Release track: Skytable Octave (0.8)
ohsayan opened this issue · comments
After several months of behind-the-scenes thinking, discussions, implementations, experimentation, and testing — we are finally just a few steps away from releasing Skytable Octave. Yes, that release that you've been hearing about forever but couldn't get your hands on!
Before everything else, I must apologize for how long it has taken. As I kept mentioning, Skytable Octave is not like a rewrite of Skytable — it has almost been like writing Skytable from scratch. Everything, starting from the query interface and engines, to the in-memory engines, to the storage engine, the network layer — every piece is new. I've spent countless hours working on this, and I hope you love it!
That being said, here's a brief timeline of where we are right now (just another few boxes to check):
- New query engine
- Models and spaces in semantics
- Remove actions (notes: we'll see what we want to do here. but for now, actions are completely removed)
- New memory engine
- Stress test memory engine (notes: still need to optimize memory usage; in later PR)
- New storage engine
- Stress test storage engine
- Final Skyhash 2 spec (publish to docs)
- Small tasks (which will take a day or two but nevertheless are important):
- PR 1: Merge new engines into
next
- Backwards compatibility
- Document the new query interface
- Document the new protocol and help the community upgrade
- Ship 🚀🚀🚀
- PR 1: Merge new engines into
For those asking, development was done in private repos (for convenience with testing and to keep it a secret!) and will be merged into this repo with a public PR.
good news, By the way, does the new version depend on the VC runtime?
good news, By the way, does the new version depend on the VC runtime?
Hmm, by VC runtime do you mean Visual C++? In that case, we haven't ever depended on VC in any way (unless you're building Skytable on Windows from source, in which case rustup
will make you install a bunch of things (MS Build Tools))
Almost at the last piece to the puzzle; waiting to fix an issue with the table data storage system (attempts to make it more efficient ending up making it a little more complex than I had aimed for which lead to a very interesting bug that I'm trying to fix).
Next up:
(1) publish the new protocol spec
(2) simply replace ~100 lines of code in the arbiter :)
Storage engine is go for launch. Finalizing protocol impl. Should hopefully have the protocol matter settled by end of day UTC :)