Douglas Bates

I am building with MKL, not OpenBLAS. My `Make.user` file is ``` USE_INTEL_MKL = 1 FFLAGS = -ff2c ``` Is there a workaround in this case?

I'm a little queasy about the idea that the type of the levels of a PooledDataArray determine whether it is treated as a factor in linear predictor expressions. I do...

I think John mentioned the possibility of `unique(x)` being much less than `length(x)`, in which case the array of indices into the levels could be stored a Uint8 or Uint16,...

Thanks for including the example and the data. We'll take a look.

I wrote up some comments on the different fits as a [Quarto](https://quarto.org/) document. It can be transformed into a Jupyter notebook (`.ipynb`) or as a PDF document or a HTML...

The peculiar fixed-effects coefficients, and the estimates of the random effects standard deviations and the extreme negative value of the objective, are all related to the fact that the response...

I tried manually setting the initial values for the parameters in the model you are trying to fit and still got the message about rescaling and convergence to nonsensical estimates....

I need to rewrite the destructuring syntax in the old, tedious syntax if we are to continue to test on the LTS version.

`BSplineKit v0.12.0` requires julia v1.8 To be able to test on Julia LTS I allowed v0.11 of `BSplineKit`. This means that after contour interpolation is available, the tests will need...

If you have time to look at that, sure. I expect that the periodic splines, which were introduced in BSplineKit v0.12.0, will be needed in MixedModelsMakie but probably not in...