has_cache reads results

[kdkavanagh]

I think this is a new issue with version 2.0.0. The function generated by has_cache attempts to read the actual cache result into memory before returning that the call is indeed cached. This is problematic for large cached dataframes, etc.

> has_cache(getter)
function (date, arg1) 
{
    mc <- match.call()
    encl <- parent.env(environment())
    called_args <- as.list(mc)[-1]
    default_args <- encl$`_default_args`
    default_args <- default_args[setdiff(names(default_args), 
        names(called_args))]
    called_args[encl$`_omit_args`] <- NULL
    args <- c(lapply(called_args, eval, parent.frame()), lapply(default_args, 
        eval, envir = environment()))
    key <- encl$`_hash`(c(encl$`_f_hash`, args, lapply(encl$`_additional`, 
        function(x) eval(x[[2L]], environment(x)))))

    res <- encl$`_cache`$get(key)  # <---- This should probably be removed?
    return(encl$`_cache`$exists(key))

    if (res$visible) {
        res$value
    }
    else {
        invisible(res$value)
    }
}

[wch]

You're exactly right about the source of the problem -- the code that modifies the memoized function had an off-by-one error. Thanks!