tests for counting edges

[Wikunia]

Before continuing with #53 I think it would be good to have these test cases. I assumed that the number is correct here just wanted to add a test such that we see when this changes. Not sure how to better test this.

[dkarrasch]

Awesome. Could you please rebase onto current master, so that the github action CI stuff runs on this PR.

You could try and see if something like length(voronoiedges(tess)) works and is equal to the result of your loop. That should be an independent "stepping through" test. Could you please also fix the random seed in the "point can be in only one triangle" test. That test turned out to give spurious errors, probably when the test point was lying on some edge.

[dkarrasch]

With #53 we won't need the collect in the tests anymore, right?

[codecov[bot]]

Codecov Report

Merging #57 (235cf75) into master (63898ec) will increase coverage by 15.05%. The diff coverage is n/a.

@@             Coverage Diff             @@
##           master      #57       +/-   ##
===========================================
+ Coverage   72.20%   87.26%   +15.05%     
===========================================
  Files           1        1               
  Lines         421      424        +3     
===========================================
+ Hits          304      370       +66     
+ Misses        117       54       -63     

Impacted Files	Coverage Δ
src/VoronoiDelaunay.jl	87.26% <0.00%> (+15.05%)	:arrow_up:

---

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update 63898ec...235cf75. Read the comment docs.

[Wikunia]

Yes that might be true though I'm not 100% how iterators handle length :smile:

[Wikunia]

Seems like there is a difference in the delaunay edges between Julia 1.6 and 1.7 though? :confused:

[dkarrasch]

Hm, I believe we've had changes in the random number generator between v1.6 and v1.7 or so. We should perhaps try to avoid hard-coding numbers, and instead compare two different implementations that we expect to give consistent results.

[Wikunia]

Oh yeah that's true maybe have a file with a list of points that we can read and use those as our test case?

[dkarrasch]

We don't need a big test. Perhaps you could manually type a couple of points, say 5, at distinct locations, and then we know everything about that specific case. It is only important to step through the iterator, even if its length is short.

[Wikunia]

True, will do

[Wikunia]

Some lines have a length of zero though:

Here you can see the 12 visible voronoi edges.

[Wikunia]

With #53 we won't need the collect in the tests anymore, right?

Actually we do because we don't provide the length of the iterator at the moment. We could walk through all and count and return that though if we want to.