Use fast string containers by default - they are much faster than the original string container when building datastructure.

Author: cnuernber

src/tech/v3/dataset/impl/column_base.clj

@@ -76,7 +76,7 @@
    (if-let [rv (get-in options [:datatype-parsers dtype])]
      (rv dtype options)
      (case dtype
-       :string (str-table/make-string-table 0 "")
+       :string (str-table/fast-string-container)
        :text
        (let [^IMutList list-data
              (try

src/tech/v3/dataset/io/column_parsers.clj

@@ -451,8 +451,8 @@
             (.add container parsed-value)
             (catch Exception e
               (throw (RuntimeException. (str "Parse failure of datatype: "
-                                             (dtype/elemwise-datatype container)
-                                             e)))))))))
+                                             (dtype/elemwise-datatype container))
+                                        e))))))))
   (finalize [_p rowcount]
     (finalize-parser-data! container missing nil nil missing-value rowcount)))
 

src/tech/v3/dataset/string_table.clj

@@ -191,6 +191,8 @@
 
 (defn fast-string-container
   ([str->int int->str]
+   (when-not (instance? java.util.HashMap str->int)
+     (throw (RuntimeException. "Invalid creation of fast string container")))
    (FastStringContainer. (hamf/long-array-list)
                          int->str str->int))
   ([]