I think it was Liz Taylor who said, "I don't care what you say about me as long as you spell my name right." You'd think that spelling names right would be high on the list for journalists, but no. You'd also think that cut-and-paste would be pretty much their most-frequently-used job skill. This is the same woman who wrote an article last year about how disappointed she was at the DNA opening party, because it was (horrors!) more like a nightclub than a trade show.
Hey netik, where's that hub you keep saying you're gonna bring me? You want to come over thursday afternoon and see if we have line of sight from the roof to Meehan's house? I don't have a GPS, but I assume you do: he's N 37.75758 W 122.44406 +464 feet.
DNA Lounge is at 37.76903 N, 122.41358 W, with a distance of 1.87 miles to his place.
You've got a few things in the way to the destination, though:
There's the central expressway, mission dolores, and a park, but you might be okay.
I'll bring my GPS and see what the elevation is, although I think the elevation in front of my place (at 7th) is 136ft, so we might not be able to make it to his place.
Here's the map, I'll post TOPO data soon.
The peak of the DNA roof is somewhere around 30', and we can easily get another 20' by attaching it to the wall of the Department of Human Services building behind us; even higher if we strap a pole to that wall. (See panorama.) I'm afraid I don't recognise enough background features in that panorama to be sure which direction his place is. Is it behind the lofts? I think it's to the left of the "Godzilla building" to the right of the leftmost panel (which is on Bryant and 14th, I think -- just past Division and the freeway.)
Here is Topo Data for the southern part of SF:
http://www.topozone.com/map.asp?z=10&n=4180332&e=551759&s=25&size=l&loc=San+Francisco
It actually looks really good!
If you put a 'Pringles Can' antenna or a decent Yagi on the roof of the DNA, I think you can hit him. It looks like a clean shot with the antenna pointing upwards, although I don't know what he's using on the top of his roof.
If he's at 464 feet, and the DNA is around 25ft, that's a 439 foot difference, and that's at the ground. Put us on top of the building (let's say the building's 40 feet high), we've got to make up 400 feet over two miles.
That's:
arctan(9873 ft / 400ft) = 2.32 degrees
Oops! Botched my trig!
It's 45.02 degrees, not 2.32.
... with heading 241.1 degrees.
Whatever you do don't bother with a pingles can antenna. They are horribly imprecise with massive side lobes. Either spend a few bucks on the modled yagi atenna from http://www.dbiplus.com/ or build a cantenna from http://www.turnpoint.net/wireless/cantennahowto.html . Greg even included some javascript to calculate the correct placement of the connector based on the diametter of your can. Also, if you check his test results you'll see that the cantenna did better than the pringles.
The map and the bearing information seem to indicate that he's either above, to the left, or right of the lofts. You've got a clear shot to the right of the lofts (see the nice opening heading towards twin peaks?) but if it's over the lofts, we have a mast to erect.
Well, first things first, as soon as you get me that wireless hub, I can hook it up to the kiosk net inside...
But as far as hooking up with the other network, do you have the pringles-oriented gear to do that, or would money have to be spent? And if you have it, when do you want to come over and try to find him? (Should I bring a telescope?)
If you use some sort of alternate queueing mechanism - ALTQ on NetBSD (and probably the others as well), dunno about lignuces - you can do traffic shaping to say "only allow n kbits/sec from this interface, or m kbits/sec if it's [some protocol]".
http://www.csl.sony.co.jp/person/kjc/kjc/software.html looks like it's one of the canonical ALTQ pages.
We've already been over this. Summary:
In the thread you mention that you're running OpenBSD w/ IPF. I've added support for OpenBSD's new PacketFilter to NoCat so that authentication and connection tracking could be done. I based the PF support of of some very rudimentry support for IPF so it shouldn't be hard to find someone to port my changes back to IPF if you watned NoCat's functionality. The next major iteration of OpenBSD support will be to include ALTQ support so that it mimics the traffic shaping abilities of the IpTables support. I just need to learn it, and it doesn't look particularly friendly.
Is NoCat just the authentication mechanism? Because I'm not interested in making people register: I don't care who uses it, when, or how long; I just want to cap total usage so that it doesn't hurt me.
Is there any reason that I would need to do anything more complicated than "screw up the antenna until the connection is sufficiently slow?" Because I'm really attracted to that ultra-low-tech approach.
Yes, NoCat is the auth mechanism. You can setup in "Open" mode so that it simply throws up a splash page saying who is providing the service and what the rules are. PersonalTelco uses it in that mode to throw an AUP up to cover their arses from potential legal problems.
Screwing up the antenna might work a little bit. You could throw an attenuator on the line or use a longer coax run to reduce signal strange and EIRP so that people are more likely to link up at 2Mb rather than 11Mb. However that won't prevent someone from simply getting a more directional antenna to circumvent your reduced emissions.
NoCat would apply rate limit rules per client. If you don't care for that granularity a global altq rule for the AP would be much simpler.