remove dependency to tibble

[javierluraschi]

Consider removing tibble dependency. Two PR are out for supporting R 3.0.0 in tibble and lazyeval:

hadley/lazyeval#87
tidyverse/tibble#187

However, we could take the code out from tibble if needed as well.

[kevinushey]

I wonder if instead of making this a hard error we could just use the default print method (and maybe warn once per session)?

[javierluraschi]

@kevinushey how do we usually warn once per session? By setting a global or there is a nicer way for doing this?

[kevinushey]

A slightly cleaner implementation would be with a factory function e.g.

make_warning <- function(message) {
    message <- message
    warned <- FALSE
    function() {
        if (warned) return()
        warning(message)
        warned <<- TRUE
    }
}

tibble_warning <- make_warning("...")
tibble_warning()

[jjallaire]

I actually think that if they ask for tibble and we can't provide it then we are just fine to error (I think this will be a very narrow case as tibble is installed with dplyr and won't be a common choice given that pagedtable is available).

[javierluraschi]

kept as is.

[jjallaire]

@kevinushey and @hadley Are we okay to have a package private S3 method here or could that interfere somehow with the tibble exported method of the same name? We could just rename it to is_s3_vector and/or just implement it with a switch/case.

[javierluraschi]

@jjallaire just in case, renamed.

[hadley]

I'd need to experiment to check - S3 + namespaces is still a little hairy. Definitely safer to rename.

[jjallaire]

Yes, I'd say let's not use S3 here.

On Thu, Nov 3, 2016 at 7:20 AM, Hadley Wickham notifications@github.com
wrote:

@hadley commented on this pull request.

In R/html_paged.R #853:

• dim_desc <- function(x) {
• dim <- dim(x) %||% length(x)
• format_dim <- vapply(dim, big_mark, character(1))
• format_dim[is.na(dim)] <- "??"
• paste0(format_dim, collapse = " \u00d7 ")
• }

• is_atomic <- function(x) {
• is.atomic(x) && !is.null(x)
• }

• is_vector <- function(x) {
• is_atomic(x) || is.list(x)
• }

• is_vector_s3 <- function(x) UseMethod("is_vector_s3")

I'd need to experiment to check - S3 + namespaces is still a little hairy.
Definitely safer to rename.

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
#853, or mute the thread
https://github.com/notifications/unsubscribe-auth/AAGXx8QnW4HgCmDURW-5BnbHIhlj67jTks5q6cOUgaJpZM4Kn1E8
.

[javierluraschi]

K, S3 removed.

[krlmlr]

FWIW, I'm planning to support R 3.0.0 in tibble 1.3.0 which will be out soon.