This month was quite a grab-bag of topics!

Hello everyone, and happy August!

⏹ It's annoying that we cannot load into our last/home region directly on Worlds grids anymore, and must always go to the default landing. Can we have that back?
⏺ Absolutely! It was never the intention that this feature stop working, if this is being experienced by users of Worlds grids then it is certainly a bug and not a planned design change. We'll have a look at it.

⏹ Any news on what the future scripting format might be yet? I have a lot of game objects to make interactive.
⏺ Good question! Short answer is you can (and should) keep using existing Lua/SDK runtime for all of your interactive needs. Don't sit around waiting for the new stuff. This is in part because it is still in flux and will evolve as we move towards a production release, but also because we will be maintaining backwards compatibility with the existing Lua SDK, and it will gain even more capabilities as we go along to keep pace with the other language options.

➤ Expanded answer: We are backporting a lot of NextGen to CurrentGen, a process of slowly merging the two together where it makes sense, and adding compatibility options where they can't fully unify. Eventually both NextGen and CurrentGen viewers might be able to talk to both backends once they're sufficiently close.

The scripting engine is being overhauled, adding support for WASM and C# with a custom compiler built with nanoFramework targeting WASM. This capability mirrors NextGen very closely, but the API is different. The NextGen API is much more complex and sweeping than CurrentGen, but the new scripting engine itself is coming to CurrentGen with support for the existing APIs. This includes scripts running async (see last month's blog post) to improve viewer performance and reduce apparent freezing.

The API in C# and Lua will be broadly similar through the use of Space.* API calls just like existing Lua scripts. In C# you'd be building a proper class and have a static entry point into that class, but the APIs are identical. We also have a Lua compiler that targets this same WASM runtime which should yield improved performance for Lua scripts.

We know that AI will probably be used increasingly to write scripts for Sinespace on both Current and NextGen in the future. So, the better our API and the better tools the agents have to test their own code the easier this'll be. We want to make sure we are placing ourselves well for the future in this regard, by being able to offer agentic workflows out of the box.

➤ Side note: Our current Lua engine is our own personal fork of MoonSharp, which is an open-source project we've made a huge number of improvements too. We had aspirations to sync back some of these changes, but it has diverged so much that it would be a major project to try and accomplish this now.

➤ In summary: Keep doing what you're doing, we'll come to you and support your existing scripts. Any Lua script you write today will continue to work going forward in CurrentGen and NextGen, it'll just become faster.

Quick clarification about NextGen: NextGen will effectively have two APIs for scripting. We talk both about it having more capabilities but also remaining compatible with existing scripts, and that may seem confusing, so we wanted to clear this up a bit: NextGen has its own entirely unique API that covers a LOT more features than CurrentGen. The existing Lua API on CurrentGen has about 1400 API calls, and this will grow to over 1900 with the new scripting engine. NextGen on the other hand will have over 4000 API calls, including more advanced inverse kinematics, better animation systems, and other functions that our CurrentGen API doesn't do particularly well. What happens is that on NextGen, if you tell it that you're running a Legacy (CurrentGen) Lua script, then it will use a translation layer to remap CurrentGen script calls to NextGen API equivalents automatically for you. This way the old APIs still work, which means existing scripts still work, but they won't have the new capabilities of the NextGen native API unless you use the new API in them directly.

➤ So where is all of this stuff and when is it coming?

Right now, most of this stuff is in a gigantic merge tree that Adam is sitting on like a dragon with a hoard of treasure, but unlike most dragons he really, really wants to share it. The bad news: That big merge tree is targeting our Unity 6 development branch, which means most of this capability has to wait for that. The good news: Indigo (our next release) will be Unity 6! Our main hurdle at this point is QA resources to verify all the work that has been done, and that's probably going to be a main challenge for us going forward as our ability to develop new code has accelerated greatly, but testing and verification still has a large and necessary human component to it.

Adam would like to thank everyone who participated in the Discord poll about whether we should put out features more quickly (and potentially less stable) or move more slowly (and have higher stability). The poll was a complete draw, so that was very useful!

➤ Some other things that are in progress:

1. Significant overhauls to content loading so we can get away from proprietary Unity formats for everything, giving us full control over the content experience. This will not only improve loading by increasing speed and reducing stalls (freezes), but it will also make it possible to edit content directly in world through viewer and update the content directly on the servers, rather than relying on re-uploading through the Unity editor. The in-world editor not only supports uploading mesh and other assets, but includes supports for MaterialX formats at runtime.

2. Differential uploading has now been merged into our main branch for next EP. This is a feature that does not rely on Unity 6, so we're moving it ahead now. Additional work is being done towards eventually eliminating processing completely, bringing us down to a few seconds for most content uploads. Shaders have always been the challenge, but we have a new solution that gives us complete control over that now. All of this reduced dependency on Unity-specific asset formats and processes has the bonus side effect of making migrating to new Unity engines much easier in the future (somewhat ironically).

3. Changing how we store data related to regions. Right now, it is stored in our backend as a giant snapshot of the current state of the region at the time you press the save button. This was fine to begin with, but not a great solution long term (and it's been left this way for far too long). The region server is becoming much more authoritative, streaming changes into and from its own database. This means we can give server scripts the ability to make some changes to scene objects, which is not currently possible on CurrentGen. Bigger worlds will be possible with this change. This will speed up download times, allow streaming partial regions, and get users into the world quicker.

Some automated regression testing is in progress on these features currently, checking all existing content for conversion and loading errors. Next step will be to hand it off to QA and figure out how to eat that elephant.

⏹ Does this mean you have or will have a big new QA team?
⏺ Not currently, our QA team remains quite small. We may have to ramp it back up in size to some degree eventually, but we're also working on automating tedious repeated testing to free up their time for more new feature testing. About 30% of regression tests are now automatic. We're hoping to keep pushing this up to 100% over time for basic regression testing of new viewers, dashboard (curator) changes, etc. This automation isn't agentic, and was started before AI really took off. Not everything benefits from AI when there's no "decisions" to be made and it's merely, "always press X first, now press Y second, now type this exact phrase in this exact box every time", etc. There are some testing areas where agents can be valuable though, and we're looking into ways to use them for this when they can add value.

⏹ New features mean new test cases?
⏺ Ohhhh yeah. So many test cases. The current regression test set for a new release is thousands of operations that have to be verified, which is why we're trying to automate them. It is extremely tedious and repetitive, but necessary to make sure something hasn't inadvertently broken.

➤ Let's talk website!

We rewrote all our backend infrastructure and modernized it onto the same stack that we use for everything else. The new backend is now in testing to make sure it is 1:1 compatible with our current system for API calls. This lets us dump dependency on older, slower web development stacks in favor of a pure C# implementation throughout our infrastructure.

But that's only half of it, we want to redo the front website as well! As you all know (and everyone who looks at it can tell), it hasn't really been touched in about 6 years, so we're looking to rebuild it to focus on Worlds. This is currently underway, and we're very open to feature suggestions for this. We are of course looking at a Worlds directory (and perhaps even Worlds mini-websites). Most of the current functionality has been migrated already, as well as adding the ability to see your friends list and mail from the website without having to log in to the viewer, plus a settings page for personal privacy features.

⏹ What about allowing event scheduling and management on the website too?
⏺ Yes, we'd definitely like to do that, and it shouldn't be a lot of work.

⏹ Will the website continue to be the only social media besides Discord?
⏺ For now, yes, we have a small team and managing a lot of social media outlets is time consuming. We know there are fan groups on other services like Facebook and we'd like to support those more, if possible. If people have ideas about how we can do that please get in touch with us.

⏹ Are there any plans to bring back the old challenges and competitions?
⏺ We'd certainly like to, but there are a number of issues that led to their original demise. The main problem was lack of participation, leading to the competitions getting stale with predictable winners in each category, especially since not everyone is or wants to be a creator. We're looking for new ideas on relaunching some form of these with a new format and a new prize structure, but we don't have anything to announce right now.

➤ Okay, so... NextGen!

Especially, where is it and why can't we log in yet? Unfortunately, Adam got very sick last week and is still not back to 100%. The last thing that needs to be done is finishing a new login portal for our new authentication backend. It is coming imminently.

As mentioned before, this will be opened up to small group, it will be pre-alpha, so you should expect incompleteness and bugs. There are still major pieces being made for it. Example: Eric just got avatar animations working properly. Even so, we will be looking for early feedback on content creation and other experience flows from the more adventurous users.

We also want to have a form soon for content creators to ask for specific pieces of their content to be migrated for testing. Keep in mind that full permission control doesn't exist yet! If we migrate your content, you should be aware that the current NextGen interface has full import/export perms and doesn't have robust IP protection yet.

Most scenes and materials are porting correctly now from CurrentGen into NextGen. We're actively working on the shim (translation/compatibility layer) to let Sinespace scripts run exactly as-is on NextGen, as discussed above. This is a focus area for the next month or two. After that will come a full roadmap for avatars and future features.

⏹ Will the content approval process change for NextGen?
⏺ Yes, it will! We're designing NextGen to have a much faster iteration cycle, and this ties back to what we mentioned earlier about bringing this better processing and automated content review to CurrentGen. We're adding ways for the upload architecture to automatically detect and enforce rules around avatar nudity, trademark checks, etc., to reduce the need for human reviewers to look at every piece of content every time. Some things, like HUDs/UIs are more complex and may need manual review. An example that came up during internal discussion is someone creating a gold-purchasing screen that looks just like ours to try and trick users into entering payment information, there are a lot of phishing possibilities for UIs especially that we need to think about.

⏹ Will NextGen be stood up as a separate environment?
⏺ Yes, it will be separate for a long time. We will eventually be in a position to merge the two together, with Sinespace becoming a federated node of NextGen. NextGen on its own is brand new from the ground up with no legacy baggage. It can do so much more, and CurrentGen Sinespace is a subset of that capability. This is where the scripting compatibility shim comes in that we mentioned before. NextGen is by design an extensible platform, so things like "Here is NextGen, but we also want it to support these extra components (legacy Sinespace components) that aren't part of the base NextGen install, on this particular world" is not just possible for us to do, but is meant to be a core feature of NextGen as an architecture.

⏹ Can you talk to us about how money will work in NextGen? Transaction histories, money transfers, sales, etc.
⏺ Transaction history can (and should) probably be added to the website for CurrentGen. For NextGen, the economic model is still being planned and we can't talk about it much at the moment. This is an area that intersects heavily between attorneys, accountants, government regulations (like KYC), etc. Goal is to keep NextGen flexible in how that works, so that commerce can happen outside the marketplace as needed, allowing people to build their own alternative models as fits their world. But we of course want to have an internal system as well, and this is going to be a very detail-oriented long-term planning process due to increasing regulations.

Okay, that's all for this month! Thank you to everyone who was able to make it to our August session. The next Q&A will be back on our US-friendly time slot of 7pm Eastern / 11pm UTC on September 15th.