Episode 517: Jordan Adler on Code Generators

Migration Process

Pinterest's transition from Python 2 to Python 3 was a meticulous process, leveraging tools like Python Future to facilitate the migration. Jordan Adler explains that Python Future, developed by Python Charmers, includes code transformers that modify Python 2 code to be more compatible with Python 3 1. This approach allowed Pinterest to gradually implement changes without disrupting existing functionalities. Adler notes, "You can migrate while using features that would typically not be available on Python two or specifically the API that changes under Python three" 2. By running Python 2 code with Python 3 behaviors, Pinterest ensured a smoother transition, ultimately leading to a cleaner, fully Python 3 codebase.

   

Migration Tools

The migration at Pinterest was heavily reliant on the Python Future tool, which utilizes the AST (Abstract Syntax Tree) class from the Python standard library. Jordan Adler describes how this tool reads and transforms code by identifying specific nodes, such as function calls, and modifying them as needed 3. This method is particularly effective for interpreted languages like Python, which can model code within the language itself. Adler highlights, "Most compiled languages don't have some version of this," emphasizing the unique capabilities of Python in facilitating such transformations 3. This approach allowed Pinterest to efficiently manage the complexities of their codebase during the migration.