people surounding an authoritative response but as a heterogeneous group of people that may not all be on the same release (or fork). This permits further exploration and enhancement of features that affect the protocol without forcing a split to the community.
I will be using a form of SDXF ( https://en.wikipedia.org/wiki/SDXF ) modified to make it both more compact and able to handle larger values it otherwise would be able to handle.