Errata in Activity 3.01

[roboli]

I got different results because your code is different from that from the book:

(def walking-speed 4)

Book says:

... and an average of walking speed of 5 km per hour.

(defn distance
	"Returns a rough estimate of the distance between two coordinate points, in kilometers. Works better with smaller distance"
	[{lat1 :lat lon1 :lon} {lat2 :lat lon2 :lon}]
	(let [deglen 110.25
				x (- lat2 lat1)
				y (* (Math/cos lat2) (- lon2 lon1))]
				(* deglen (Math/sqrt (+ (* y y) (* x x))))))

Book has:

110.25 * sqrt((lat2 - lat1)^2 + cos(lat1) * (lon2 - lon1)^2)

Please notice that has cos(lat1) not cos(lat2)

[dsletten]

Roberto, you only identified the smallest problem with this terrible code. Yes, they switched from cos(lat1) to cos(lat2). But they shouldn't be computing cos() of either. Java's Math.cos() method takes radian arguments, not degrees of latitude. You have to first call Math.toRadians(): Math.cos(Math.toRadians(lat1))
Furthermore, they are squaring the cos() factor in (* y y) above, so the result is (cos(lat2)*(lon2-lon1))^2 not what the formula in the book shows at all...

[twoe-hub]

Bitten by it. However, thanks to @dsletten for, Math.cos(Math.toRadians(lat1)); never occurred to me.

[mrwizard82d1]

I, too, just encountered this issue.

Any ideas on the actual solution. The internet seemed to recommend different approaches:

• Take the cosine of lat1
• Take the cosine of lat2
• Take the cosine of the average latitude, (/ (+ lat1 lat2) 2)

None of these approaches produce the result in "expected output" of the itinerary function printed in the book. From the book source code, that calculation squares the cosine of lat1 but the printed formula does not squire the cosine. This approach seems similar to what Google AI recommends - except that Google AI averages the two latitudes.

Sigh...

Help.