Closes #688. dast preserves attributes except when value.var is fac…

…tor/ordered factor.

README.md

@@ -213,6 +213,9 @@ DT[, list(.N, mean(y), sum(y)), by=x] # 1.9.3+ - will use GForce.
 
   *  `dcast.data.table` provides better error message when `fun.aggregate` is specified but it returns length != 1. Closes git [#693](https://github.com/Rdatatable/data.table/issues/35). Thanks to Trevor Alexander for reporting [here on SO](http://stackoverflow.com/questions/24152733/undocumented-error-in-dcast-data-table).
 
+  *  `dcast.data.table` tries to preserve attributes whereever possible, except when `value.var` is a `factor` (or ordered factor). For `factor` types, the casted columns will be coerced to type `character` thereby losing the `levels` attribute. Closes git [#688](https://github.com/Rdatatable/data.table/issues/688). Thanks to juancentro for reporting.
+
+
 #### NOTES
 
   *  Reminder: using `rolltolast` still works but since v1.9.2 now issues the following warning :

inst/tests/tests.Rraw

@@ -4674,6 +4674,10 @@ test(1313.22, DT[, list(y=max(y, na.rm=TRUE)), by=x], DT[c(5,10)])
 dt <- data.table(x=c(1,1), y=c(2,2), z = 3:4)
 test(1314, dcast.data.table(dt, x ~ y, value.var="z", fun.aggregate=identity), error="Aggregating function provided to argument 'fun.aggregate' should return a length 1")
 
+# bug #688 - preserving attributes
+DT = data.table(id = c(1,1,2,2), ty = c("a","b","a","b"), da = as.Date("2014-06-20"))
+test(1315, dcast.data.table(DT, formula = id ~ ty, value.var="da"), data.table(id=c(1,2), a=as.Date("2014-06-20"), b=as.Date("2014-06-20"), key="id")) 
+
 ##########################
 
 # TO DO: Add test for fixed bug #5519 - dcast.data.table returned error when a package imported data.table, but dint happen when "depends" on data.table. This is fixed (commit 1263 v1.9.3), but not sure how to add test.

man/dcast.data.table.Rd

@@ -38,7 +38,7 @@ If the formula's LHS contains the same column more than once, ex: \code{dcast.da
 
 Names for columns that are being cast are generated in the same order (separated by a \code{_}) from the (unique) values in each column mentioned in the formula RHS.
 
-The only feature that's not implemented from \code{reshape2:::dcast} *yet* is the argument \code{margins}. 
+From \code{v1.9.3}, \code{dcast.data.table} tries to preserve attributes whereever possible, except when \code{value.var} is a factor (or ordered factor). For \code{factor} types, the resulting casted columns will be coerced to character type thereby losing \code{levels} attribute.
 }
 \value{
     A keyed \code{data.table} that has been cast. The key columns are equal to the variables in the \code{formula} LHS in the same order.

src/fcast.c

@@ -584,6 +584,7 @@ SEXP fcast(SEXP DT, SEXP inames, SEXP mnames, SEXP vnames, SEXP fill, SEXP fill_
             break;
             default : error("Unknown 'value' column type '%s'", type2char(TYPEOF(tmp)));
         }
+        copyMostAttrib(vdt, tmp);
         UNPROTECT(1); // tmp
         SET_VECTOR_ELT(ans, i+ilen, tmp);
     }