R/formula.util.q

In lasso2: L1 Constrained Estimation aka `lasso'

###  Copyright (C) 1998
###  Berwin A. Turlach <bturlach@stats.adelaide.edu.au>
###  Bill Venables <wvenable@stats.adelaide.edu.au>
### --> ../COPYRIGHT for more details

is.formula <- function(x) inherits(x, "formula")

merge.formula <- function(x, y, ...)
{
    if(!is.formula(x) || length(x) != 3)
        stop("First argument is invalid")
    if(!is.formula(y)) stop("Second argument is invalid")
    if(length(list(...))) warning("extraneous arguments discarded")
    is.gEnv <- function(e) identical(e, .GlobalEnv)

    str <- paste(c(deparse(x[[2]]), "~",
                   deparse(x[[3]]), "+",
                   deparse(y[[length(y)]])), collapse = "")
    f <- as.formula(str)
    ## MM: try to keep environment (where reasonable)
    ex <- environment(x)
    ey <- environment(y)
    if(!is.gEnv(ex)) {
        environment(f) <- ex
        if(!is.gEnv(ey) && !identical(ex,ey)) {
            warning("`x' and `y' have different environments; x's is used")
        }
    } else if(!is.gEnv(ey))
        environment(f) <- ey
    f
}