Re: [PATCH 14/25] qemuxmlconftest: Add test data for an ARM machine where qemu doesn't honour -usb

On Wed, Feb 28, 2024 at 06:43:06 -0800, Andrea Bolognani wrote: > On Tue, Feb 27, 2024 at 05:36:16PM +0100, Peter Krempa wrote: > > +++ b/tests/qemuxmlconftest.c > > @@ -1248,6 +1248,8 @@ mymain(void) > > DO_TEST_CAPS_ARCH_LATEST_ABI_UPDATE("armv7l-versatilepb-minimal", "armv7l"); > > DO_TEST_CAPS_ARCH_LATEST("aarch64-realview-minimal", "aarch64"); > > DO_TEST_CAPS_ARCH_LATEST_ABI_UPDATE("aarch64-realview-minimal", "aarch64"); > > + DO_TEST_CAPS_ARCH_LATEST("aarch64-nousb-minimal", "aarch64"); > > + DO_TEST_CAPS_ARCH_LATEST_ABI_UPDATE("aarch64-nousb-minimal", "aarch64"); > > Other test cases are named after the machine type, so I think these > should also be aarch64-borzoi-*. I opted to do something more generic here to make it obvious that we don't care about the specific machine type if it ever were to get deprecated in qemu. In such case this test case should not be removed but rather a different machien type picked. I can express that in a comment but it still requires the poor soul upgrading capabilities in the future to rename the files once that happens.

> As an additional note, it would IMO be preferable to add all these > test cases at the very beginning of the series, before making any > functional changes. This is nicer because it establishes a baseline > for the current behavior, and as the various patches alter that > behavior we can see this reflected in the test output. This is > already the case to some extent, I'm just suggesting going all the > way instead of adding some tests after the first round of functional > changes. Sure thing. This is an artifact of how I was developing this it should be trivial to shuffle things around.