You are working on the Kludex/uvicorn repository. The source code is checked out at /testbed at commit…

You are working on the Kludex/uvicorn repository. The source code is checked out at /testbed at commit…: a task in SWE-Lego-RL-2699 (Harbor dataset). So, I think uvicorn should allow for non-ascii chars in header. It is definitely the only reason that we can't use uvicorn in production yet.