I’ve been working on making llbuild itself be a Swift package, and one of the things I built up was a demo of computing Conway’s “Game of Life” using llbuild, with a web frontend.
I am hitting some reliability issues, which could completely be the fault of llbuild, but when I went to run under TSAN I noticed I am getting loads of TSAN errors coming from the HTTP server implementation. Has anyone looked at this?
The current implementation has been deemed insufficiently asynchronous for our purposes, so work has shifted to focus on https://github.com/swift-server/http/pull/96 - which would effectively replace it.
That said, I have a branch that has additional DispatchSemaphores that allows the tests to pass with ASAN/TSAN enabled. I'll be happy to put up a PR for that if it would be useful to you (or the community) for the time being until the tests for PR#96 get fixed and it gets merged.
-Carl
···
On Nov 30, 2017, at 11:17 AM, Daniel Dunbar via swift-server-dev <swift-server-dev@swift.org> wrote:
Hi all,
I’ve been working on making llbuild itself be a Swift package, and one of the things I built up was a demo of computing Conway’s “Game of Life” using llbuild, with a web frontend.
I am hitting some reliability issues, which could completely be the fault of llbuild, but when I went to run under TSAN I noticed I am getting loads of TSAN errors coming from the HTTP server implementation. Has anyone looked at this?