Test_pedigree with internal samples

[abureau]

I have modified TestSimulateThroughPedigree and the simulate_pedigree function to specify generations of subjects as internal samples. Setting internal_sample_gen=[True, False] on line 562 specifies that founders i.e. parents of a two-generation pedigree are samples, but not their children (probands). This results in Fatal Python error: Aborted with core dump in my setup.

[jeromekelleher]

Would you mind reopening this PR from a feature branch please @abureau?

git checkout -b internal-samples
git push 

opening PRs from your fork's main generally confuses things. See the tskit dev docs for more info

[jeromekelleher]

What's happening here is that an assert has been tripped in the C code:

 Bug detected in lib/msprime.c at line 5610. Please report this issue on GitHub, ideally with a reproducible example. (https://github.com/tskit-dev/msprime/issues)
Fatal Python error: Aborted

The line it's referring to is here

I'm not sure whether this is actually important, or if it's just over-zealous asserting of stuff that's no longer relevant.

It would be good to try this example in the algorithms.py prototype, and see what it makes of it. To do this, save the .trees file from your test, and pass it in as a parameter on the CLI for algorithms.py

The msprime developer docs should help with this.

I'm away for a week now, so won't be able to look at this again for a bit.

[abureau]

I tried closing and reopening this pull request, but I saw no option to change the branch it is from. Do you mean that I need to open a new PR from my branch internal-samples @jeromekelleher ?

[jeromekelleher]

Yes, that's the idea. The dev docs should go through this

[mergify]

⚠️ The sha of the head commit of this PR conflicts with #2326. Mergify cannot evaluate rules on this PR. ⚠️