<br><div class="gmail_quote">2009/1/16 Arjen de Korte <span dir="ltr"><<a href="mailto:nut%2Bdevel@de-korte.org" target="_blank">nut+devel@de-korte.org</a>></span><br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Citeren Arnaud Quette <<a href="mailto:aquette.dev@gmail.com" target="_blank">aquette.dev@gmail.com</a>>:<br>
<div><br>
> about the filtering, I thought Arjen already applied the 2nd macro over<br>
> USB_DEVICE...<br>
<br>
</div>Not yet, since I don't think we already reached a consensus on which<br>
devices should go into which script. Basically, there are three<br>
categories:<br>
<br>
1) valid VID:PID that uniquely matches a UPS<br>
2) valid VID:PID that uniquely matches a USB to serial converter<br>
3) bogus VID:PID<br>
<br>
Only for #1 there is no discussion that we can safely put this in the<br>
hotplug/udev/HAL scripts.<br>
<br>
Regarding #2 and #3, it is questionable since a device showing up with<br>
this VID:PID combination *might* be a UPS, but it could also be<br>
something totally different. I know of at least one non-UPS device (a<br>
GPS) that uses a VID:PID combination that is in category #2. I don't<br>
think we should use these in the HAL script and I'm uncertain if it is<br>
a good idea to use it in the hotplug/udev scripts. Putting them in<br>
risks that we're setting the device up for NUT (while it is something<br>
else) and not putting them in will probably lead to many questions on<br>
the mailinglist why device X worked in nut-2.2.2 and no longer works<br>
for nut-2.4.0.<br></blockquote><div><br>thanks for the clarification on #2 since until now, I was only aware of #1 and #3 <br></div></div><br>I've logged a tracker as a reminder:<br><a href="https://alioth.debian.org/tracker/index.php?func=detail&aid=311356&group_id=30602&atid=411542" target="_blank">https://alioth.debian.org/tracker/index.php?func=detail&aid=311356&group_id=30602&atid=411542</a><br>
<br>what comes first to my mind is:<br>- as you stated, remove #3 from all these helpers. NUT support should only be kept through forcing the regex matching. This leaves the permissions issue... see below<br>- manage permissions more globally within a group, as for serial devices (dialout, uucp, ...)<br>
This solution would also solve the #2 issue!<br>USB groups already exists, like usbusers on Debian. But as for serial devs, this opens a wider door.<br>so we'll have to weight the pros and cons and make a decision here.<br>
<br>Arnaud<br>