Github user vanzin commented on a diff in the pull request:
https://github.com/apache/spark/pull/10205#discussion_r53716337
--- Diff: core/src/main/scala/org/apache/spark/config/ConfigEntry.scala ---
@@ -0,0 +1,247 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.spark.config
+
+import java.util.concurrent.TimeUnit
+
+import org.apache.spark.SparkConf
+import org.apache.spark.network.util.{ByteUnit, JavaUtils}
+
+/**
+ * An entry contains all meta information for a configuration.
+ *
+ * @param key the key for the configuration
+ * @param defaultValue the default value for the configuration
+ * @param valueConverter how to convert a string to the value. It should
throw an exception if the
+ * string does not have the required format.
+ * @param stringConverter how to convert a value to a string that the user
can use it as a valid
+ * string value. It's usually `toString`. But
sometimes, a custom converter
+ * is necessary. E.g., if T is List[String], `a, b,
c` is better than
+ * `List(a, b, c)`.
+ * @param doc the document for the configuration
+ * @param isPublic if this configuration is public to the user. If it's
`false`, this
+ * configuration is only used internally and we should not
expose it to the user.
+ * @tparam T the value type
+ */
+private[spark] class ConfigEntry[T] (
+ val key: String,
+ val defaultValue: Option[T],
+ val valueConverter: String => T,
+ val stringConverter: T => String,
+ val doc: String,
+ val isPublic: Boolean) {
+
+ def defaultValueString: String =
defaultValue.map(stringConverter).getOrElse("<undefined>")
+
+ /**
+ * Returns a new ConfigEntry that wraps the value in an Option, for
config entries that do
+ * not require a value.
+ */
+ def optional: OptionalConfigEntry[T] = {
+ require(!defaultValue.isDefined, s"$this has a default value, cannot
be optional.")
+ new OptionalConfigEntry(key, valueConverter, stringConverter, doc,
isPublic)
--- End diff --
I agree it may not super intuitive, but it's not only for optional configs.
`get(FOO)` may return `String` while `get(BAR)` may return a `Long` and
`get(OTHER)` might return `Option[Boolean]`. But read on.
But I disagree that having a `getOption` for `ConfEntry` objects is better.
That completely breaks the goal of this change, which is to tie a type to the
configuration entry. A configuration entry that has a default value will always
return `Some(foo)` for that method, so it doesn't really make a lot of sense to
call that method to retrieve that config's value.
An optional config has type `Option[Something]`, not `Something`, which
makes way more sense to me. With the current code, if you try to use an
optional config as if it were an actual config, you'll get a compile-time error
instead of a runtime error telling you the config option is not set.
The last point is why I think the slight awkwardness in having `get` return
different types is actually better in the end, because now you have the
compiler doing proper type checks instead of relying on your code knowing what
the type of a config is.
If you're worried about having to call `.optional` to create optional
configs, that can be changed in other ways. For example you could have
overloaded methods, one that doesn't take a default value and returns an
optional config, and one that does take a default value and returns a different
config. Or you could use a builder-style pattern (e.g.,
`ConfigEntry.intConf(...).withDefault(...)`), although now you have the
verbosity the other way.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
