<div dir="ltr">+1<br></div><br><div class="gmail_quote"><div dir="ltr">On Wed, 12 Oct 2016 at 09:09 Said Sikira via swift-evolution <<a href="mailto:swift-evolution@swift.org">swift-evolution@swift.org</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div style="word-wrap:break-word" class="gmail_msg"><div id="m_-8426030209915870348bloop_customfont" style="font-family:Helvetica,Arial;font-size:13px;color:rgba(0,0,0,1.0);margin:0px;line-height:auto" class="gmail_msg">+1</div></div><div style="word-wrap:break-word" class="gmail_msg"> <br class="gmail_msg"> <div id="m_-8426030209915870348bloop_sign_1476256035667259904" class="m_-8426030209915870348bloop_sign gmail_msg"></div> <br class="gmail_msg"><p class="m_-8426030209915870348airmail_on gmail_msg">On October 12, 2016 at 8:54:45 AM, Rien via swift-evolution (<a href="mailto:swift-evolution@swift.org" class="gmail_msg" target="_blank">swift-evolution@swift.org</a>) wrote:</p> <blockquote type="cite" class="m_-8426030209915870348clean_bq gmail_msg"><span class="gmail_msg"><div class="gmail_msg"><div class="gmail_msg"></div><div class="gmail_msg">Beautiful, +1<br class="gmail_msg"><br class="gmail_msg">Rien<br class="gmail_msg"><br class="gmail_msg">> On 11 Oct 2016, at 23:28, Nate Cook via swift-evolution <<a href="mailto:swift-evolution@swift.org" class="gmail_msg" target="_blank">swift-evolution@swift.org</a>> wrote:<br class="gmail_msg">> <br class="gmail_msg">> Introduction<br class="gmail_msg">> <br class="gmail_msg">> This proposal addresses significant unexpected performance gaps when using dictionaries. It introduces type-specific collections for a Dictionary instance's keys and values properties.<br class="gmail_msg">> <br class="gmail_msg">> New DictionaryKeys and DictionaryValues collections provide efficient key lookup and mutable access to dictionary values, enabling updates to be performed in-place and allowing copy-on-write optimization of stored values.<br class="gmail_msg">> <br class="gmail_msg">> Motivation<br class="gmail_msg">> <br class="gmail_msg">> This proposal address two problems:<br class="gmail_msg">> <br class="gmail_msg">> • The Dictionary type keys implementation is inefficient, because LazyMapCollection doesn't know how to forward lookups to the underlying dictionary storage.<br class="gmail_msg">> • Dictionaries do not offer value-mutating APIs. The mutating key-based subscript wraps values in an Optional. This prevents types with copy-on-write optimizations from recognizing they are singly referenced.<br class="gmail_msg">> This proposal uses the following [String: [Int]] dictionary to demonstrate these problems:<br class="gmail_msg">> <br class="gmail_msg">> var dict = ["one": [1], "two": [2, 2], "three": [3, 3, 3]]<br class="gmail_msg">> Inefficient dict.keys Search<br class="gmail_msg">> <br class="gmail_msg">> Swift coders normally test key membership using nil checks or underscored optional bindings:<br class="gmail_msg">> <br class="gmail_msg">> if dict["one"] != nil<br class="gmail_msg">> {<br class="gmail_msg">> <br class="gmail_msg">> // ...<br class="gmail_msg">> <br class="gmail_msg">> }<br class="gmail_msg">> <br class="gmail_msg">> if let _ = dict["one"<br class="gmail_msg">> ] {<br class="gmail_msg">> <br class="gmail_msg">> // ...<br class="gmail_msg">> <br class="gmail_msg">> }<br class="gmail_msg">> <br class="gmail_msg">> These approaches provide the expected performance of a dictionary lookup but they read neither well nor "Swifty". Checking keys reads much better but introduces a serious performance penalty: this approach requires a linear search through a dictionary's keys to find a match.<br class="gmail_msg">> <br class="gmail_msg">> if dict.keys.contains("one") {<br class="gmail_msg">> // ...<br class="gmail_msg">> }<br class="gmail_msg">> <br class="gmail_msg">> A similar dynamic plays out when comparing dict.index(forKey:) and dict.keys.index(of:).<br class="gmail_msg">> <br class="gmail_msg">> Inefficient Value Mutation<br class="gmail_msg">> <br class="gmail_msg">> Dictionary values can be modified through the keyed subscript by direct reassignment or by using optional chaining. Both of these statements append 1 to the array stored by the key "one":<br class="gmail_msg">> <br class="gmail_msg">> // Direct re-assignment<br class="gmail_msg">> <br class="gmail_msg">> dict[<br class="gmail_msg">> "one"] = (dict["one"] ?? []) + [1<br class="gmail_msg">> ]<br class="gmail_msg">> <br class="gmail_msg">> <br class="gmail_msg">> // Optional chaining<br class="gmail_msg">> <br class="gmail_msg">> dict[<br class="gmail_msg">> "one"]?.append(1)<br class="gmail_msg">> Both approaches present problems. The first is complex and hard to read. The second ignores the case where "one" is not a key in the dictionary. It forces its check into a higher branch and encourages forced unwrapping. Furthermore, neither approach allows the array to grow in place. They introduce an unnecessary copy of the array's contents even though dict is the sole holder of its storage.<br class="gmail_msg">> <br class="gmail_msg">> Adding mutation to a dictionary's index-based subscripting isn't possible. Changing a key stored at a particular index would almost certainly modify its hash value, rendering the index incorrect. This violates the requirements of the MutableCollection protocol.<br class="gmail_msg">> <br class="gmail_msg">> Proposed Solution<br class="gmail_msg">> <br class="gmail_msg">> This proposal adds a custom collection for the keys and values dictionary properties. This follows the example set by String, which presents multiple views of its contents. A new DictionaryKeys collection introduces efficient key lookup, while a new DictionaryValues collection provides a mutable collection interface to dictionary values.<br class="gmail_msg">> <br class="gmail_msg">> These changes introduce a simple and efficient way of checking whether a dictionary includes a key:<br class="gmail_msg">> <br class="gmail_msg">> // Performant<br class="gmail_msg">> if dict.keys.contains("one"<br class="gmail_msg">> ) {<br class="gmail_msg">> <br class="gmail_msg">> // ...<br class="gmail_msg">> <br class="gmail_msg">> }<br class="gmail_msg">> <br class="gmail_msg">> As a mutable collection, values enables modification without copies or clumsy code:<br class="gmail_msg">> <br class="gmail_msg">> if let i = dict.index(forKey: "one"<br class="gmail_msg">> ) {<br class="gmail_msg">> dict<br class="gmail_msg">> .values[i].append(1) // no copy here<br class="gmail_msg">> <br class="gmail_msg">> } <br class="gmail_msg">> else<br class="gmail_msg">> {<br class="gmail_msg">> dict[<br class="gmail_msg">> "one"] = [1<br class="gmail_msg">> ]<br class="gmail_msg">> }<br class="gmail_msg">> <br class="gmail_msg">> Both the keys and values collections share the same index type as Dictionary. This allows the above sample to be rewritten as:<br class="gmail_msg">> <br class="gmail_msg">> // Using `dict.keys.index(of:)`<br class="gmail_msg">> if let i = dict.keys.index(of: "one"<br class="gmail_msg">> ) {<br class="gmail_msg">> dict<br class="gmail_msg">> .values[i].append(1<br class="gmail_msg">> )<br class="gmail_msg">> } <br class="gmail_msg">> else<br class="gmail_msg">> {<br class="gmail_msg">> dict[<br class="gmail_msg">> "one"] = [1<br class="gmail_msg">> ]<br class="gmail_msg">> }<br class="gmail_msg">> <br class="gmail_msg">> Detailed design<br class="gmail_msg">> <br class="gmail_msg">> • The standard library introduces two new collection types: DictionaryKeys and DictionaryValues.<br class="gmail_msg">> • A Dictionary's keys and values property types change from LazyMapCollection to these new types. <br class="gmail_msg">> • The new collection types are not directly constructable. They are presented only as views into a dictionary.<br class="gmail_msg">> struct Dictionary<Key: Hashable, Value>: ...<br class="gmail_msg">> {<br class="gmail_msg">> <br class="gmail_msg">> var keys: DictionaryKeys<Key, Value> { get<br class="gmail_msg">> }<br class="gmail_msg">> <br class="gmail_msg">> var values: DictionaryValues<Key, Value> { get set<br class="gmail_msg">> }<br class="gmail_msg">> <br class="gmail_msg">> <br class="gmail_msg">> // Remaining declarations<br class="gmail_msg">> <br class="gmail_msg">> }<br class="gmail_msg">> <br class="gmail_msg">> <br class="gmail_msg">> /// A collection view of a dictionary's keys.<br class="gmail_msg">> struct DictionaryKeys<Key: Hashable, Value><br class="gmail_msg">> : Collection {<br class="gmail_msg">> <br class="gmail_msg">> typealias Index = DictionaryIndex<Key, Value><br class="gmail_msg">> <br class="gmail_msg">> <br class="gmail_msg">> subscript(i: Index) -> Key { get<br class="gmail_msg">> }<br class="gmail_msg">> <br class="gmail_msg">> <br class="gmail_msg">> // Other `Collection` requirements<br class="gmail_msg">> <br class="gmail_msg">> }<br class="gmail_msg">> <br class="gmail_msg">> <br class="gmail_msg">> /// A mutable collection view of a dictionary's values.<br class="gmail_msg">> struct DictionaryValues<Key: Hashable, Value><br class="gmail_msg">> : MutableCollection {<br class="gmail_msg">> <br class="gmail_msg">> typealias Index = DictionaryIndex<Key, Value><br class="gmail_msg">> <br class="gmail_msg">> <br class="gmail_msg">> subscript(i: Index) -> Value { get set<br class="gmail_msg">> }<br class="gmail_msg">> <br class="gmail_msg">> <br class="gmail_msg">> // Other `Collection` requirements<br class="gmail_msg">> <br class="gmail_msg">> }<br class="gmail_msg">> <br class="gmail_msg">> A sample implementation of this proposal can be found in this branch.<br class="gmail_msg">> <br class="gmail_msg">> Impact on existing code<br class="gmail_msg">> <br class="gmail_msg">> The performance improvements of using the new DictionaryKeys type and the mutability of the DictionaryValuescollection are both additive in nature.<br class="gmail_msg">> <br class="gmail_msg">> Most uses of these properties are transitory in nature. Adopting this proposal should not produce a major impact on existing code. The only impact on existing code exists where a program explicitly specifies the type of a dictionary's keysor values property. The fix is to change the specified type. <br class="gmail_msg">> <br class="gmail_msg">> Alternatives considered<br class="gmail_msg">> <br class="gmail_msg">> The Generics Manifesto lists nested generics as a goal. This could impact the naming and structure of these new collection types. <br class="gmail_msg">> <br class="gmail_msg">> Instead of DictionaryKeys<Key, Value> and DictionaryValues<Key, Value>, these types could be Dictionary<Key, Value>.Keys and Dictionary<Key, Value>.Values. However, because many types in the standard library may be revisited once such a feature is available (indices, iterators, etc.), the current lack of nesting shouldn't prevent consideration of this proposal.<br class="gmail_msg">> <br class="gmail_msg">> It could be possible to add additional compiler features that manage mutation through existing key-based subscripting without the copy-on-write problems of the current implementation. I don't know enough about how that would be implemented to speak to its feasibility or level of effort. Such a feature would reduce the need for a mutable DictionaryValues collection.<br class="gmail_msg">> _______________________________________________<br class="gmail_msg">> swift-evolution mailing list<br class="gmail_msg">> <a href="mailto:swift-evolution@swift.org" class="gmail_msg" target="_blank">swift-evolution@swift.org</a><br class="gmail_msg">> <a href="https://lists.swift.org/mailman/listinfo/swift-evolution" class="gmail_msg" target="_blank">https://lists.swift.org/mailman/listinfo/swift-evolution</a><br class="gmail_msg"><br class="gmail_msg">_______________________________________________<br class="gmail_msg">swift-evolution mailing list<br class="gmail_msg"><a href="mailto:swift-evolution@swift.org" class="gmail_msg" target="_blank">swift-evolution@swift.org</a><br class="gmail_msg"><a href="https://lists.swift.org/mailman/listinfo/swift-evolution" class="gmail_msg" target="_blank">https://lists.swift.org/mailman/listinfo/swift-evolution</a><br class="gmail_msg"></div></div></span></blockquote></div>
_______________________________________________<br class="gmail_msg">
swift-evolution mailing list<br class="gmail_msg">
<a href="mailto:swift-evolution@swift.org" class="gmail_msg" target="_blank">swift-evolution@swift.org</a><br class="gmail_msg">
<a href="https://lists.swift.org/mailman/listinfo/swift-evolution" rel="noreferrer" class="gmail_msg" target="_blank">https://lists.swift.org/mailman/listinfo/swift-evolution</a><br class="gmail_msg">
</blockquote></div>