On Wednesday 14 February 2007, Paul Vriens wrote:
Maybe we should start thinking (again) of having several sources for the winetest executable. The only issue I'm seeing there is that the builds could be different (if build by different people at an unknown time of the day/state of GIT) hence test.winehq.org should make a distinction as well (to not compare apples with pears).
Many if not most failures are due to missing imports on MinGW. Using Wine's import libraries should make the service more robust, at the cost of somewhat reduced testing exposure.
-Hans