There are some issues with 0.1.13, please wait for @Ubospica to release a new version.

@zhyncs understood, thanks for the review and info.

I presume the issues you reference are build related? Installing from wheel passed our internal smoke tests fwiw.

@qeternity Yes the CI fails because xgrammar's new version contains a kernel not compatible with CUDA<12.4. It should work properly for CUDA>=12.4. We have fixed this issue (mlc-ai/xgrammar#206) and will release a new version later.

@Ubospica gotcha. We are running 12.4 so I guess that explains things :)

Ah Currently SGLang is using CUDA 12.1 for the CIs :)
ref https://github.com/sgl-project/sglang/blob/main/docs/developer/setup_github_runner.md#step-1-start-a-docker-container
Maybe we should upgrade CUDA 12.4 @merrymercy @Ying1123

xgrammar just releases 0.1.14, which supports lower version of glibc mlc-ai/xgrammar#113 . Is it possible to support latest version xgrammar to ease the installation on older os?

I have updated this PR to 0.1.14

Why did this PR skip the test? @shuaills @zhaochenyang20

Note: python/sglang/srt/constrained/xgrammar_backend.py has changed.

@zhyncs I think this should be fine

@zhyncs I think this should be fine

@zhaochenyang20 I don't think so. When we update python/pyproject.toml or python files, we must trigger the pr test.

Sure. We are fixing this today:

#3566

If failed, we can revert it.