Hey! I'm the author of the decision to "open-source" it in its current form.
Previous commenters nailed it -- it's much better to make the LSP generally accessible now -- to use it, to explore, to iterate on the actual feedback and not on the internal echo-chamber instead of cooking it internally for ~one year to make it fully OSS.
And not having to fight an uphill battle with the internal codebase (we have plenty of experience here -- Kotlin itself is fully OSS while IDEA part is mirrored through the internal monorepo) makes wonders.
> smuggles some kind of binary
These are basically IJ parts. No new dependencies, no statistics, no internet access from this code. It is also not obfuscated, so if you really worry about the contents -- any public bytecode decompiler will do.
Previous commenters nailed it -- it's much better to make the LSP generally accessible now -- to use it, to explore, to iterate on the actual feedback and not on the internal echo-chamber instead of cooking it internally for ~one year to make it fully OSS.
And not having to fight an uphill battle with the internal codebase (we have plenty of experience here -- Kotlin itself is fully OSS while IDEA part is mirrored through the internal monorepo) makes wonders.
> smuggles some kind of binary
These are basically IJ parts. No new dependencies, no statistics, no internet access from this code. It is also not obfuscated, so if you really worry about the contents -- any public bytecode decompiler will do.