Java Performance and Garbage Collection

On this episode of The Programming Podcast, Danny Thompson and Leon Noel unpack the biggest programming myths that confuse developers at every level. From “AI will take all dev jobs” to “DRY at all costs,” they separate hype from reality and share hard-won lessons from real teams in production.You will hear why paper Big-O is not the whole story, how cache behavior and data size impact real performance, and why map/reduce vs for loops is a wash on modern engines. We get into testing culture too: why E2E does not replace unit tests, how to use AI for test scaffolding without losing your engineering brain, and what actually improves product reliability. Danny also tackles the myths that Java is slow and GC is always bad, and both hosts talk about the cost curve where cloud is not cheaper than on-prem.The conversation closes with an “Ask Danny and Leon” mailbag on what really separates junior, mid, and senior engineers: independence, guardrails, impact, and the quality of questions you ask.If this helped, drop a comment with a myth you want us to tackle next, and subscribe for more practical, no-fluff engineering talk.Topics include:AI as a productivity tool vs one-click magicBig-O vs real-world performance and memory behaviorjQuery, Deno, Bun, and the hype cycleJavaScript the language vs browser APIsmap/reduce vs for loops on modern enginesUnit tests, integration tests, E2E, and using AI wiselyJava performance and garbage collection tuningDRY vs duplication and over-abstractionAccessibility as a defaultCloud costs vs on-prem at scaleCareer ladder: junior, mid, senior traitsHosts: Danny Thompson (Director of Technology, This Dot Labs; Commit Your Code Conference) and Leon Noel (Managing Director of Engineering, Resilient Coders; instructor at 100Devs)Chapters00:00 Intro and why myths still persist00:58 Host intros and setup for “Gem City” episode02:00 Myth 1: “AI is taking all dev jobs”03:33 When AI image gen goes sideways and why it is a tool, not a replacement06:03 Leon’s motion-blur trick for more believable AI images07:08 Myth 2: Big-O vs real performance in the wild09:52 Cache misses, allocation, data size, and why paper math can mislead11:30 Myth 3: “jQuery is dead” and the reality in legacy estates12:23 Deno and Bun hype vs actual employer adoption13:32 Why jQuery still ships and what we lost chasing complexity15:30 JavaScript the language vs DOM and host environment APIs16:58 Myth 4: “map/reduce are slower than for loops” on modern engines18:11 Myth 5: “E2E replaces unit tests”20:57 When testing cultures go wrong and how to course-correct22:46 Using AI for tests without losing critical thinking25:03 The 80-20 way to use AI on tickets and test suites27:01 Danny gets baited, Leon laughs28:01 Myth 6: “Java is slow” and “GC is always bad”30:30 Region and concurrent collectors, and why allocation patterns matter31:02 Engine differences and mental models across stacks31:58 Myth 7: “Everything must be DRY” vs useful duplication33:01 Strong opinions held weekly and leaving dogma behind34:22 How dev opinions evolve with experience34:52 Accessibility as a default, not a later task36:29 Myth 8: “Cloud is always cheaper than on-prem”37:17 Real-world cost surprises and pulling workloads back38:30 Hype cycles, Jamstack memories, and maintenance pain41:56 On-prem done right and budget realities43:49 Mailbag: junior vs mid vs senior, company variance45:25 Danny’s framework for levels: guardrails, impact, and ownership51:10 The power of high-quality questions at senior and staff levels52:10 Leveling up from mid: own initiatives and become the firefighter53:31 Wrap-up and sign-off