Skip to main content

What Is RAII

RAII is a frequently used idiom in C++ that ensures the safe usage of resources by releasing them when an object's scope ends. In C++, resources allocated on the heap are not released unless explicitly done so, but those allocated on the stack are automatically released when their scope ends, triggering their destructor. Originally, RAII was used to guard against unexpected changes in control flow, such as exceptions.

In the above code example, the unsafeFunction() function is not safe. If the thisFunctionCanThrowException() throws an exception, the resource may not be released. The unmaintanableFunction releases the resource, but it is not easy to read and maintain.

The safeFunction example uses unique_ptr, a smart pointer introduced at C++11, for RAII. unique_ptr automatically releases the memory it holds when it is destroyed, ensuring that the resource is released when the function exits.

The resource does not only refer to heap memory but also includes files, databases, and other things that can be safely used with RAII. Furthermore, RAII can be used to handle code that must always be executed when a specific scope ends, similar to the finally statement in other languages. In fact, Bjarne Stroustrup, the creator of C++ and the term RAII, stated that there was no need for a finally statement in C++ due to the existence of RAII.

This article is a translation of a Korean post written in 2014. If you would like to view the original, please refer to this link.

Labels:

Comments

Post a Comment

Popular posts from this blog

Understanding Aspect-Oriented Programming with Python Examples

Object-Oriented Programming (OOP) manages code by grouping it into independent modules known as objects, emphasizing the crucial principle of Separation of Concerns. This means each object should focus on its specific responsibilities. However, real-world applications often feature functionalities that are common across multiple objects or modules, such as logging, security, transaction management, and performance monitoring. These functionalities are called Cross-cutting Concerns. Challenge of Cross-Cutting Concerns Scattering These cross-cutting concerns, when handled solely with OOP, create two major problems. The first problem is Scattering , which is when code for a specific functionality is spread across multiple places through copying and pasting. For instance, imagine adding user permission checks and logging code to every function. The same logging and permission checking code would repeatedly appear within each method. Tangling The other issue is Tangling . This refe...
Post a Comment
Read more

Cursor Movement with CSI Sequences

Code Abbr Name CSI # A CUU CUrsor Up CSI # B CUD CUrsor Down CSI # C CUF CUrsor Forward CSI # D CUB CUrsor Backward CSI # E CNL CUrsor Next Line CSI # F CPL CUrsor Previous Line CSI # I CHT Cursor Horizontal forward Tabulation CSI # Z CBT Cursor Backward Tabulation CSI # G CHA Cursor Horizontal Absolute CSI # ; # H CUP CUrsor Position Today, we will continue from the  previous article to explore how to move the cursor using CSI sequences. The types of CSI sequences for moving the cursor can be summarized as follows. CUU, CUD, CUF, CUB These are the abbreviations for CUrsor Up, CUrsor Down, CUrsor Forward, and CUrsor Backward; as the names suggest, they move the cursor up, down, forward, and backward. They take a single number as an argument; if the argument is omitted, it is treated as 1. Thus, 0x1b[A is equivalent to 0x1b[1A . In this case, CUF and CUB move only within the same line. In other words, CUB rece...
Post a Comment
Read more

[C++] enum class

Traditional C++ enum had several issues. To solve these problems, C++11 introduced a new feature called enum class . In this article, I will examine the problems with the traditional enum and how they are solved with enum class . First, traditional enum could not be forward-declared. The reason was that if the values in the enumerator were unknown, it was impossible to determine their size . However, enum class is treated as int if an underlying type is not specified, assigning values outside the range of an int will raise a compilation error. If you want to use values outside the range of an int , you need to specify the underlying type. Another problem with traditional enum was that the scope of enumerator names was not limited. Let's see the following example. Here, we try to represent the results of IO and Parse functions with enum s. However, this code will not compile because the Error and Ok of IOResult conflict with those of ParseResult . To resolve t...
Post a Comment
Read more