better error handling in join functions. closes #371

NEWS.md

@@ -76,6 +76,10 @@
 
 * internal `DataDots` class protects against missing variables in verbs (#314). 
 
+* join functions throws error instead of crashing when there are no common
+  variables between the data frames, and also give a better error message when
+  only one data frame has a by variable (#371). 
+
 # dplyr 0.1.3
 
 ## Bug fixes

inst/include/dplyr/DataFrameJoinVisitors.h

@@ -16,7 +16,13 @@ namespace dplyr{
             std::string name ;
             for( int i=0; i<nvisitors; i++){
                 name = names_[i] ;
-                visitors[i] = join_visitor( left[name], right[name], name) ;
+                try{
+                    visitors[i] = join_visitor( left[name], right[name], name) ;
+                } catch( ... ){
+                    std::stringstream s ;
+                    s << "cannot join on column '" << name << "'" ;
+                    stop(s.str());
+                }
             }
         }
 

inst/tests/test-joins.r

@@ -169,3 +169,18 @@ test_that("indices don't get mixed up when nrow(x) > nrow(y). #365",{
   expect_equal( res$V2, c("a", "b"))
   expect_equal( res$V3, c("n", "m"))
 })
+
+test_that("join functions error on column not found #371", {
+  expect_error( 
+    left_join(data.frame(x=1:5), data.frame(y=1:5), by="x"), 
+    "cannot join on column 'x'"
+  )
+  expect_error( 
+    left_join(data.frame(x=1:5), data.frame(y=1:5), by="y"), 
+    "cannot join on column 'y'"
+  )
+  expect_error(
+    left_join(data.frame(x=1:5), data.frame(y=1:5)), 
+    "no common variables"
+  )
+})