On Fri, Apr 28, 2017 at 10:07:43AM +0200, Andrea Bolognani wrote:
On Thu, 2017-04-27 at 20:45 +0200, Pavel Hrdina wrote:
@@ -1416,6 +1416,10 @@ mymain(void) DO_TEST_PARSE_ERROR("usb-controller-xhci-limit", QEMU_CAPS_CHARDEV, QEMU_CAPS_NODEFCONFIG, QEMU_CAPS_PIIX3_USB_UHCI, QEMU_CAPS_NEC_USB_XHCI, QEMU_CAPS_NEC_USB_XHCI_PORTS); + DO_TEST("usb-controller-qemu-xhci", QEMU_CAPS_DEVICE_QEMU_XHCI); + DO_TEST_FAILURE("usb-controller-qemu-xhci-unavailable", NONE); ... and just use the usb-controller-qemu-xhci input file again for the failing test case. No, this is not just what file should be used, it's also a name of the test case and it should be unique.
There are already several test cases where we use a single input file with different capabilities, eg. all the aarch64-gic-* (those were introduced by me, so they don't count for the sake of argument), machine-aeskeywrap-on-caps, shmem...
I could point out more if I bothered looking further:
$ VIR_TEST_DEBUG=1 ./tests/qemuxml2argvtest 2>&1 | \ grep -E '^[0-9]+)' | awk '{print $4}' | wc -l 698 $ VIR_TEST_DEBUG=1 ./tests/qemuxml2argvtest 2>&1 | \ grep -E '^[0-9]+)' | awk '{print $4}' | sort -u | wc -l 644
Since you clearly have a strong preference for one approach, arguably the most correct one, let's go with that.
It was clear the first time, but it was not a good suggestion so I've used link instead.
It would have been easier if you'd let me know you disagreed with my comments by replying to them instead of me having to find out and ask, wouldn't it? :)
I didn't know that there is already a case where we use the name twice so I didn't thought about that at all, otherwise I would let you know, I just thought that it was a honest mistake, not an intention :). Thanks Pavel