[Nut-upsdev] reassessing the HCL ratings scale

[Charles Lepple]

Here is our current rating scale for the HCL:

1 = protocol based on reverse engineering
2 = based on fragments of publicly available protocol
3 = based on publicly available protocol
4 = vendor provided protocol
5 = vendor provided protocol and hardware

I was going to suggest a few changes to the ratings we have assigned (for instance, not having a rating of "5" with an "experimental" tag, and increasing the ratings of the Tripp Lite models which they tested in-house against NUT), but I wonder if we should redefine the scale a bit. After all, just because a vendor provides hardware or a protocol spec doesn't automatically mean that the end-user experience will be satisfactory.

Any thoughts? At the very least, I would like to differentiate between "should work" and "really does work", possibly replacing 4 and 5.

While we're on the topic of the HCL, I also think it might be handy to list some of the more common vendors at the top of the "Manufacturer" filter dropdown. Suggestions for criteria would be appreciated.

-- 
Charles Lepple
clepple at gmail