Data and techie stuff thread for BAFS 2016

[hozn]

@hozn 135017 wrote:

It is likely a bug; I will look into it. (Maybe the instagram photos aren’t syncing at all; I haven’t been using Instagram so my “testing” has been incomplete.)

This was a bug in how I was syncing primary instagram photos specifically. They were getting pulled into database, but not associated with the ride. This is fixed now and I’m re-processing (there were ~160 missed photos). Thanks for bringing this to my attention.

[dkel]

@Judd 135020 wrote:

I usually see about a four to eight hour delay from when I upload a picture on Strava and when it appears in the photos on Freezing Saddles. It appears that there’s a little bit of a lag for the people I follow too. Not complaining, just observing.

In my case we’re closing on 12 hours. Like you, though, not complaining. Frankly, managing this has to feel a lot like the old day in, day out job for @hozn; one wonders why he volunteers for this gig in the first place.

@hozn, you’re the best.

[hozn]

@Judd 135020 wrote:

I usually see about a four to eight hour delay from when I upload a picture on Strava and when it appears in the photos on Freezing Saddles. It appears that there’s a little bit of a lag for the people I follow too. Not complaining, just observing.

Yeah, so we sync the activity details (which include photos and segments) a couple times an hour; if we sync the activity before the photo has been associated then it won’t show up until we do a “rewrite” of the data, which only happens every couple of hours, since it is more API-intensive to re-ask for data we have already retrieved.

[dkel]

@hozn 135021 wrote:

This was a bug in how I was syncing primary instagram photos specifically. They were getting pulled into database, but not associated with the ride. This is fixed now and I’m re-processing (there were ~160 missed photos). Thanks for bringing this to my attention.

Wow. This crossed my last post mid-stream. I’ll reiterate, though: @hozn, you’re the best! (Except possibly for those 160 missed photos…but who’s counting? )

[hozn]

@dkel 135025 wrote:

Wow. This crossed my last post mid-stream. I’ll reiterate, though: @hozn, you’re the best! (Except possibly for those 160 missed photos…but who’s counting? )

I think things are all fixed this AM. I didn’t check in on my computer, but it was chugging along when I went to bed. Since there are over 10k rides now, reprocessing all of them takes a bit of time. I probably could have used a slightly smaller hammer for that job, since it would only have been rides affected since I updated my sync logic. But computer time is cheaper than my time.

(I did see your photo there, so that’s at least good. At some point I’ll make the interface less confusing regarding photo sync in the “My Rides” section.)

[dkel]

@hozn 135032 wrote:

(I did see your photo there, so that’s at least good. At some point I’ll make the interface less confusing regarding photo sync in the “My Rides” section.)

My photo is there, despite the Ride Data page grumpily declaring none of my photos have been fetched for weeks. But I would only look at the Ride Data page if I didn’t see my pic, so the problem is fixed, as far as I’m concerned.

[whunke]

Not a complaint by any means, but just a note that all data might not be syncing correctly – specifically HP laps since I did 8 last night and they didn’t show on the leader board

[hozn]

@whunke 135124 wrote:

Not a complaint by any means, but just a note that all data might not be syncing correctly – specifically HP laps since I did 8 last night and they didn’t show on the leader board

I have been seeing errors on the detail-sync (which is also responsible for segments); I haven’t had time to look into this, but will get it fixed. Somewhat-expected errors in photo sync are prematurely bailing out the script, it seems.

[whunke]

@hozn 135142 wrote:

I have been seeing errors on the detail-sync (which is also responsible for segments); I haven’t had time to look into this, but will get it fixed. Somewhat-expected errors in photo sync are prematurely bailing out the script, it seems.

No rush for sure! Thanks for all the hard work in setting this up and keeping it together – it’s made for an awesome winter cycling experience so far!

[hozn]

@whunke 135152 wrote:

No rush for sure! Thanks for all the hard work in setting this up and keeping it together – it’s made for an awesome winter cycling experience so far!

My pleasure! — All should be right in the world again now.

[bentbike33]

This questions is not strictly related to Freezing Saddles, but it seems to me this is the most likely thread to ask. It concerns inaccuracies in the GPS recording of my rides. My employer recently switched my smartphone from a BlackBerry to an iPhone 6. I used a freebie 3rd-party app on the BlackBerry to record rides, and the errors were generally associated with places where the GPS signal would be weakest (next to sound walls, in the Valley of the Shadow of the Two Sisters, etc.), but otherwise looked random, and over the same daily commute route, there was little variance in distance recorded. With the iPhone, I’m using the Strava app, and it seems to want to “correct” my route to keep me on streets, or put me on streets, when I am riding on the W&OD, Custis or MVT. This lengthens my commute rides by about 3%. There seem to be fewer such “corrections” when I record with the Strava app and the iPhone in Airplane Mode. Any other helpful hints (and, no, I don’t care to buy a “real GPS” for my bike)?

Thanks.

[vvill]

@bentbike33 135345 wrote:

This questions is not strictly related to Freezing Saddles, but it seems to me this is the most likely thread to ask. It concerns inaccuracies in the GPS recording of my rides. My employer recently switched my smartphone from a BlackBerry to an iPhone 6. I used a freebie 3rd-party app on the BlackBerry to record rides, and the errors were generally associated with places where the GPS signal would be weakest (next to sound walls, in the Valley of the Shadow of the Two Sisters, etc.), but otherwise looked random, and over the same daily commute route, there was little variance in distance recorded. With the iPhone, I’m using the Strava app, and it seems to want to “correct” my route to keep me on streets, or put me on streets, when I am riding on the W&OD, Custis or MVT. This lengthens my commute rides by about 3%. There seem to be fewer such “corrections” when I record with the Strava app and the iPhone in Airplane Mode. Any other helpful hints (and, no, I don’t care to buy a “real GPS” for my bike)?

I don’t use the iPhone Strava app much, but I’m guessing there’s not much you can do about it. Even with a “real GPS” you are subject to the whims of the device and the limitations of GPS. One of my bike GPS units drifts easily with any cloud/forest cover. If Airplane Mode does work better then maybe stick with that, annoying as it is.

If you are really concerned about logging too many miles for Freezing Saddles, you can go into Strava and trim your ride accordingly. (At least on a computer – not sure about in the app.) The route will still be wrong but the mileage will be better. There are also third-party websites that will let you alter a GPS track (exported from your Strava ride) which you could then re-upload to Strava. I have done that once before as my GPS drift was bad enough to remove certain segments I wanted to see from my ride.

[TwoWheelsDC]

@bentbike33 135345 wrote:

This lengthens my commute rides by about 3%. There seem to be fewer such “corrections” when I record with the Strava app and the iPhone in Airplane Mode. Any other helpful hints (and, no, I don’t care to buy a “real GPS” for my bike)?

Thanks.

Lengthens your rides compared to the Blackberry, or according to an accurate measurement? The former can probably be explained by variances in GPS tech in each phone (not sure if BB uses aGPS, but Apple does) and differences in how the app looks at your track. If you view the same track from the same device in Strava and GarminConnect, for example, you might see varied distances simply because of how the apps interpret GPS data. Strava probably interprets GPS data with a bent toward matching the track up to thoroughfares or known segments, hence the differences you’re seeing.

If your routes are lengthy compared to some other accurate measurement, I simply attribute that to inherent inaccuracies in GPS. GPS has an accuracy rate of +/- ~3 meters (IIRC), so spread that out over a long ride (at a typical sampling rate of 1 second) and you can see quite a bit of variance, even with the same device under similar conditions. Given that everyone experiences these types of inaccuracies, I think it’s probably a wash as far as BAFS scoring goes.

[wheelswings]

@TwoWheelsDC 135358 wrote:

Given that everyone experiences these types of inaccuracies, I think it’s probably a wash as far as BAFS scoring goes.

Dear bentbike33,
Given that you are an important member of the mighty Team 12 Cyclo-Paths, I believe it would be wise to follow Two-WheelsDC’s sage advice and refrain from trimming any of your rides.
Fondly,
w&w (your teammate)

[Steve O]

@bentbike33 135345 wrote:

This questions is not strictly related to Freezing Saddles, but it seems to me this is the most likely thread to ask. It concerns inaccuracies in the GPS recording of my rides.

Okay, so I know this has come up before, but the variance in elevation is just enormous. All of these are from yesterday’s Monster Cross, which presumably was the same course for everyone:

[ATTACH=CONFIG]11031[/ATTACH][ATTACH=CONFIG]11030[/ATTACH][ATTACH=CONFIG]11032[/ATTACH][ATTACH=CONFIG]11035[/ATTACH][ATTACH=CONFIG]11029[/ATTACH]

There’s a 1000 foot difference between Dave K and Joe B. And Amalitza’s is off by even more. +/- 10% seems acceptable, but wow.
(Note Amalitza’s was one lap, but 1,056 x 2=2,112)

A – No wonder your bubble is so low on the bubble chart; if you used Dave K’s device, your one lap would be 1,939ft

[Author]

Posts