josephmure

These comments are: https://github.com/openturns/openturns/pull/2451#discussion_r1549659570 and https://github.com/openturns/openturns/pull/2451#discussion_r1549321628

This is not really related (except it concerns ChiSquared), but is there a reason the Python syntax is different whether we are doing the hypothesis test or the fitting test...

- We could add some way to automatically detect relevant ResourceMap keys in the doc of the various classes. - If the openturns.conf.in file could be automatically generated from the...

Let's document it and maybe change the name (suggestion : dNonCentralStudentExplicitImplementation)

https://github.com/openturns/openturns/blob/4e02123b94c67f930958b15febc1884c89946e2b/lib/src/Uncertainty/Distribution/StudentFunctions.cxx#L538

I agree with Julien, this change would make the maintenance of factories easier

Here is the relevant line: https://github.com/openturns/openturns/blob/4f3d29c83f6f37b6c6e07ddfecceac225066ffc4/lib/src/Uncertainty/Model/DistributionImplementation.cxx#L2440 And quantileEpsilon_ is by default set to 1e-12 in the ResourceMap: Distribution-DefaultQuantileEpsilon | 1e-12 | float

Another weird thing: the same quantileEpsilon_ attribute is also used further down the code, as tolerance for the x values instead of the cdfs: https://github.com/openturns/openturns/blob/4f3d29c83f6f37b6c6e07ddfecceac225066ffc4/lib/src/Uncertainty/Model/DistributionImplementation.cxx#L2484

You need to rebase (again) on master to get build_linux and build_mingw to work. After that, I think the PR could be marked as ready for review.

@mbaudin47 Could you please take over this PR? A rebase is needed and you still need to do your review, so maybe the best way to proceed is that you...