Re: [802.3af] comment #178
If and implementer chooses, they could differentiate between devices that
used a common detection & classificationsignature or seperate detection and
classification signatures, there would be nothing stopping them. We've
also never been keen no implementor using a form of sampling to plan power
utilization, and statistics relies on sampling - but I suppose an
implementor could do that too. It would seem cleaner, to me at least, to
drop class 0 entirely.
Class 4 is needed, if anone ever comes up with a clever way to expand our
work. While I said it could be dropped from the PD as long as it was still
defined for the PSE, that would not be my preference. We should say RFE
everywhere, so if the expansion of clause 33 happens after the demise of
the whole group, those who follow will know where we left them space.
Clearly and concisely - drop class 0, fold it into class 3, keep class 4.
Mike
Dave Dwelley <ddwelley@xxxxxxxxxx>@majordomo.ieee.org on 01/22/2003
08:18:06 PM
Sent by: owner-stds-802-3-pwrviamdi@xxxxxxxxxxxxxxxxxx
To: Mike S McCormack/US/3Com, stds-802-3-pwrviamdi@xxxxxxxxxxxxx
cc:
Subject: Re: [802.3af] comment #178
I recall suggesting this (St. Louis 1?) and being roundly abused for "not
understanding statistical power management". Seems that "unknown" is
different from "will be full power in some mode", at least in someone's
opinion. I personally agree that Class 3 seems redundant, at least from a
hardware point of view.
I think we need to keep Class 4 in the PSE half of the spec - this lets me
build a PSE chip that supports Class 4, and system vendors can ship
firmware upgrades to recognize class 4 when it gets defined.
Dave
At 06:04 PM 1/22/2003 -0500, Mike_S_McCormack@xxxxxxxx wrote:
>I would not be adverse to folding Class 0 into Class 3 since we have
>determined that PDs must provide a stable classification signature, its
>just that a stable detection signature over the classification voltage is
>also a valid classification signature now. (That is if I remember what we
>had done in Vancouver2.)
>
>As far as dropping Class 4, we need that if we want to leave room for
>future engineers to expand the power stuff.
>
>We reserve the range in the PD as class 4, which I imagine we can drop.
>End the classification region for PDs at the end of Class 3, and thereby
>reserve Class 4 be default. But a PSE built to day needs to know what to
>do when it runs into a device in what we intend for future stuff, and our
>only rule for future Class 4 devices is that Class 4 device better not
>start out more than class 3 requirements because that is what a legacy
>(legacy once the next group updates the spec.) PSE will do to get started.
>
>My two cents.
>
>Mike
>
>PS - is there anyway we can get a rule in .3 that we will not recycle
>meeting cities until all groups that met in that city have finished? It
>makes it more difficult to specify when a decision was made. While we're
>doing that, we should insert a rule about no plenary meetings at airport
>hotels.
>
>
>
>
>"Chad Jones" <cmjones@xxxxxxxxx>@majordomo.ieee.org on 01/22/2003 05:48:22
>PM
>
>Please respond to <cmjones@xxxxxxxxx>
>
>Sent by: owner-stds-802-3-pwrviamdi@xxxxxxxxxxxxxxxxxx
>
>
>To: "AF Reflector \
>cc:
>Subject: [802.3af] comment #178
>
>
>Colleagues: anyone have thoughts on this comment? I'm not sure which way
>to
>go with this comment.
>
>-----------
>CommentID: 178
>CommenterName: Thaler, Pat
>CommenterEmail: pat_thaler@xxxxxxxxxxx
>CommenterPhone: 916-788-5663
>CommenterFax:
>CommenterCo: Agilent Technologies
>Clause: 33
>Subclause: 33.2.3.2
>Page: 41
>Line: 46
>CommentType: E
>
>Comment:
>It would be more reader friendly to have the value match the class number
>that is supported. It also isn't clear why Classes 3 and 4 are lumped
>together. Also, it isn't clear why the last value isn't simply Class 3
>since
>the text says it is the highest power supported. Class 4 is currently
>undefined but the table says it is limited to the same max power as Class
>3.
>Class 0 means that the power will be less than or equal to Class 3.
>Therefore the highest power would be Class 3.
>
>Same comment applies to do_classification on page 42 line 44
>CommentEnd:
>
>SuggestedRemedy:
>Values: 1 Class 1
> 2 Class 2
> 3 Class 3
>RemedyEnd:
>
>Response:
>
>ResponseEnd:
>CommentStatus: X
>ResponseStatus: O
>Topic:
>CreateDate: 1/2/2003
>LastModDate: 1/2/2003
>DispatchDate:
>WrittenDate:
>Accept_RejectDate:
>Closed_UnsatisfDate:
>VoterStatus:
>
>Chad Jones cmjones@xxxxxxxxx
>Hardware Engineer Phone: 330-664-7818
>WNBU Engineering Fax: 330-664-7990
>Cisco Systems
>320 Springside Drive
>Suite 350
>Akron, OH 44333-4500 www.cisco.com
>
>(See attached file: C.htm)
>
>
>