While working with the Windows (and now Windows 7) build slaves I've been noticing that a number of "unexpected" test skips are occurring, mostly for external modules like ssl, bz2, tk, etc... However, I know for a fact that those libraries are being fetched and built, so I was a little perplexed that they were being skipped for testing. I don't think it's a regression due to the recent slave work, but something that may have been happening for a while, if not always. Some manual testing I did would seem to indicate that there's a problem related to loading the right MS CRT when loading these modules. But they seem to reference both debug and non-debug versions, so even if I copy the CRT DLLs (from the SxS tre) into pcbuild manually, then I get a CRT assertion about a bad initialization. I suspect this wouldn't be a problem when everything is built in release mode. I would have sworn they used to get run, but now I'm not so sure. Perhaps I'm remembering older Python releases with VS.NET 2003, since the MSVC9 versions of the CRT and the SXS stuff was new with VS 2008 I think. Does anyone happen to know if these tests have ever run recently on the Windows build slaves? -- David