Latest (Oct 2, 2025) Windows nightly toolchain DOA?

Hello,

I don't know where else to post this, but I installed the latest main nightly Swift Windows toolchain (October 2, 2025) and there appears to be an issue with the toolchain's command line

PS C:\Users\jenkins> swift --version
PS C:\Users\jenkins> swift build --help
PS C:\Users\jenkins> swift build --build-tests
PS C:\Users\jenkins> swift --help
PS C:\Users\jenkins>

Restarting did not fix the issue.

The toolchain is a month old. Has this been fixed/addressed? I created swiftlang/swift#85252 to track it

Cheers,

Sam

Hi @bkhouri,

Yes, it seems that a new library was added and missed in the distribution. While the CI has been stable for Windows, the nightly builds have not been produced due to other issues. @mishal_shah should be able to give more details on that.

I know that I and others are also waiting on a new toolcahin snapshot. Hopefully we get one soon!

Saleem

A few questions:

  1. This is not the first time I encountered this symptom(swiftlang/swift#80356). What strategy is in place to ensure we don't re-introduce this behaviour when a new library will be added in the future?
  2. Do we know what is preventing a new Windows nightly toolchain?

The best option is to actually consider using GitHub - thebrowsercompany/swift-build: Swift toolchain builds by The Browser Company for your toolchain needs as that is currently being smoke tested and did catch this issue when it occurred. That CI is much more open and allows you to see what is happening on the releases.

IMIO, in general, you should be testing your toolchain on all the platforms when you add a new library.

No, we do not know what is going on there. The nightly release process on swift.org is not public and you would need to ask @mishal_shah about the details.

@mishal_shah , @compnerd : The November 3 nightly windows toolchain image is still unusable due to Nighty Windows toolchain is not usable · Issue #85252 · swiftlang/swift · GitHub

When can we expect a fix for this?

Do we know which change introduced this regression?

I don't remember the PR, but it was the one that Doug made to the swift-syntax repo to introduce a new type of construct.