#490 It’s a vibe coding party

Impossibile aggiungere al carrello

Puoi avere soltanto 50 titoli nel carrello per il checkout.

Riprova più tardi

Rimozione dalla Lista desideri non riuscita.

Riprova più tardi

Non è stato possibile aggiungere il titolo alla Libreria

Per favore riprova

Non è stato possibile seguire il Podcast

Per favore riprova

Esecuzione del comando Non seguire più non riuscita

#490 It’s a vibe coding party

Ascolta gratuitamente

Vedi i dettagli del titolo

Topics covered in this episode: Some more things about Django I've been enjoyingWho cleans up after the vibe-coding party?Where Did All Your AI Tokens Go? AgentsView to the rescue!Careful with phishing allExtrasJokeWatch on YouTube About the show Sponsored by us! Support our work through: Our courses at Talk PythonConsulting from Six Feet Up Connect with the hosts Michael: Mastodon / BlueSky / X / LinkedInCalvin: Mastodon / BlueSky / X / LinkedInShow: Mastodon / BlueSky / X Join us on YouTube at pythonbytes.fm/live to be part of the audience. Usually Tuesday at 7am PT. Older video versions available there too. Finally, if you want an artisanal, hand-crafted digest of every week of the show notes in email form? Add your name and email to our friends of the show list, we'll never share it. Calvin #1: Some more things about Django I've been enjoying Julia Evans is learning "2010-style" web dev (Django + SQL + server-rendered HTML) after years of Go backends and JS-heavy frontendsQuery builders: likes defining custom QuerySet classes with chainable filter methods (.approved().future().with_tags()) — more readable than raw SQLTemplate filters: highlights urlize, linebreaksbr, json_script, and especially querystring for building/modifying query-string links in templatesMigrations: still loves Django's auto-generated migrations — 19 and counting on her projectSkips inheritance for class-based views; prefers function-based views for sharing code, though fine using Django's own mixins/interfacesPerformance surprise: CPU profiling (via py-spy) — not slow DB queries — revealed the culprit; she'd accidentally disabled the cached template loader, and re-enabling it took throughput from ~2-3 req/s to ~12 req/s on a $10/mo VM Michael #2: Who cleans up after the vibe-coding party? FT Magazine piece by Sam Learner (July 11) on AI coding tools overwhelming open source maintainers - sent in by listener Dylan McConnell, whose main point was that this ran in the Financial Times, not a dev blog. cURL as the case study - Daniel Stenberg has been the only full-time person on it for years; libcurl has been installed an estimated 20+ billion times with 3,000+ listed contributors.Bug bounty killed - cURL ended its paid security bounty program in January, citing an "explosion of AI slop reports" that take real time to debunk and drain morale.Extractive contributions - authoring a PR is now nearly free, reviewing one still costs a human; tldraw's Steve Ruiz closed outside contributions entirely, asking why he'd want someone else writing the easy part.Guido weighs in - van Rossum says projects are holding emergency meetings over the slop flow, and notes LLM patches tend to touch unrelated parts of a file, making review more tedious."Vibe Coding Kills Open Source" - paper from Miklós Koren's group: packages frequently recommended by coding models saw big download jumps with no matching engagement, breaking the reputation loop that sustains maintainers.Stack Overflow flatlined - over 100,000 questions a month before ChatGPT, under 1,500 last month, with the response rate cut roughly in half; the public archive is now stale training data.The course-creator angle - Josh Comeau's newest web dev course launched at about a third of prior enrollment, and he worries about devs who never learn which questions to ask. But the most interesting portion is what was omitted. Focused on: The end of the curl bug-bountyOmitted: High-Quality Chaos Why the omission is interesting It fits a narrative. The FT piece is a maintenance-and-decline story, and January-Stenberg is a perfect witness for it. April-Stenberg complicates it - same person, same project, better data, opposite direction on the specific claim being used.The tell is already in the article. Learner quotes Stenberg saying AI tools are much better at finding problems than fixing them. That's the April thesis in one line, and it goes undeveloped.Reason for the shift is process, not vibes. Killing the bounty removed the cash incentive and the venue change filtered the rest. Worth saying out loud, because "AI reports got better" isn't quite it - "no bounty plus a real triage platform" is closer. Joke too: Sarah O’Connor wrote a related piece (is this just before skynet launches?) Calvin #3: Where Did All Your AI Tokens Go? AgentsView to the rescue! Local-first desktop/web app for browsing, searching, and analyzing your past AI coding agent sessions (Claude Code, Codex, Copilot, Cursor, Gemini, Aider, and dozens more)Auto-discovers session files on your machine — no config needed; everything stored locally in SQLite, no cloud/accountsagentsview usage is a drop-in ccusage alternative — reads from pre-indexed SQLite, reports run 80–220× faster on large historiesNew Activity dashboard shows peak concurrency, active vs. idle time, agent-minutes, and cost — filterable by project/agent/machine, with a -json CLI report tooFull-text + optional semantic search across every session; ...

Ancora nessuna recensione