Johannes Terblanche

Yes I agree. The bad case is if the names are variable. The functions will compile a new method for each name.

It will be a tradeoff between compile time and runtime. Every new NamedTuple will require a new method to be compiled.

This tip is relevant here: https://docs.julialang.org/en/v1/manual/performance-tips/#The-dangers-of-abusing-multiple-dispatch-(aka,-more-on-types-with-values-as-parameters)

I checked v0.28 and it was failing intermittently there as well, so think we can safely eliminate the recent `ArrayPartition` and Manifolds v0.8 upgrade.

From #1580, it looks like this will only be done after #1010

Ok, just saw it is caused by `Prior(I)` I'm changing the issue to the question: What is the correct way of defining the mechanics in a mixture? For now the...

Also, should we perhaps use the nullhypo (spreadNH) case in initialization: ```julia certainidx, allelements, activehypo, mhidx = IIF.assembleHypothesesElements!(mh, N, 1, 3, Bool[0, 0, 1]) #currently: #samples for 0: 0.0%

I didn’t miss it in the discussion. I understand that part. The issue is even if you give the probability of 0.999-0.001 it wil go to 33% nullhypo and 66%...

This subfg from a range only example.  Results in these 2 factors after deconvolution.  x0 and x1 are connected by 2 range only factors and the differential factor...

For now `useMsgLikelihoods=false` should be used for better performance `useMsgLikelihoods=true`  `useMsgLikelihoods=false`