terms.formula(y ~ (x + fn(z = NA)) - fn(z = NA), simplify=TRUE) does not simplify

Dear all,

I'm trying to use update.formula for removing a function in formula, but when the function has argument set as NA, this doesn't seem to work. I was able to track the issue to terms.formula function (.Internal(terms.formula(..))), but couldn't figure out anything useful from corresponding C code in src/main/model.c. Here's an example of the issue:

rm(list=ls())
tmp<-terms.formula(y ~ (x + fn(z = NA)) - fn(z = NA),simplify=TRUE)
tmp

y ~ x + fn(z = NA)
attr(,"variables")
list(y, x, fn(z = NA), fn(z = NA))
attr(,"factors")
           x fn(z = NA)
y          0          0
x          1          0
fn(z = NA) 0          1
fn(z = NA) 0          0
attr(,"term.labels")
[1] "x"          "fn(z = NA)"
attr(,"order")
[1] 1 1
attr(,"intercept")
[1] 1
attr(,"response")
[1] 1
attr(,".Environment")
<environment: R_GlobalEnv>

formula(tmp)

y ~ x + fn(z = NA)

with fn(z=a) it works fine, and also with NA_character_ (but not with NA_real_):

rm(list=ls())
tmp<-terms.formula(y ~ (x + fn(z = NA_character_)) - fn(z = NA_character_),simplify=TRUE)
tmp

y ~ x
attr(,"variables")
list(y, x, fn(z = NA_character_))
attr(,"factors")
           x
y          0
x          1
fn(z = NA) 0
attr(,"term.labels")
[1] "x"
attr(,"order")
[1] 1
attr(,"intercept")
[1] 1
attr(,"response")
[1] 1
attr(,".Environment")
<environment: R_GlobalEnv>

formula(tmp)

y ~ x

Is this what is supposed to happen?

Best regards,

Jouni Helske