Closes #899. rbindlist combines factor levels properly.

README.md

@@ -47,6 +47,8 @@
 
   14. `subset` handles extracting duplicate columns in consistency with data.table's rule - if a column name is duplicated, then accessing that column using column number should return that column, whereas accessing by column name (due to ambiguity) will always extract the first column. Closes [#891](https://github.com/Rdatatable/data.table/issues/891). Thanks to @jjzz.
 
+  15. `rbindlist` handles combining levels of data.tables with both ordered and unordered factor columns properly. Closes [#899](https://github.com/Rdatatable/data.table/issues/899). Thanks to @ChristK.
+
 #### NOTES
 
   1. Clearer explanation of what `duplicated()` does (borrowed from base). Thanks to @matthieugomez for pointing out. Closes [#872](https://github.com/Rdatatable/data.table/issues/872).

inst/tests/tests.Rraw

@@ -5408,6 +5408,13 @@ test(1391.2, subset(DT, select=c("V2", "V1")), DT[, c("V2", "V1"), with=FALSE])
 DT = data.table(x=sample(c(1:2, NA), 30, TRUE), y=sample(c(1:5, NA, NaN), 30, TRUE))
 test(1392, na.omit(DT), DT[!is.na(x) & !is.na(y)])
 
+# Fix for #899. Mix of ordered and normal factors where normal factors in more than 1 data.table has identical levels.
+DT1 = data.table(A = factor(INT(7,8,7,8,7)), B = factor(6:10), C = 0)
+DT2 = data.table(D = ordered(1:5), A = factor(INT(1:2,1:2,1L)), C = 0)
+DT3 = data.table(A = factor(INT(7:8)), C = 0)
+ans = data.table(A=factor(INT(7,8,7,8,7,1,2,1,2,1,7,8), levels=c("7", "8", "1", "2")), B=factor(INT(6:10, rep(NA,7))), C=0, D=ordered(INT(rep(NA,5), 1:5, rep(NA,2))))
+test(1393, rbindlist(list(DT1, DT2, DT3), fill = TRUE), ans)
+
 ##########################
 
 

src/rbindlist.c

@@ -277,6 +277,9 @@ static SEXP combineFactorLevels(SEXP factorLevels, int * factorType, Rboolean *
                 while (h[idx] != NULL) {
                     pl = h[idx];
                     if (data.equal(VECTOR_ELT(factorLevels, pl->i), pl->j, elem, j)) {
+                        // Fixes #899. "rest" can have identical levels in 
+                        // more than 1 data.table.
+                        if (!(pl->i == i && pl->j == j)) break;
                         record = TRUE;
                         do {
                             // if this element was in an ordered list, it's been recorded already