This is an alternative to the official Zig Datastar SDK https://github.com/starfederation/datastar/tree/main/sdk/zig

The main difference is that this API uses stream processing, instead of building up a buffer. As a result, you can connect this to any code that simply print() or writeAll() to the stream, and the lib will inject the necessary protocol commands.

When you zig build it will compile several apps into ./zig-out/bin/ to demonstrate different aspects of the api

• example_1 shows using the DataStar API using basic SDK handlers
• example_2 shows an example multi-user auction site for cats with realtime updates
• example_3 shows an example multi-user pigeon racing betting site with realtime updates
• example_4 shows an example multi-game, multi-player TicTacToe site, using the backstage actor framework

This repo is currently in semi-stealth DEV mode

That means sweeping breaking changes, as we are targetting DataStar 1.0 which is only in RC release at the moment

And ... probably a reworked http.zig as writergate changes may be implemented soon

And ... Zig 15 which doesnt come out till August 2025, and has many breaking changes

So ... lots of breaking changes, and yolo merges straight onto the master branch

After things settle, we will do the whole documentation - CI - tests - FeatureBranches engineering goodness

Until then - wild west development

All contribs welcome.

Please raise an issue first before adding a PR, so there is room for discussion before any code hits

Until we get out of stealth mode and get into a proper engineering cycle, PR reviews will be light and easy

Once we hit release ... expect PR reviews to be harsh and well reasoned

If you want to modify what someone else has already worked on - by all means put in a PR to change it, but also ping them offline or put them on the review list to comment on the change

Realistically, once code is 'merged' then we ALL own it, but try and keep others in the the loop anyway

This is partially for common courtesy, but also for the Chesterton's Fence principle https://fs.blog/chestertons-fence/

Longer term - I would like this to hit 1.0 at the same time as Zig 15 and writergate changes settle down, and then ONLY add new features to keep it in track with D* updates (or Zig / http.zig updates)

This is intended to be a professionally supported API for the DataStar community after we hit 1.0. Im committed to that at least, and I believe we have enough good ppl on the team already to provide good support coverage long term

Its not a particulary difficult bit of code to maintain

Happy to advocate for DataStar and Zig and this API very strongly

But ... we dont say anything until we have reproducable benchmarks that people can apply themselves and come to their own conclusions

If we do it right, we shouldnt need to sell anything - it will sell itself

Avoid LLM like the plague please. By all means use it for rubber ducking, but dont trust any code it produces

Its just not there yet (even if it looks convincing)