BUG: radolan_test.cpp failing

[kmuehlbauer]

When I run radolan_test, the test breaks with:

FAILED:geographical coordinates of anchor point don't match

When I change the indices here to (450,450), then I get the correct results. However, the test still breaks, because the latitude values differ 1.42109e-14.

So there are two issues:

• indices, is 450, 450 correct?
• precision in comparing doubles

This is blocking conda-forge/staged-recipes#591 from adding the test.

[kmuehlbauer]

After some testing, this turns out to be a bug within the coordinate_system.cpp, IMHO.

[kmuehlbauer]

@JuergenSimon Did you have time to look into this? Is this cruzial for using libradolan, or is this just minor?

[JuergenSimon]

Hi, I had no time to look at any of this yet. I think it is a minor difference and wanted to replace the value test with a range test instead but never got round to it. Unfortunately I won't have any time in the foreseeable months for this project, sorry!

[kmuehlbauer]

@JuergenSimon Didn't want to stalk you. It would be very helpful if you just have a look, when pinging. I promis, I'll not overuse it 😀.