Ben & Ryan Show Episode 24

In this episode, your hosts Ben Nadel and Ryan Brown go guest-free and take the opportunity to dive into ColdFusion bugs, guinea pig maintenance, and their latest experiments in app building and AI usage.They kick things off with some light banter, guinea pig parenting tips, and custom shoes, before transitioning into deep debugging talk, asynchronous programming quirks in ColdFusion 2025, and the evolving relationship developers have with AI tooling.Key Points• Ryan surprises Ben with custom ColdFusion-themed shoes for the upcoming CF Summit.• Ben shares two deep ColdFusion 2025 bugs related to asynchronous iteration and CFQuery behavior.• Big Sexy Poems—a side project by Ben—is being built using HTMX and AlpineJS with a ColdFusion backend.• Debugging ColdFusion edge cases requires intense reduction and isolation of code to find root causes.• Ryan and Ben discuss AI voice mode frustrations and using AI for sales and dev support—highlighting both promise and pitfalls.Ben introduces his new ColdFusion 2025 app, “Big Sexy Poems,” and why he built it.• The app helps poets count syllables, find rhymes, and synonyms using the Datamuse API.• Ben is rebuilding it to learn HTMX and AlpineJS in a real-world context.• ColdFusion 2025 forces him to rethink and modernize old code written for CF2021.• Features include caching, shareable links, and saving user poems.Ben outlines two complex ColdFusion bugs that derailed development.• Bug #1: CFQuery inside asynchronous iteration dumps SQL statements to the page.• Bug #2: A three-level deep array iteration causes closure variables to vanish.• Both bugs are confirmed to go back as far as ColdFusion 2021.• Fixes required rewriting code to remove async parallelism and use manual loops.They explore how to isolate and debug ColdFusion bugs in large codebases.• Ben shares his method: reduce files down to minimal reproducible cases.• Progressively merge and remove components until only the bug trigger remains.• Use CF’s ability to simulate services to isolate logic layers during debugging.• Debugging complex async issues often takes hours but can result in 20-line reproductions.Conversation pivots to AI tools, frustrations, and how Ryan uses AI for sales support.• Ben tried ChatGPT voice mode again and found it frustratingly sycophantic.• Ryan uses persona-driven GPTs to simulate sales coaches and customer role-play.• Operator and GPT Agents have shown promise in configuring servers from specs.• Even advanced tools sometimes fail mid-task or require lots of oversight.• The team plans a live Ben and Ryan Show at CF Summit to talk AI and dev tools in person.