Spark - deprecated registerTempTable() function

In Apache Spark, the function registerTempTable() was an old API (deprecated since Spark 2.0 and removed in Spark 3.0) that allowed you to register a DataFrame as a temporary table (view) so it could be queried using SQL syntax.

---

📌 Usage (before Spark 2.0)

df = spark.read \
    .format("jdbc") \
    .option("url", "jdbc:postgresql://10.0.0.1:3433/zzhdb") \
    .option("dbtable", "people") \
    .option("user", "zzh") \
    .option("password", "pwd") \
    .load()# Register as a temporary table
df.registerTempTable("people")

# Now you can query it using Spark SQL
result = spark.sql("SELECT name, age FROM people WHERE age > 20")
result.show()

 

---

🚨 Deprecation

• registerTempTable("name") was replaced with createOrReplaceTempView("name").

• Reason: registerTempTable only worked with SQLContext and was being phased out in favor of SparkSession unified APIs.

---

📌 Modern Equivalent

df.createOrReplaceTempView("people")

result = spark.sql("SELECT name, age FROM people WHERE age > 20")
result.show()

 

---

📌 Difference Between Views

1. createOrReplaceTempView(name)

   • Session-scoped (disappears when the SparkSession ends).

   • Can be replaced if you register the same name again.

2. createGlobalTempView(name)

   • Global across sessions but tied to the Spark application.

   • Registered under the global_temp database.

    

   SELECT * FROM global_temp.people;

---

✅ In short:
registerTempTable() = old API (now gone).
createOrReplaceTempView() = the correct replacement.