Proposal: Indexing access syntax for Lists and Maps

Shams Mahmood shams.mahmood at gmail.com
Mon Mar 30 06:32:04 PDT 2009


The proposed syntax for Set seems counter intuitive to me.
While in the case of Arrays, Lists and Maps it will be to access the
elements
the Set ones will perform a contains check. Perhaps this should be proposed
in a separate
proposal with a different operator. e.g. set?[x]

---------- Forwarded message ----------
From: Tim Keith <tim.keith at gmail.com>
To: coin-dev at openjdk.java.net
Date: Sun, 29 Mar 2009 22:40:31 -0700
Subject: Re: Proposal: Indexing access syntax for Lists and Maps
Is it possible to include Set as well?

E.g.  "bool = set[x]"  meaning  "bool = set.contains(x)"
and "set[x] = bool"  meaning  "bool ? set.add(x) : set.remove(x)"

-- Tim

On Sun, Mar 29, 2009 at 5:12 PM, Shams Mahmood <shams.mahmood at gmail.com>
wrote:

> Indexing access syntax for Lists and Maps
>
> VERSION
> This is version 1.0.
>
> AUTHOR(S):
> Shams Mahmood Imam
>
> OVERVIEW
>
> FEATURE SUMMARY:
> Collection classes are among the most frequently used in the Java SDK.
> Currently Lists and Maps do not provide any additional language
> feature to access individual elements unlike Arrays. This proposal
> aims to provide these Collection citizens of java additional
> language support to access elements like Arrays have currently.
>
> MAJOR ADVANTAGE:
> Will provide a consistent syntax for accessing elements of Arrays,
> Lists and Maps. In addition, the language grammar will not change
> much since the subscript operator is already supported for Arrays.
>
> MAJOR BENEFIT:
> Apart from the consistency mentioned above, implementation fo this
> feature will result in fewer characters needed to be typed to achieve
> simple access to elements in Maps/Lists.
>
> MAJOR DISADVANTAGE:
> Like the for-each loop construct, it will expose the client to
> NullPointerException(NPE)s when used with a null List/Map. However,
> this shouldn't be such a major issue as NPEs are also generated
> by arrays when the operator is used in a null array.
>
> ALTERNATIVES:
> The comparatively more verbose get/set methods for Lists and get/put
> methods for Maps.
>
> EXAMPLES
>
> SIMPLE EXAMPLE:
>
> public class Main {
>  public static void main(String[] arguments) {
>    List<String> l1 = Arrays.asList(new String[] {"a", "b", "c"});
>    String firstElement = l1[0];
>    Map<Integer, String> m1 = new HashMap<Integer, String>(4);
>    m1[Integer.valueOf(1)] = "One";
>  }
> }
>
> ADVANCED EXAMPLE:
>
> public class Main {
>  public static void main(String[] arguments) {
>    List<String> l1 = Arrays.asList(new String[] {"a", "b", "c"});
>    Map<Integer, String> m1 = new HashMap<Integer, String>(4);
>    Map<String, Integer> m2 = new HashMap<String, Integer>(4);
>
>    m2[l1[2]] = m2[m1[1]] = 4; // same as m2.put(l1.get(2),
> m2.put(m1.get(1), 4));
>  }
> }
>
> DETAILS
>
> SPECIFICATION:
> Java Language Specification changes:
>
> 15.29 (NEW CHAPTER): Collection Access Expressions
> A collection access expression contains two subexpressions, the List/Map
> reference expression (before the left bracket) and the index expression
> (within the brackets). Note that the List/Map reference expression may be
a
> name or any expression that evaluates to a List/Map. The index experssion
is
> expected to evaluate to an int for Lists and a valid key type for Maps.
>
> CollectionAccess:
>  Expression [ Expression ]
>
> 15.8 Primary Expressions
> original:
> ---------
> PrimaryNoNewArray:
>  Literal
>  Type . class
>  void . class
>  this
>  ClassName.this
>  ( Expression )
>  ClassInstanceCreationExpression
>  FieldAccess
>  MethodInvocation
>  ArrayAccess
>
> replaced with:
> --------------
> PrimaryNoNewArray:
>  Literal
>  Type . class
>  void . class
>  this
>  ClassName.this
>  ( Expression )
>  ClassInstanceCreationExpression
>  FieldAccess
>  MethodInvocation
>  ArrayAccess
>  CollectionAccess
>
> 15.26 Assignment Operators
> original:
> ---------
> LeftHandSide:
>  ExpressionName
>  FieldAccess
>  ArrayAccess
>
> replaced with:
> --------------
> LeftHandSide:
>  ExpressionName
>  FieldAccess
>  ArrayAccess
>  CollectionAccess
>
>
> COMPILATION:
>
> After successful creation of the AST handling the additional grammar for
> Collection Access expressions, the syntactic sugar will be replaced by
> JDK1.4 compatible code during the Code Generation phase. This is
consistent
> with how JDK5.0 constructs like the for-each loop is handled by the
> compiler.
>
> e.g.
> public class TestConcise {
>  public static void main(String[] args) {
>    java.util.Map<Integer, String> m1 = new java.util.HashMap<Integer,
> String>();
>    m1[2] = "two";
>
>    java.util.LinkedList<String> l1 = java.util.Arrays.asList( new String[]
> {"a", "b", "c" });
>    m1[3] = l1[2];
>    l1[0] = m1[0];
>    l1[1] = "one";
>  }
> }
>
> is converted to
> public class TestConcise {
>  public TestConcise() {
>    super();
>  }
>
>  public static void main(String[] args) {
>    java.util.Map<Integer, String> m1 = new java.util.HashMap<Integer,
> String>();
>    m1.put(Integer.valueOf(2), "two");
>
>    java.util.LinkedList<String> l1 = java.util.Arrays.asList( new String[]
> {"a", "b", "c" });
>    m1.put(Integer.valueOf(3), l1.get(2));
>    l1.set(0, m1.get(Integer.valueOf(0)));
>    l1.set(1, "one");
>  }
> }
>
>
> TESTING:
>
> LIBRARY SUPPORT:
> No additional library support is needed.
>
> REFLECTIVE APIS:
> This proposal does not require any reflective API changes.
>
> OTHER CHANGES:
> No changes required.
>
> MIGRATION:
> No migration is needed.
>
> COMPATIBILITY
>
> BREAKING CHANGES:
> No breaking changes.
>
> EXISTING PROGRAMS:
> Existing programs are not affected by this change.
>
> REFERENCES
> My Java7 Wishlist regarding Collections,
>
http://shamsmi.blogspot.com/2008/04/my-java7-wishlist-regarding-collections.html
> Implementation of My Java7 Wishlist,
>
http://shamsmi.blogspot.com/2008/05/implementation-of-my-java7-wishlist.html
>
> EXISTING BUGS:
> None.
>
> URL FOR PROTOTYPE (optional):
> Projects kijaro's concisecollections branch:
>
https://kijaro.dev.java.net/source/browse/kijaro/branches/concisecollections/
>
>
>
>
>


-- 
Shams Mahmood



More information about the coin-dev mailing list