From: Alexandre Julliard
"Greg Geldorp" writes:
The "overlong instruction" test produces different results on different test machines. Since it fails on other machines besides the TestBot VMs I don't think this is a TestBot problem.
Do you have evidence that it also fails on real hardware?
No, I don't, since I'm not sure which test results are generated on real hardware and which on virtual hardware. I found the following failures on test.winehq.org:
http://test.winehq.org/data/17e66e06d613026387753f350aa2eaedbcb02ac1/2000_Ne... http://test.winehq.org/data/17e66e06d613026387753f350aa2eaedbcb02ac1/2000_Wi... http://test.winehq.org/data/b51f8d48128ff0876341e4f31b4d4e1f31f9dad4/2000_Sw... http://test.winehq.org/data/5f694ddf4c211c1c167a15541c2654175ce42a62/2000_Wi... http://test.winehq.org/data/5f694ddf4c211c1c167a15541c2654175ce42a62/2000_Wi... http://test.winehq.org/data/1c7d349f26c89c9be410978148c64f59c987cf82/2000_aj... http://test.winehq.org/data/1c7d349f26c89c9be410978148c64f59c987cf82/2000_aj... http://test.winehq.org/data/17e66e06d613026387753f350aa2eaedbcb02ac1/xp_fg-w...
Perhaps the submitters of those results can let us know whether they're using a VM?
Of course, if you prefer I can also either mark the failing tests as broken, or add an alternative set of allowable results.
Greg.