Re: [PATCH 0/1] MR7968: d3d9/tests: Replace incorrect "broken(1)" condition with flaky.

[Elizabeth Figura (＠zfigura)]

Thanks. In that case it seems like the prudent thing to do would in fact be to call Reset() from WM_WINDOWPOSCHANGED, as the application does, and probably also test that we did so at most twice? I imagine we might have been afraid to do so lest a failing test result in a hang or crash rather than a simple "test failed" message, but I at least am of the school of thought that a failing test is a failing test, regardless of the mode of failure.