If we want to go for this solution, we have to scan the archive for usage of this header file, and fix them by submitting a patch.With the hope it won't come in a middle of a transition.Speaking of which, we probably shouldn't be trying too long with kfreebsd 8.0 unless we have a plan and a timeline. The blockers you mentioned make it look like we're already on people's way. So I guess it makes sense to revert ASAP and to contact -release@ once we're ready to roll.
For rebuild of libsdl1.2, it is sufficient to use this: --- /usr/include/usbhid.h~ 2009-11-28 23:38:42.000000000 +0000 +++ /usr/include/usbhid.h 2009-12-02 18:43:17.000000000 +0000 @@ -30,6 +30,12 @@ */ #include <sys/cdefs.h> +#include <dev/usb/usb_ioctl.h> + +struct usb_ctl_report { + int ucr_report; + u_char ucr_data[1024]; /* filled data size will vary */ +}; typedef struct report_desc *report_desc_t; It means we have to fix only our freebsd-libs for libsdl1.2. I am now rebuilding qemu, usb-hid.o is already compiled. The remaining known failing packages are freeglut and hal. What else ?
So I guess it makes sense to revert ASAP and to contact -release@ once we're ready to roll.
What about inform -release@ about our problem and ask them to ignore our dependecies in testing propagation. According to edos.debian.net,
the kfreebsd* testing have 177/195 binary architecture specific packages already uninstallable. This way it should not hurt other architectures. Petr