"Native" support for "X" language

[DeityLamb]

I have a question — how do you decide which languages are supported natively in Zed?
Are you considering expanding that list, or do you plan to keep it limited to a specific set of languages?
Do you accept pull requests that add native support for new languages?
If so, are there any minimum requirements for native support?

For example, does it need to include LSP integration, debugger support, or a test runner right away — or is it acceptable to start with just a Tree-sitter grammar and incrementally add features through separate pull requests?

I'd like to implement native support for Java, but I'm not sure what format you expect for such contributions and how you evaluate and approve them.

I understand that Java is already supported through an extension, but the integration of critical functionality is currently very limited — such as debugger invocation, dependency decompilation, and test runners.
Unfortunately, the extension API doesn’t currently cover all the capabilities needed to implement full support for those features.

[SomeoneToIgnore]

It seems the native languages list is frozen unless something really fundamental happens.

The list was formed iteratively, where first users submitted a lot of PRs with the new language grammars, and then team moved many of them into the extensions.
I think the move out process is not finished, but somehow stalled.

The rest of the languages that are supported natively fall back into 3 tiers, as it seems:

• the ones used in the repository (note how proto is not uncluded into that list for some reason)
• the ones very popular (TypeScript) or used by the community members (that's how Lua was added at some point but then moved away into the extension)
• the ones that are not removed out of the repository for some reason

Each extra dependency, esp. the one that carries tree-sitter grammar and certain Rust code, takes a toll on the final linking process and overall compilation process — hence, the less is better for this, Zed Core, repo, ergo no plans to add new things, unless miracles happen, exist.

native support for Java

Ohoho, why that among all others...

I'm almost certain no one in the team had touched Java for a long time at best (if touched at all), and does not use it anyhow regularly.
Native support does not matter THAT much except a few hardcoded scenarios where nothing ever matches what Intellij Idea can do with their fully native support: Zed will still have to rely on language servers for Java, and last time I've checked, their quality was sub-par, take the indexing times alone.

Given all that and other things, it's never likely for Java support to appear anywhere in Zed Core: no way a not used language (even in Zed, Java's extension usage part is very small) with odd language servers will get as much traction as the popular ones (that already have issues, even with "native" support!).

Besides, we do have wonderful, motivated people who have built two extensions for Java and, as you may spot in this tracker, struggle with the capabilities both Zed and language servers are able to provide.
A set of words in the discussion from a person, random for this project matters nothing, frankly — taking on a huge scope churn based on that alone is "peculiar".

There's a Roadmap to 1.0 with no Java in it, should we drop it all and start supporting a random language natively?
Or should Zed first solve its baby issues first, mature an extension API and then consider options?

[DeityLamb]

Thank you for the response!

I wasn’t trying to push for native Java extension support in Zed or anything like that — sorry if it came across that way.
All of these questions were actually aimed at helping a “random” person like myself understand how things are structured on your end, and where exactly I could invest my own effort to satisfy my own needs or curiosities :)

Thanks for all your work!