The post responds to Jesper Cockx’s article on syntax with binding representation, adding a method not mentioned before: randomly generated, globally unique identifiers for bound variables. The writer explores various criteria for comparison, highlighting the issue with de Bruijn indices failing to encode variables consistently. They argue in favor of unique IDs over de Bruijn indices, citing ease of understanding, simplicity of implementation, and efficiency in practical programming. The writer presents a comparison table and concludes that unique IDs are preferable for practical programming tasks. This post sparks discussion on various platforms, including Lobsters, Mastodon, and Reddit. Jesper’s response is also linked.
https://safinaskar.writeas.com/