value() returns a vector instead of a list

[fangly]

Hi Christopher,
I have stumbled across a small issue with the hash package:

h <- hash(c("a","b","c"), list("bar", 1L, TRUE))
print(h$a); print(h$b); print(h$c)
res <- values(h)
print(class(res)); print(res)

As you see, the list of values is stored properly in the hash, but instead of returning a list, values() returns a vector.
Cheers,
Florent

PS/ This is for hash version 2.2.6 (the latest on CRAN)

[ctbrown]

I believe you are right. This is a problem. But the docs show the solution. By default values tries to simplify the response since it raps sapply. Will get the behavior you want.

     values( h, simplify=FALSE ) 

If there is no objection, I think the best thing to due is either:

• make simplify=FALSE the default
• or make simplify=TRUE only valid when all the values are of the same class.

Thoughs?

[fangly]

You are entirely right, Chris. I must have overlooked the ... argument passed to sapply(), which is a function regarded by some as potentially dangerous.

Personally, I like the principle of least surprise, and would be happy if the return value was always a named list, i.e. have simplify=FALSE (or even no "simplify" option).

Here is a suggestion if you want to keep the "simplify" option:

• Use lapply() instead of vapply(). You get a list "x" back.
• If simplify=TRUE, then convert to vector if all keys have the same type using

types <- unique(vapply(x, typeof, character(1L)), nmax=10L)
atomicTypes <- c("logical", "integer", "double", "complex", "character", "raw")
if (length(types) == 1L && types %in% atomicTypes) {
    x <- unlist(x)
}

[ctbrown]

I am went with always returns a named list. The user can simplify it themselves with `unlist` if needed. Thanks for the input. C-

…

On Fri, Mar 3, 2017 at 2:15 AM, Florent Angly ***@***.***> wrote: You are entirely right, Chris. I must have overlooked the ... argument passed to sapply(), which is a function regarded by some as potentially dangerous <r-lib/lintr#149>. Personally, I like the principle of least surprise, and would be happy if the return value was always a named list, i.e. have simplify=FALSE (or even no "simplify" option). Here is a suggestion if you want to keep the "simplify" option: - Use lapply() instead of vapply(). You get a list "x" back. - If simplify=TRUE, then convert to vector if all keys have the same type using types <- unique(vapply(x, typeof, character(1L)), nmax=10L)atomicTypes <- c("logical", "integer", "double", "complex", "character", "raw")if (length(types) == 1L && types %in% atomicTypes) { x <- unlist(x) } — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#13 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AC5xa8a6H7NRPB3ZMDl6YcN77OWnJCmcks5rh-grgaJpZM4KHwS-> .

[ctbrown]

This was resolved in 2017.