[Nut-upsdev] usbhid-ups: shutdown testing needed

[Arjen de Korte]

> ok, I'll retain the 2.2.1 release until we have enough confidence in this
> part.
> IMO, the APC support should be fine since at least Peter and Charles
> have tested it.
> For the others, I've no clue.
>
> Ping me back when it's ok.

You won't like the answer, but there *is* some issue with nut-2.2.0 and
what we have in Testing right now. I can't pinpoint what it is, but
apparently there are problems with both the megatec_usb driver and some
CyberPower units for usbhid-ups. Neither of these seem to be present in
the development version.

There are currently two major differences between Testing and trunk:

1) All the fixes in usbhid-ups. Mainly the caching the conversion from
text to numeric HID path, but also fixing the reports that we receive over
the interrupt pipeline for instance. The latter is broken for devices
where the logical and physical values for a variable are different (ie,
where a conversion is needed). The only functional change is in the
shutdown function, which I think was partially broken in previous versions
(ie, didn't reboot when power was present at the time of running it).

2) Splitting of the non-HID specific code from libhid.c and putting that
in libusb.c. For all USB drivers except the usbhid-ups driver, this means
that they use libusb.c only and the hidparser is not longer included. It
doesn't change the functionality of the drivers, but prevents having
dependencies that shouldn't be there.

Neither of these can be backported easily, since they require changes in
*many* places of the code, so we would basically have to copy what is in
the trunk to Testing to make these effective.

Best regards, Arjen