What I mean by worthwhile is more than just the number of devices they sell. We'd have to determine if there is a significant enhancement to the usability/capability of the device with a hard-float v6 vfp version opposed to soft-float v5. If it's only a marginal increase, it wouldn't make sense to build and maintain that repo. We'd also need to make sure hard-float drivers are available for anything proprietary in it, and compatible with Arch's bleeding-edge versions.
Additionally, I'd need to have at least two of the boards on my desk to do the building, preferably more so that it can get done faster. I could possibly employ my v7h fleet using a v6 rootfs, since they'd support vfp and the v6 instructions. For a good time comparison, it took my Beagle-XM, Panda, and TrimSlice months to get the v7h repo where it is today. I got the base hard-float system in place around beginning to mid July, when I was finally able to start building out the rest of the repo. And you know well how many issues we've had with packages and getting things building properly.