On 17 April 2012 21:24, Blue Swirl <blauwir...@gmail.com> wrote: > On Tue, Apr 17, 2012 at 07:33, Peter Maydell <peter.mayd...@linaro.org> wrote: >> Just testing a device shouldn't require running a particular >> board model either, of course. > > The goal is obviously to make comprehensive tests for all devices in > all boards. Also, if a device is only used by a certain board then > yes, that particular board model is required.
Device testing should be done by instantiating the device in a test harness and prodding it according to whatever tests you're doing. Trying to use the board as an ad-hoc test harness means you can't test aspects of the device which the board doesn't use, and if the board changes then half your test cases probably break. >>> I think we should probably refactor the boards to do something more useful >>> if a kernel isn't specified verses just buggering out. >> >> I don't inherently object if you can define something that's genuinely >> more useful. I'm not sure "start the CPU and have it execute through >> zeroed out RAM so the user gets no indication they've forgotten something" >> really counts, though... > > The error could be downgraded to a warning. Well, it could. But we should make that decision based on whether it makes sense and has a use case for actual users of the board, not because we're trying to get away with not having a setup that lets us properly unit test devices. -- PMM
