{"id":683,"date":"2017-11-05T13:54:42","date_gmt":"2017-11-05T13:54:42","guid":{"rendered":"http:\/\/www.rangakrish.com\/?p=683"},"modified":"2017-11-05T14:00:59","modified_gmt":"2017-11-05T14:00:59","slug":"c17-stdvariant","status":"publish","type":"post","link":"https:\/\/www.rangakrish.com\/index.php\/2017\/11\/05\/c17-stdvariant\/","title":{"rendered":"C++17: std::variant<>"},"content":{"rendered":"

C++17 introduces a new type-safe union in the form of std::variant<\/strong><\/em>. At any time, it can store a single value from one of many types. We need to include <variant><\/strong><\/em>\u00a0to use this feature.<\/p>\n

Let us look at a simple example to get started:<\/p>\n

\"Example1\"<\/a>
Example1<\/strong><\/figcaption><\/figure>\n

First, we define a variant object v1<\/strong><\/em> to hold either an int<\/strong><\/em> or string<\/strong><\/em> type, and initialize it with an integer value. The alternative elements can each be accessed by an index value, starting from 0. The left-most alternative gets zero, the next one gets 1, and so on. In this example, int<\/strong><\/em> will corrspond to index 0 and string<\/strong><\/em> will get index 1. The index()<\/strong><\/em> member function returns the index for the valid alternative.<\/p>\n

We use the std::get<>()<\/strong><\/em> function to reference the currently valid alternative. Since the current value is integer, we use get<int>(v1)<\/strong><\/em>. We could also use get<0>(v1)<\/strong><\/em>.<\/p>\n

We can assign a value directly to the variant object, or through the active index. Note that trying to access the invalid alternative using the wrong index will\u00a0 throw a run-time exception of type std::bad_variant_access<\/strong><\/em>. On the other hand, it is a compile-time error to use an index outside the valid range.<\/p>\n

The correct variant alternative will be selected automatically based on the initialization expression. If we initialize with a string literal, the string<\/strong><\/em> alternative is selected. And in this case, the valid active index will be 1, not zero.<\/p>\n

See the next example:<\/p>\n

\"Example2\"<\/a>
Example2<\/strong><\/figcaption><\/figure>\n

What if we do not initialize at all? In this case, the left-most alternative must allow default initialization, else it is a compile-time error. In the above example, the int<\/strong><\/em> element will be initialized with 0, and the index()<\/strong><\/em> function will return 0.<\/p>\n

Let us try using some user-defined type. Consider the struct X<\/strong><\/em> given below:<\/p>\n

\"User<\/a>
User Type<\/strong><\/figcaption><\/figure>\n

Note that X<\/strong><\/em> does not have a default constructor. Let us consider using X<\/strong><\/em> as a type alternative in std::variant<\/strong><\/em>.<\/p>\n

\"Example3\"<\/a>
Example3<\/strong><\/figcaption><\/figure>\n

The first definition (commented out) does not compile because we are depending on default initialization and X<\/strong><\/em>, being the left-most type, does not have a default constructor.<\/p>\n

When we change the order of types, it works because int<\/strong><\/em> will become the active type with a default value of 0.<\/p>\n

In the next definition, we supply a proper initialization value for X<\/strong><\/em>. This results in a call to X<\/strong><\/em> constructor to instantiate a new object and X<\/strong><\/em> becomes the active type.<\/p>\n

The next statement is interesting. We assign an integer value to the variable. Since the previous value was of type X<\/strong><\/em>, the compiler cannot just overwrite that area with the integer value, but it must destroy X<\/strong><\/em> first! This is indeed what happens, and this is the most important difference between a union and a variant. std::variant<\/strong> <\/em>thus guarantees type safety!<\/p>\n

We saw earlier that if we pass an incorrect index (or type) to std::get<\/strong><\/em>, we get std::bad_variant_access<\/strong><\/em> exception. Alternatively, we can\u00a0 use std::get_if<\/strong><\/em>. In this case we have to pass a pointter to the variant object. If the index (or type) matches the active element, then a pointer to the value will be returned. If not, a nullptr<\/strong><\/em> will be returned. This is illustrated in the following example:<\/p>\n

\"Example4\"<\/a>
Example4<\/strong><\/figcaption><\/figure>\n

Alternatively, we can use the function std::holds_alternative()<\/strong><\/em> to check if the variant currently holds a value of given type. See the following example:<\/p>\n

\"Example5\"<\/a>
Example5<\/strong><\/figcaption><\/figure>\n

The next example shows how we can use variant::emplace()<\/strong><\/em> member function to construct an object in-place.<\/p>\n

\"Example6\"<\/a>
Example6<\/strong><\/figcaption><\/figure>\n

The library guarantees that the previous object will be destroyed before constructing the new object in its place.<\/p>\n

The last funcationality we will discuss is std::visit<\/strong><\/em>. Let us suppose that we have a collection of variant objects, and we want to visit each element of the collection and do something on it. How can we do this? There are a few interesting ways, but to keep the discussion brief, we will look at a basic approach.<\/p>\n

Let us first define a function object that has overloaded versions for the different variant alternatives (we use member template for this). We can pass the function object to the std::visit<\/strong><\/em> function along with the variant. This causes the appropriate function (based on the active variant alternative) to be applied on the variant element. This is shown in the following figure.<\/p>\n

\"Example7\"<\/a>
Example7<\/strong><\/figcaption><\/figure>\n

Thus std::variant<\/strong><\/em> is a useful concept when you are looking for a type-safe discriminated union. Check it out.<\/p>\n

I used Visual Studio Profession 2017, Version 15.4.0 for this article. The source code can be downloaded from\u00a0here<\/a>.<\/p>\n","protected":false},"excerpt":{"rendered":"

C++17 introduces a new type-safe union in the form of std::variant. At any time, it can store a single value from one of many types. We need to include <variant>\u00a0to use this feature. Let us look at a simple example to get started: First, we define a variant object v1 to hold either an int […]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"advanced_seo_description":"","jetpack_seo_html_title":"","jetpack_seo_noindex":false,"_jetpack_newsletter_access":"","_jetpack_dont_email_post_to_subs":false,"_jetpack_newsletter_tier_id":0,"_jetpack_memberships_contains_paywalled_content":false,"_jetpack_memberships_contains_paid_content":false,"footnotes":"","jetpack_publicize_message":"","jetpack_publicize_feature_enabled":true,"jetpack_social_post_already_shared":false,"jetpack_social_options":{"image_generator_settings":{"template":"highway","default_image_id":0,"font":"","enabled":false},"version":2},"jetpack_post_was_ever_published":false},"categories":[49,17],"tags":[118,121],"class_list":["post-683","post","type-post","status-publish","format-standard","hentry","category-c","category-programming","tag-c17","tag-stdvariant"],"jetpack_publicize_connections":[],"jetpack_featured_media_url":"","jetpack_sharing_enabled":true,"jetpack_shortlink":"https:\/\/wp.me\/p9OLnF-b1","jetpack-related-posts":[{"id":717,"url":"https:\/\/www.rangakrish.com\/index.php\/2017\/12\/03\/c17-stdany\/","url_meta":{"origin":683,"position":0},"title":"C++17: std::any","author":"admin","date":"December 3, 2017","format":false,"excerpt":"In the previous two posts, I talked about std::variant<> and std::optional<>. Today, I want to take up std::any for discussion. The type any (implemented by the class any) allows a variable to hold a single value of any type. More interestingly, the type of the value held by a variable\u2026","rel":"","context":"In "C++"","block_context":{"text":"C++","link":"https:\/\/www.rangakrish.com\/index.php\/category\/c\/"},"img":{"alt_text":"Example1","src":"https:\/\/i0.wp.com\/www.rangakrish.com\/wp-content\/uploads\/2017\/12\/Example1.png?resize=350%2C200","width":350,"height":200,"srcset":"https:\/\/i0.wp.com\/www.rangakrish.com\/wp-content\/uploads\/2017\/12\/Example1.png?resize=350%2C200 1x, https:\/\/i0.wp.com\/www.rangakrish.com\/wp-content\/uploads\/2017\/12\/Example1.png?resize=525%2C300 1.5x"},"classes":[]},{"id":700,"url":"https:\/\/www.rangakrish.com\/index.php\/2017\/11\/21\/c17-stdoptional\/","url_meta":{"origin":683,"position":1},"title":"C++17: std::optional","author":"admin","date":"November 21, 2017","format":false,"excerpt":"Suppose we want to write a function that returns a value, but with the possibility that\u00a0 the computation might fail. This failure can be represented as an exception, or as a return value that unamibiguously denotes failure (for example, -1). Throwing an exception is a strong form of failure and\u2026","rel":"","context":"In "C++"","block_context":{"text":"C++","link":"https:\/\/www.rangakrish.com\/index.php\/category\/c\/"},"img":{"alt_text":"Example1","src":"https:\/\/i0.wp.com\/www.rangakrish.com\/wp-content\/uploads\/2017\/11\/Example1-3.png?resize=350%2C200","width":350,"height":200,"srcset":"https:\/\/i0.wp.com\/www.rangakrish.com\/wp-content\/uploads\/2017\/11\/Example1-3.png?resize=350%2C200 1x, https:\/\/i0.wp.com\/www.rangakrish.com\/wp-content\/uploads\/2017\/11\/Example1-3.png?resize=525%2C300 1.5x"},"classes":[]},{"id":1928,"url":"https:\/\/www.rangakrish.com\/index.php\/2020\/03\/15\/stdis_destructible\/","url_meta":{"origin":683,"position":2},"title":"std::is_destructible","author":"admin","date":"March 15, 2020","format":false,"excerpt":"In the last article, I explained the deleted destructor in some detail. Today, I would like to talk about a related construct, a type trait called std::is_destructible. Type traits, defined in the header , are a big help when it comes to implementing template metaprogramming. \u00a0See this\u00a0article for a nice\u2026","rel":"","context":"In "C++"","block_context":{"text":"C++","link":"https:\/\/www.rangakrish.com\/index.php\/category\/c\/"},"img":{"alt_text":"Private\/Protected Destructor","src":"https:\/\/i0.wp.com\/www.rangakrish.com\/wp-content\/uploads\/2020\/03\/Example7.jpg?resize=350%2C200&ssl=1","width":350,"height":200,"srcset":"https:\/\/i0.wp.com\/www.rangakrish.com\/wp-content\/uploads\/2020\/03\/Example7.jpg?resize=350%2C200&ssl=1 1x, https:\/\/i0.wp.com\/www.rangakrish.com\/wp-content\/uploads\/2020\/03\/Example7.jpg?resize=525%2C300&ssl=1 1.5x"},"classes":[]},{"id":1973,"url":"https:\/\/www.rangakrish.com\/index.php\/2020\/04\/26\/stdis_base_of\/","url_meta":{"origin":683,"position":3},"title":"std::is_base_of","author":"admin","date":"April 26, 2020","format":false,"excerpt":"In our on-going series on C++ Type Traits, today's topic is about the trait std::is_base_of<>. For the official description, see this. std::is_base_of::value takes two arguments, both classes (or structs), and returns true if A is a base class of B and false otherwise. Trivially, std::::value is true. Let\u2026","rel":"","context":"In "C++"","block_context":{"text":"C++","link":"https:\/\/www.rangakrish.com\/index.php\/category\/c\/"},"img":{"alt_text":"Output","src":"https:\/\/i0.wp.com\/www.rangakrish.com\/wp-content\/uploads\/2020\/04\/Output.jpg?resize=350%2C200&ssl=1","width":350,"height":200},"classes":[]},{"id":1960,"url":"https:\/\/www.rangakrish.com\/index.php\/2020\/04\/12\/stdcommon_type-type-trait\/","url_meta":{"origin":683,"position":4},"title":"std::common_type Type Trait","author":"admin","date":"April 12, 2020","format":false,"excerpt":"In today's post, I would like to go over the type trait std::common_type<>. This trait was introduced in C++11. As per the specification, std::common_type::type refers to a type Tx in the given list, which the rest of the types in the list can be implicitly converted to. This\u2026","rel":"","context":"In "C++"","block_context":{"text":"C++","link":"https:\/\/www.rangakrish.com\/index.php\/category\/c\/"},"img":{"alt_text":"Program Output","src":"https:\/\/i0.wp.com\/www.rangakrish.com\/wp-content\/uploads\/2020\/04\/Fig5.jpg?resize=350%2C200&ssl=1","width":350,"height":200},"classes":[]},{"id":198,"url":"https:\/\/www.rangakrish.com\/index.php\/2016\/04\/08\/c14-return-type-deduction-for-normal-functions\/","url_meta":{"origin":683,"position":5},"title":"C++14: Return Type Deduction for Normal Functions","author":"admin","date":"April 8, 2016","format":false,"excerpt":"In C++11, a function must specify its return type in one of the following two ways: int aFunction(int arg) { \u00a0 \u00a0 return arg * 2; } (or) auto aFunction(int arg) -> int { \u00a0 \u00a0 return arg * 2; } The first is the classic function definition syntax. The\u2026","rel":"","context":"In "C++"","block_context":{"text":"C++","link":"https:\/\/www.rangakrish.com\/index.php\/category\/c\/"},"img":{"alt_text":"","src":"","width":0,"height":0},"classes":[]}],"_links":{"self":[{"href":"https:\/\/www.rangakrish.com\/index.php\/wp-json\/wp\/v2\/posts\/683","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.rangakrish.com\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.rangakrish.com\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.rangakrish.com\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.rangakrish.com\/index.php\/wp-json\/wp\/v2\/comments?post=683"}],"version-history":[{"count":0,"href":"https:\/\/www.rangakrish.com\/index.php\/wp-json\/wp\/v2\/posts\/683\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.rangakrish.com\/index.php\/wp-json\/wp\/v2\/media?parent=683"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.rangakrish.com\/index.php\/wp-json\/wp\/v2\/categories?post=683"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.rangakrish.com\/index.php\/wp-json\/wp\/v2\/tags?post=683"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}