Exponential to Pareto Test

[mkhattab940]

### Failure in: 6:RoundTrip:0:6:t_exponential_to_pareto:1
haskell/Tests/TestTools.hs:130
expected:
X <~ uniform(+0/1, +1/1)
return 3/2 * real2prob(X) ** (-3/2)
but got:
X3 <~ gamma(1/1, 2/3)
return exp(prob2real(X3)) * (2/3)

Cases: 182  Tried: 97  Errors: 0  Failures: 9
Cases: 182  Tried: 98  Errors: 0  Failures: 9

[JacquesCarette]

This is getting really tiresome... too late now, but you've somehow managed to create all your PRs so that every single one of them causes conflits, and so has to be hand-merged. You need to get your use of git updated, else your co-workers are going to hate you...

[mkhattab940]

I've been struggling to avoid these merge conflicts, because they don't exist until you merge another branch that adds a test to the same set in RoundTrip.hs. I've tried my best to minimize these kinds of merge conflicts, but it seems to be a problem inherent with merging multiple branches that write to the same file (and in the case of adding single tests, to the same line). Is it possible to avoid a merge conflict in this scenario?

[JacquesCarette]

I think you need to build these incrementally, rather than build them all from a really old base.

[mkhattab940]

All of our pull requests should have all the latest commits that were on hakaru-dev:master at the time of submission. I've been making sure of this since investigating how the Erlang tests ended up going missing.

[JacquesCarette]

Right - that's not the issue. The issue is that your commits do not seem to 'accumulate' well, one on top of another.

There's probably weird issues to do with line endings mixed in there as well. And editor settings that introduce a trailing blank when you edit, which causes a spurious diff that throws things off.

[mkhattab940]

So could it be an issue introduced by my text editor, then?

[JacquesCarette]

It could be.