UserAgent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10_12_5) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/59.0.3071.86 Safari/537.36

Steps to reproduce the problem:
1. Nurture a desire to put USB interaction on a separate thread
2. Become disappointed that it's beyond your reach

What is the expected behavior?
WebUSB has a lot of potential, probably a lot more than people are creative enough to think of right now. I am trying to port SANE (scanner drivers) to WebUSB. 

I thought it was natural to put the USB code on a Worker or SharedWorker, because USB devices (and particularly scanners) are timing-sensitive. If I leave the device in an undefined state, or wait too long to process something, it could eventually cause damage or excessive wear to the device.

I think that being off the main event loop is probably important for most non-trivial USB devices, so I would like to be able to access WebUSB from a Worker, and preferably from the other types of worker as well (within the limits of SOP and all that lovely jazz).

What went wrong?
I'm not emotionally prepared for this question.

Did this work before? No 

Chrome version: 61.0.3126.0  Channel: canary
OS Version: OS X 10.12.5
Flash Version: 

WebUSB is awesome, I hope it can firm up over the next year so that the latency testers, hardware diagnostic tools, sporadic-use drivers, drivers for ChromeOS, and device configurators can some day all function reasonably reliably within a browser.