Building a Compiler (Step-by-Step)

Based on Pixeled's (youtube.com/@pixeled-yt) "Creating a Compiler" series.

Why choose C++ for this task? No idea, but enjoy the following quip from uncyclopedia.com's page on the C++ language.

"It gives programmers more control by allowing them to overload and override functions, while being backwards compatible with the C language's ability to overrun buffers. The language is also incredibly hard."

Anyway, here's the version notes:

• v1/a Corresponds to timestamp of 0:00-19:55 in part one of the video series. At this point, the directory contains the .asm file of a program to return an exit code, object file output by NASM (nasm), executable produced by GNU Linker (ld), and a Bash script to automatically check the program.

• v1/b Corresponds to timestamp of 19:56-27:51 in part one of the video series. This is an interim version used to check cmake and C++ are working on the system. Version 2 only contains a basic CMakeLists.txt file and a C++ "Hello, World!" program. There is an included Bash script to automate testing of the compiler at this point.

• v1/c Corresponds to timestamp of 27:52-38:21 in part one of the video series. The third version builds upon the CMake and C++ files from v2 and introduces the code needed to read the contents of the Hydrogen file (test.hy).A step is added to the Bash script which runs the hydro executable on test.hy to return the Hydrogen file contents as a string.

• v1/d Corresponds to timestamp of 38:22-57:56 in part one of the video series. File structure remains the same as v03 with tokenization added to the main.cpp file. With the exception of main.cpp all file contents remain unchanged.

• v1/e Corresponds to timestamp of 57:57-1:05:10 in part one of the video series. All conditions related to v04 are the same. File structure and contents with exception of main.cpp are unchanged. Introduces conversion of tokens to Assembly code, but lacks the functionality to output the Assembly code to a file of its own.

• v1/f Corresponds to timestamp of 1:05:11-1:11:02 in part one of the video series. At this step, pre-existing file structure is unchanged. Additional files are added, included out.asm, out.o, and the out executable. The test.sh script receives an update to continue providing a complete and automated test of the compiler at this point in development. Functionality to output Assembly code to a file of its own is added for this version.

• v2/g Corresponds to timestamp of 0:00-28:54 in part two of the video series. Refactoring takes place in this version. By now, the tokenization has been abstracted to a header file. Several processes related to tokenization have been improved. Corresponding code within main.cpp has been reduced. The test.sh script has been improved with absolute paths which allow it to be from anywhere on the local host. All other file contents and structure have remain the same.

• v2/h Corresponds to timestamp of 28:55-1:05:47 in part two of the video series. Parsing and code generation are abstracted to .hpp files. A grammar.md file is now included. Code in main.cpp has been modified or reduced to use the contents of header files accordingly. Other file contents and structure are unchanged.

• v3/i Corresponds to timestamp of 0:00-TBD in part three of the video series. At the time of writing, this version is a work-in-progress.