What happened with iCache?asked by TheMarcusArts
TL;DRThe developer says the earlier iCache timing was only an internal estimate that should not have been shared, and that discovered work, borrowed resources and extra testing and tooling pushed it back, though the service fleet is largely done.
When I wrote that (perhaps somewhat optimistically) that was indeed a reflection of the current predicted schedule for icache+global persistence. In retrospect I think I should probably have not said anything even vaguely close to dates, because really that is more part of the coordinated messaging that goes out with things like the roadmap. Our predicted internal roadmap for icache wasn't communicated outward and now this is being directly quoted as some kind of guaranteed delivery date, which it never was, just an internal predicted estimate at that time.
To answer your question "what happened?" Well a few things:
- There ended up being some discovered work along the way in order to both deliver on the goals of icache/global persistence, while also setting us up for server meshing. Without going into too many details, for example we have filled out more of the design of icache to scale better by allowing us to query on data in a way that leverages the shard aggregation of the data. Additionally we've laid out some more ambitious plans to deal with some inventory design work to set us up for planned features like the separation of players and characters (death of a spaceman kind of stuff), and to support additional design asks for things like physical inventory and cargo. There's more, but you get the idea.
- Some resources have been borrowed away from icache work here and there for things like the Long Term Persistence (LTP) implementation and support and some vital internal engine work to support the team. This hasn't been a huge amount but it did impact the schedule.
- We increased the amount of internal testing we are planning for the feature set, and increased the amount of time dedicated to tooling for debugging so that we have more confidence when it goes live for the initial release. This is an incredibly deep set of changes to the game and backend, rushing it out is to the benefit of no one.
In the last video I did I said the icache service fleet was largely done and most of the work is focused on integrating it into the game engine and updating all the game systems accordingly. That is still true.