also closed #2245; test added

NEWS.md

@@ -228,7 +228,7 @@
 
 27. `integer64` sum-by-group is now properly optimized, [#1647](https://github.com/Rdatatable/data.table/issues/1647), [#3464](https://github.com/Rdatatable/data.table/issues/3464). Thanks to @mlandry22-h2o for the report.
 
-28. `setkey`/`setkeyv` would mangle the ordering of tables with memory-duplicate (i.e., sharing the same `address`) columns, [#3496](https://github.com/Rdatatable/data.table/issues/3496) and [#3766](https://github.com/Rdatatable/data.table/issues/3766). This is fixed by forcing copies of duplicated columns before reordering. Relatedly, `[` could also produce such columns when subsetting. Thanks @kirillmayantsev and @alex46015 for reporting and @jaapwalhout and @Atrebas helping to debug&isolate the issue.
+28. `setkey`/`setkeyv` would mangle the ordering of tables with memory-duplicate (i.e., sharing the same `address`) columns, [#3496](https://github.com/Rdatatable/data.table/issues/3496) and [#3766](https://github.com/Rdatatable/data.table/issues/3766). This is fixed by forcing copies of duplicated columns before reordering. Relatedly, `[` could also produce such columns when subsetting. Finally, `[` might leave its output with `.data.table.locked`, preventing downstream use of `:=`, [#2245](https://github.com/Rdatatable/data.table/issues/2245). Thanks @kirillmayantsev, @alex46015, and @grayskripko for reporting and @jaapwalhout and @Atrebas helping to debug&isolate the issue.
 
 #### NOTES
 

inst/tests/tests.Rraw

@@ -15680,6 +15680,9 @@ x$b = x$a
 setDT(x)
 test(2078.2, setkey(x, a, verbose=TRUE), output='Deep duplicate columns')
 test(2078.3, x, data.table(a=paste0(1:2), b=paste0(1:2), key='a'))
+## #2245 unset .data.table.locked even if is.null(irows)
+x <- data.table(a=c(0.85, -0.38, 1.19), b=c(0.56, 0.63, -1.30))
+test(2078.4, x[ , round(.SD, 1)][ , c := 8], data.table(a=c(.8, -.4, 1.2), b=c(.6,.6,-1.3), c=8))
 
 
 ###################################