Brenton's dashboard

Thanks for your efforts!

I'm not seeing the changes on all of them. EveryTrail didn't record trip duration, so I can only compare my most recent trips:

Z9 shorter:
http://zednine.com/workouts/466
Z9 - 0:16:45
MX - 0:19:26

Z9 shorter:
http://zednine.com/workouts/494
Z9 - 0:15:31
MX - 0:18:35

Z9 longer:
http://zednine.com/workouts/493
Z9 - 1:03:32
MX - 0:37:02

Z9 shorter:
http://zednine.com/workouts/496
Z9 - 00:09:24
MX - 00:11:43

Z9 longer:
http://zednine.com/workouts/497
Z9 - 02:43:33
MX - 02:25:43

Z9 longer:
http://zednine.com/workouts/509
Z9 - 01:38:12
MX - 00:28:14

I don't have a comparison for this one, but I know 9 hours is much too high:
http://zednine.com/workouts/446

The charts still seem to be off. EveryTrail behaves well in this situation. Although their total travel time includes untracked time, they usually cut it out of the graph. More importantly, their graph is always properly synced to the map:
http://www.everytrail.com/view_trip.p...
http://www.everytrail.com/view_trip.p...

The trip your algorithm thought was 9 hours totals to 12 on EveryTrail. (I promise I wasn't biking that long). In this example, there is some dead time that made it into the graph, but the graph always correlates to the map:
http://www.everytrail.com/view_trip.p...

MotionX counts the time between start and stop for each leg. (They call it a stopwatch). I don't mind from a trip duration POV if you cut out time at stoplights, etc., but I would like to see those points marked on the graph. =)

The GPX file looks correct to me:
...
< trkpt lat="37.293942" lon="-121.958198">
< ele>26.986446</ele>
< time>2009-07-17T21:11:06Z</time>
< /trkpt>
< /trkseg>
< trkseg>
< trkpt lat="37.294061" lon="-121.957496">
< ele>13.000000</ele>
< time>2009-07-17T21:46:40Z</time>
< /trkpt>
...
MotionX elapsed time: 11:43.9
Zed9 elapsed time: 45:33

http://zednine.com/workouts/496

(I added spaces above to keep GS from breaking the XML)

Velocity issue
---
iPhone A-GPS seems like a reasonable explanation.

The correct way to handle this situation may be to run a normalization algorithm over each point: If point B is more than x% above the expected value, as computed with A & C, redact it. Show a message saying "your graph contains anomalies," with a toggle to count/discount outliers.

Another solution would allow the user to edit his own data to filter out such points by hand.

Idle time
---
I've seen the idle issue in both MotionX and BikeMyDrive (EveryTrail). In both apps, I stop the track, go about my business, and resume again when I finish.

ZedNine's location and duration features behave as if idle time (where I have stopped the tracking) is included in the workout. The speed graph omits it.

I don't mean to litter your app with toggles, but like the ability to redact anomalies, you may want to include a setting for this in the Edit workout section.

Customer service
---
I appreciate that you are actively tracking the issues I report. It makes me much more likely to continue testing Z9. =)

Good luck!

Glad to hear that's working better.

Though I really don't understand how a GPX file could be so far off, it sounds like you mostly fixed the problem. For whatever reason, maybe EveryTrail just sucks at making GPX files. The ones I created in Motion-X are all within a couple MPH of your algorithm.

It looks like the graph is still using the old algorithm. #384 still has 200 mph on the Y axis.

The blue hover indicator seems to position itself based on where I actually was on the x (time) axis; however, the graph still ignores idle time. Hence, the location indicator does not correlate speed and location.

I suspect you may be doing something naive on the speed graph (such as evenly spacing every data point, regardless of how long it was time-wise from the previous point). That said, I like that it doesn't count idle time (like the trip duration indicator does).

There should be a threshold (1 minute?) whereby you consider a user to be idle if there's no data for greater than that amount of time. Using this algorithm, you could iterate over all the points in the GPX file and calculate the actual travel part of a trip's duration.

Glad you guys are hard at work! I've had a hard time recommending this to anybody when the important parts (speed and duration) were/are so inaccurate. I do like your UI much better than EveryTrail though. As your site improves, I'll keep telling people about it!

I can confirm that this is a Zed Nine problem, not an iPhone GPS problem.

I used a different app to track my most recent trip (MotionX). It emailed me the GPX file along with a list of trip statistics:
Name: To Orchard Valley Coffee
Date: Jul 7, 2009 12:39 pm
Distance: 3.25 miles
Elapsed Time: 19:26.0
Avg Speed: 10.1 mph
Max Speed: 21.5 mph
Avg Pace: 05' 58" per mile
Min Altitude: 49 ft
Max Altitude: 141 ft
Start Time: 2009-07-07T19:39:39Z
Start Location:
Latitude: 37.300162o N
Longitude: 121.952273o W
End Time: 2009-07-07T19:59:05Z
End Location:
Latitude: 37.300019o N
Longitude: 121.952412o W

Notice the top speed, 21.5 MPH. Zed9 thinks it's 31.3 MPH:
http://zednine.com/workouts/466

It's not so much the multiple mailboxes thing that I'm complaining about. If a domain is flagged, I want to be able to turn it off with one click, without even entering the settings.

As far as the UI on the mailboxes page itself, I don't think the checkbox metaphor is very intuitive. That iPhone-style move/copy slider in my mind means enable/disable. I click it every time this happens, then I realize the alternative is "copy" instead of "don't move," and I have to find the checkbox.

As someone who knows how to use a computer and how to use the Internet, I shouldn't have to read copy to figure out which widget to click to disable OIB filtering on a domain.

No, I'm just getting emails from real people that get flagged as OIB and archived. The workflow to disable OIB filtering on messages from a certain domain right now sucks.

Thanks Hoon.

There ought to be a widget in the mailbox UI to ignore emails from a domain. (I know it's in the settings, but it's also a pain in the ass to get in there everytime OIB starts catching my human-mail). Your bot has a tendency to be over-zealous about what it thinks is spam.

Also, there's no allowance for domains that send both types of email. A friend of mine runs a company with a newsletter, but also sends me personal mail from the same domain. His personal messages were getting eaten as if they were spam.

It's been a while since I looked into this. I just did a search for Facebook and I have messages pre-Gmail that are tagged OIB.

Looks like they got scanned sometime after I posted the original request.

On a side note, I never actually check my OIB. I just like that it takes crap out of my inbox stream. Maybe be good for you guys to know for bus-dev.

They have the original senders. I pulled them down from Yahoo as EMLs, converted them to MBOXes, and posted them with Gmail's Email Uploader utility.

I don't think you understand my point.

I started using GMail, then setup OIB, then imported my old Yahoo! Mail. OIB scanned my empty GMail for quasi-spam, but didn't find any because my GMail was empty. My Yahoo! Mail had plenty for OIB to sort. I was wondering if there was a way to make OIB rescan my GMail archive, to sort all my pre-OIB mail, after I imported it from my old mail accounts.

It's not a huge deal for me right now, but it would have been nice.