Various glitches

[ekstroem]

This is more like a to do list/reminder of various glitches that we need to fix/address

1. A dataargument is needed for glm output to work with publish. For example this doesn't work

   set.seed(1)
   x <- rnorm(100)
   y <- rnorm(100)
   g <- factor(rbinom(100, size=2, p=.5))
   df <- data.frame(x, y, g)
   
   m <- glm(y ~ g*x, family=gaussian)
   publish(m)

   while this does:

   m <- glm(y ~ g*x, family=gaussian, data=df)
   publish(m)

2. Add a way to allow for other than 95% confidence intervals of parameters?

[ekstroem]

And while I remember it:

Would it be a good idea to replace anova with oneway.test for Gaussian/numeric data?

[ekstroem]

And yet another idea .... it would be nice if univariateTable played nicely with knitr::kable