Jeremy White wrote:
On a side note, there is something that worries me a bit. Specifically, it appears that we have a dirty secret about the regression tests - that they only work reliably on Alexandre's machine. Is that true? Shouldn't we do something about that? Or am I on crack, and lots of people run all the tests cleanly every day...
Hi Jeremy,
No crack: you can see the variation graphically. Because bug 3666[1] has become stale, I sent a winetest report[2] from an SMP machine to demonstrate problems running Wine on the terminal server at our office: a Dual AMD Opteron with Fedora Core 5 x86 (32-bit software only). Then, I sent another report from a single CPU machine. The former report had many more failures.
[1] http://bugs.winehq.org/show_bug.cgi?id=3666 [2] http://test.winehq.org/data/200605161000/#Wine
(Later, I suppose should run the tests on the same machine with an SMP kernel and a normal kernel.)
Andrew