Formula environment causing error

[ngreifer]

See this example:

X <- rnorm(100)
Y <- rnorm(100)
F <- function(f) summ(glm(f))
F(Y ~ X)

## Error: object of type 'symbol' is not subsettable 

The traceback is:

10. notnull(x$formula) 
9. formula.default(object, env = baseenv()) 
8. formula(object, env = baseenv()) 
7. as.formula(old) 
6. update.formula(call$formula, formula) 
5. j_update(object, formula = form, weights = .weights, offset = .offset, 
    data = frame) 
4. pR2(model) 
3. summ.glm(glm(f)) 
2. summ(glm(f)) 
1. F(Y ~ X) 

I think this is occurring because the environment in which you are evaluating the formula when decomposing the glm object is not the same environment with the data. Note that this occurs even when a data set is included in glm. This isn't a huge problem but points to a possible underlying issue. The way around it is to pass the glm object to F() rather than just the formula.

[jacob-long]

Programmatic model updating is a bear. I think I've fixed it.

[ngreifer]

Yup! Thank you :)