More mapping accuracy trouble (Jackson County, MO)

[Guest]

I've been doing some stumbling in the Kansas City area, particularly Jackson County. From what I see when I compare the JiGLE maps and what my GPS shows, there is a large error range.

Example:

The fartest south I've taken the GPS is about N38deg48.135',W094deg22.441' according to the built-in maps of my Garmin eTrex Legend, whcih shows that spot on as the north jog/intersection of highways 58 and 291. That is where if you continue on 58 you end up going east on 175th ST. The online maps show that I have found APs nearly as far south as 347th ST. WHOA!

Also, the cluster of APs here should really be at the intersection of hwy 150 and Ward road, centered here. That cluster has a definite line extending south, which should be on Ward, and a two point E-W line below the tight cluster which should be on hwy 150. The tight cluster is from where I dstumbled a new subdivision just NE of the intersection, and the roads have yet to make the maps.

This info should be in intransaction id 20040729-00018 (My first log, a big ~10mb one). I've also got a couple more stumbles to u/l. Would this be the maps or my GPS? My GPS is mapping me right where I should be, but it's spitting NMEA to dstumble 1.0 on a FreeBSD 4.9 laptop, if that helps any.

[frogman]

Arg, I swear I was logged in when I posted that. I logged in on purpose just to do it...


Frogman

[uhtu]

yeah, that appears a bit odd.
out of curiosity, what are you using as your stumbler software?

[bobzilla]

it's spitting NMEA to dstumble 1.0 on a FreeBSD 4.9 laptop, if that helps any.

There was an issue with dstumble using hexigesimal instead of decimal, wasn't there?

[uhtu]

hi, i'm dumb. le sigh. someday i'll learn to read.
get yer hot dstumbler patches here:

https://hugh-kennedy.chicago.il.us/dstu ... tch.latest

turns freaky babylonian numbers into proper human decimals.

[frogman]

I nabbed that patch, it'll be applied before I pollute more logs... Is there anything much to do for my old logs? And I'm assuming you guys are fixing what's in the DB from my uploads?

[uhtu]

depending on the extent of the effect we may either re-run your logs (if they're still around) after doing some parser repairs, allow them to self-correct based on other observations, or (very rarely) isolate and do sexigesimal corrections in the dataset itself.
single observations don't mean much (for this very reason), which is why they are given a low QoS value in WiGLE. it may take a while to correct, however, if it ends up being more than a few minute fix.
thanks very much for the catch, let us know if anything else shows up awry!

[Guest]

hi, i'm dumb. le sigh. someday i'll learn to read.
get yer hot dstumbler patches here:

https://hugh-kennedy.chicago.il.us/dstu ... tch.latest

turns freaky babylonian numbers into proper human decimals.

The patch is not worky. I.e. the link is broken.

I upload almost exclusively with dstumbler recently... so a lot of my data is way off. Also, I deliver pizza part time, and so I see a lot of the same access points in the same location, several times per week, for months straight... Is there any way to figure out what logs are tainted? You probably wanna re-run any of my files that weren't .ns1...

[frogman]

And that would have been my friend ax0n (currently#126 on the stats) posting that... And the links works for me, but Mozilla grumbles about certificates.

Last edited by frogman on Tue Aug 10, 2004 9:33 pm, edited 1 time in total.

[bobzilla]

uhtu has a self-signed cert for the ssl, so it will complain, as does WiGLE, since we want crypto but refuse to pay the $100 to the evil powers in order to get a signed one. (of course, maybe you trust that we are who we say we are a bit less...)

[kleine18]

http://www.CAcert.org

CAcert Inc: The Free Community Digital Certificate Authority

[bobzilla]

They aren't part of the default trusted set for browsers, so it wouldn't help the above issue