I introduced a stupid bug in the function that parses dates from the "run" string, i.e. chose the wrong subscripts (see issue #210). That is very unfortunate as I don't know in how many samples this wrong date is now encoded :-/
On the other hand, this is a wonderful - because its small - example why you should always have unit-tests in place, which I had not.
With this PR, you see the ideal procedure: 1) create a new test that replicates the reported wrong behaviour and leads to a failing test (29ec512) 2) fix the bug and show that tests now pass (1128d31).
I introduced a stupid bug in the function that parses dates from the "run" string, i.e. chose the wrong subscripts (see issue #210). That is very unfortunate as I don't know in how many samples this wrong date is now encoded :-/ On the other hand, this is a wonderful - because its small - example why you should always have unit-tests in place, which I had not.
With this PR, you see the ideal procedure: 1) create a new test that replicates the reported wrong behaviour and leads to a failing test (29ec512) 2) fix the bug and show that tests now pass (1128d31).