[SOLVED] Alternatives to reshape2::melt() for matrices with named rows/columns?

Alternatives to reshape2::melt() for matrices with named rows/columns?

The melt() function from the reshape2 package has a convenient behavior where it will convert a matrix with named rows/columns into a three-column data frame with the matrix row/column names becoming the first two columns:

testmat <- matrix(1:25,nrow=5,dimnames=list(LETTERS[6:10],LETTERS[1:5]))
#   A  B  C  D  E
# F 1  6 11 16 21
# G 2  7 12 17 22
# H 3  8 13 18 23
# I 4  9 14 19 24
# J 5 10 15 20 25

longmat <- reshape2::melt(testmat)
head(longmat)
# Var1 Var2 value
#    F    A     1
#    G    A     2
#    H    A     3
#    I    A     4
#    J    A     5
#    F    B     6

But it has the annoying behavior of making Var1 and Var2 factors, and it does not support a stringsAsFactors=F option. The README for reshape2 says "reshape2 is retired: only changes necessary to keep it on CRAN will be made. We recommend using tidyr instead.", so it seems unlikely this behavior will ever be changed in melt().

I have experimented with alternatives, but haven't found something that works yet:

tidyr::pivot_longer() will not take a matrix as input. Even if I force the matrix to become a data.frame with tidyr::pivot_longer(as.data.frame(testmat),cols=all_of(colnames(testmat))), the matrix row names are discarded instead of captured in a column in the result.
data.table::melt() just redirects to reshape2::melt().

Main question: Is there some other function (tidyverse or non-tidyverse) that will convert a matrix to long form in a manner similar to reshape2::melt()?

Second question: I've been experimenting with writing my own. Are there any hidden "gotchas" I haven't thought of that might make the function below produce unexpected/incorrect behaviors?

melt_by_hand <- function(mat) {
  return(data.frame(row=rep(rownames(mat),ncol(mat)),
                      col=rep(colnames(mat),each=nrow(mat)),
                      value=as.vector(mat)))
}

Solution

Nothing else but as.data.frame.table from base R should help

> as.data.frame.table(testmat, stringsAsFactors = FALSE)
   Var1 Var2 Freq
1     F    A    1
2     G    A    2
3     H    A    3
4     I    A    4
5     J    A    5
6     F    B    6
7     G    B    7
8     H    B    8
9     I    B    9
10    J    B   10
11    F    C   11
12    G    C   12
13    H    C   13
14    I    C   14
15    J    C   15
16    F    D   16
17    G    D   17
18    H    D   18
19    I    D   19
20    J    D   20
21    F    E   21
22    G    E   22
23    H    E   23
24    I    E   24
25    J    E   25