bobbydigitales

You can't fit all that in the title of a post. The readme is very clear about the claim. There's no way to get a wide variety of real world program benchmarks for a brand new language. You can look at the benchmarks and determine if you think they're fair.

Well it's their original idea and it's based on the Lobster compiler they wrote, so it isn't just a random person doing it. I'd say it's more like an expert language designer using a tool to explore an idea they otherwise wouldn't have time for.

I suppose to be distinctive and have a "new idea", you sort of have to do that. Otherwise there's really no point in making a new language.

Yeah I hate the Claude writing style. It would be better with a cut down README. I suppose the point is that this was a latent idea that they got out there using these new tools, and that this would probably have not been done otherwise.

Yeah it's very clear at the bottom that it's created by AI. I think this is an interesting case where someone with some great ideas they never got to can actually implement them. I personally hate the Claude style, but just looking at the samples was enough to get a feel for the language.

What kinds of things do you think might not translate well?

Thanks again for the feedback. We moved this information to our blog now: https://blog.dotbigbang.com/2023/10/10/dot-big-bang-v0-8/

Thanks!

Thanks for testing! I do see around a 2 second load time on my current device (s23) see the video here: https://youtube.com/shorts/OQ8x92hPfkA?si=6ugW-JGSgIVWtqkv

I'll check out the SoC for your phone and see if we have specific problems there, and what we need to optimize, thanks!

Ah interesting! It's great to know what the expectations are! I'll think about doing that in a future post once we've done some more optimization!

Thanks for testing and going the info! Firefox is about 40% slower than Chromium-based browsers and Safari for us, we focus on the default browser on each platform so Firefox isn't tested often. Firefox also accounts for about 0.4% of our traffic and their performance analysis tools aren't very deep, so unfortunately it's hard to justify allocating time to improve there.

We also can't load in 2 seconds on any device we can run on. There are obviously some devices that ship Chrome on Android that can load the game, but are incredibly slow. So yes, not always 2 seconds!

Finally, we still have a huge amount of known performance we can gain in our loading, so I do hope I can come back in the future and show that 2 second loading on a lot more devices!

Hey! Thanks for checking it out. Twitter broke threads for people who aren't signed in sometime this year so unless your signed in, they will only show you the first tweet in a thread. Previously you could use Twitter as a place to post general information like this, but since they broke it I should really move the information somewhere else!

We have a fix for this that will be rolling out soon, thanks!

Thanks for reporting, we have a fix for this that will be rolling out soon!

You should be able to see that loading performance on a "cold load" yes, but it's difficult to account for all network conditions. It's also possible for the multiplayer connection to take a little longer. I think if we messaged that more clearly then it would explain what's happening. We also considered just putting you in the game before you got the multiplayer connection, but that might be a bit jarring when it added all the players and entities in. We should maybe still think about it!

There's also a lot more room for loading optimization, I think we can maybe go 2-5x faster than what we have currently!

We do use some WASM yes and intend to use more in the future, but the majority of the engine is still TypeScript compiled to Javascript.

The pricing model is not yet announced, but it'll be a rev share between the creator of the game and us. More info here: https://developers.dotbigbang.com/

Thanks for reporting, we're working on it!

Thanks for testing! Could you try this game: https://dotbigbang.com/game/cab2b545338144d68fb8934801a1cd97...

And could you let me know which browser you're using?

Thanks!

Great, thanks :)

Thanks!

Ah awesome, well well done, it's such an awesome example of what can be done with the web!

We went for Go as it's 1) about 100x faster than Python in testing 2) Garbage collected so prevents many memory-related bugs (many of us have a C++ background so we're familiar with manual memory management and its pitfalls) and 3) It's super stable.

We did look at Rust, but Rust suffers from a little bit of the "shiny new thing" syndrome. We have simple needs and wanted something fast and boring :)

The server-side design is also as boring as possible, just standard web stack for serving, and websockets for multiplayer. We use EC2 for everything yep!

Thanks!, This really is something we just haven't got to. I'm adding a task right now for it!

Thanks for trying it out! I see a 2 second load time on Chrome on my macbook air on WiFi, what device and browser are you using?

Just adding on to this, I have a work-in-progress GLTF exporter that enables you to export your game level from dot big bang into Godot(https://bobbydigitales.github.io/twitter/data/tweets_media/1...), Unreal Engine 5 (https://bobbydigitales.github.io/twitter/data/tweets_media/1...) Unity (https://bobbydigitales.github.io/twitter/data/tweets_media/1...) and Blender (https://bobbydigitales.github.io/twitter/data/tweets_media/1...)

It's nice because all the game entities end up being separate entities in the destination tools, so you can rearrange and move things around vs it just being a monolithic blob.

I would like to consider open sourcing the engine at some point in the future, at least for single-player games, but my view is that with a small team, we have to make the engine excellent and the platform successful before we do that.

Lol sounds like an epic last play :) In terms of performance, Firefox is about 40% slower than Chromium-based browser or Safari, basically purely in the performance of Spidermonkey. We have tried to optimize on Firefox in the past, however the development tools there make it difficult to determine what the problem might be beyon "everything is slower". For V8 we have the Chromium tracing tools that les us understand where V8 is spending time on potentially unoptimized code, or on megamorphic property access. Finally, unfortunately Firefox is currently used by 1% of our players, so it's hard to find time to improve it.

I would love to know how vanilla Chromium performed on the same machine! https://download-chromium.appspot.com/

Thanks! In our testing, Firefox is about 40% slower than Chromium-based browsers or Safari. Would be interesting to know what the performance was like with Chromium on that device! https://download-chromium.appspot.com/

Thanks, glad it worked for you! We want to make sure it actually works on a broad range of devices! Our lead artist regularly tests on his Pixel 2 :)