#code-review — Public Fediverse posts
Live and recent posts from across the Fediverse tagged #code-review, aggregated by home.social.
-
One of the annoying things when having to review vibe-coded changes, is what I would call "method of the week": A quirk of the currently used LLM model found in a majority of PRs/MRs that vibe-coders don't learn to change before requesting a review despite being told multiple times.
#LLMs #vibeCoding #codeReview #peerReview #noAI #AI #artificialIntelligence #genAI #generativeAI #amCoding #amProgramming #softwareDevelopment #softwareEngineering
-
One of the annoying things when having to review vibe-coded changes, is what I would call "method of the week": A quirk of the currently used LLM model found in a majority of PRs/MRs that vibe-coders don't learn to change before requesting a review despite being told multiple times.
#LLMs #vibeCoding #codeReview #peerReview #noAI #AI #artificialIntelligence #genAI #generativeAI #amCoding #amProgramming #softwareDevelopment #softwareEngineering
-
AI-generated code looks just as clean as human code in review. That is exactly the problem: review was built to catch reasoning, not syntax. https://hackernoon.com/your-ai-agents-pull-request-looks-clean-thats-the-problem #codereview
-
AI-generated code looks just as clean as human code in review. That is exactly the problem: review was built to catch reasoning, not syntax. https://hackernoon.com/your-ai-agents-pull-request-looks-clean-thats-the-problem #codereview
-
"how many tons of CO₂ added to the atmosphere or liters of fresh water supplies disrupted is a tolerable price for a better code review?"
-
"how many tons of CO₂ added to the atmosphere or liters of fresh water supplies disrupted is a tolerable price for a better code review?"
-
JenkinsでGitHub Copilot CLIにPRを自動レビューさせる ― E2BIG・プロンプトインジェクション・トークン地獄との戦い
https://qiita.com/jqit-yukiono/items/21c54529410ba960c388?utm_campaign=popular_items&utm_medium=feed&utm_source=popular_items -
Drei Fragen und Antworten: Wie KI wirklich bei der Schwachstellensuche hilft | heise online
https://heise.de/-11377461 #KI #Codereview #Coding #Programmieren -
Drei Fragen und Antworten: Wie KI wirklich bei der Schwachstellensuche hilft | heise online
https://heise.de/-11377461 #KI #Codereview #Coding #Programmieren -
#fragefur1freund Es ist ja bekannt das bei vielen #Developer #Engineeering spezialisten im Bereich der #Softwareentwicklung die Nutzung von #KI gar nicht so schwarz gesehen wird wie von vielen Influenzern.
Wenn Ihr nun also eine KI euren #CodeReview machen lassen wollen würdet, wie #claude
welchen Skill empfehlt Ihr dafür?
-
Как я автоматизировал превращение вайбкодерского PoC в production-ready MVP
Как я автоматизировал превращение вайбкодерского PoC в production-ready MVP За несколько часов с помощью AI можно собрать работающий PoC: интерфейс открывается, кнопки нажимаются, основной сценарий проходит. Потом кто-нибудь спрашивает: А это уже можно выкатывать в прод? И тут начинается самое интересное. В проекте обнаруживаются дублирование логики, слабая типизация, почти отсутствующие тесты, случайная архитектура, забытые debug-логи и один файл на несколько тысяч строк, который страшно открывать и ещё страшнее менять. Мне регулярно приходится заниматься именно второй половиной этой работы — превращать быстро собранные прототипы в поддерживаемые MVP. В какой-то момент я понял, что каждый раз повторяю примерно один и тот же инженерный процесс. Так появился Pre2Prod — CLI на базе Codex, который последовательно проверяет репозиторий, составляет план исправлений, выполняет его и независимо перепроверяет результат. Под капотом — один постоянный Reviewer, временные Workers, 41 специализированное ревью и простой цикл: Review → Plan → Implement → Re-review
https://habr.com/ru/articles/1061702/
#AI #agents #openai #vibecoding #codereview #cli #npm #npx #poc #mvp
-
Ep. 18: Ha ancora senso leggere il codice generato dall'IA? (Rispondiamo ad Antirez)
https://www.spreaker.com/episode/ep-18-ha-ancora-senso-leggere-il-codice-generato-dall-ia-rispondiamo-ad-antirez--73011539
#podcast #goodvibrations #runtimeradio #pensieriincodice #aigenerativa #antirez #codereview #intelligenzaartificiale #llm #programmazione #qualitadelcodice #refactoring #runtimeradio #salvatoresanfilippo #sviluppatori #svilupposoftware #vibecoding #waltervannini -
Ep. 18: Ha ancora senso leggere il codice generato dall'IA? (Rispondiamo ad Antirez)
https://www.spreaker.com/episode/ep-18-ha-ancora-senso-leggere-il-codice-generato-dall-ia-rispondiamo-ad-antirez--73011539
#podcast #goodvibrations #runtimeradio #pensieriincodice #aigenerativa #antirez #codereview #intelligenzaartificiale #llm #programmazione #qualitadelcodice #refactoring #runtimeradio #salvatoresanfilippo #sviluppatori #svilupposoftware #vibecoding #waltervannini -
CW: Loooooooooooong read on the tech identity crisis, AI, and developer careers
Late-night thoughts... sparked by an engineer on Mastodon asking how, with all the "vibe coders" out there, the demand for actual, competent code reviews is ever going to be met: https://mastodon.world/@paulschoe/116894894916264050
So I just started writing...
The end of the tech exception
For almost thirty years, IT had this sort of "immunity." It was (yeah, was...) the ultimate safe-haven sector. Almost non-stop growth, incredibly comfortable (sometimes even indecent) conditions, and companies always begging for more people than were actually available. Cutthroat, sure, but the kind where you could just hop from gig to gig in Silicon Valley to always land a better deal.
But we have to look at two things. First, historical instability. The way we absorbed past major crises (oil shocks, the dot-com crash, subprime)... those were mostly economic or geopolitical. They hit the global economy, and tech felt the shockwaves, but never as hard as other sectors like real estate or the auto industry.
The second point is the current instability. Today, the crisis is homegrown. It’s coming from inside the tech industry itself (whether that’s Fintech, space tech, or even gov tech). Generative AI and agentic tools are changing the very core of how value is produced. (And the money and physical resources are taking a hit because of it, by the way).
This isn't just an economic dip or a structural pivot. It’s a full-blown identity crisis for a massive, multi-layered profession that honestly thought it was untouchable.
This existential crisis about an individual's actual usefulness in a sector points back to a really old concept... one theorized by Émile Durkheim in The Division of Labour in Society (he called it anomie) back in the late 1800s during the Industrial Revolution.
The human need to look ahead
No drama here, we’re just talking about the need for stability. Not economic stability pushed by some brand-new liberal market, no, but purely personal stability. The ability to see your own future. A human being needs a marked path. Until very recently, a developer’s career path was pretty linear, predictable, and reassuring (with the usual life bumps we all go through). But looking a bit deeper, the sector had these very rigid, schematic rules for the social escalator:
Learn the syntax and the basics of engineering.
Practice like crazy (the famous "eating lines of code") to actually become autonomous.
Move up to Senior, Tech Lead, or Architect roles (for the people who want that).
Today, the fog comes from the fact that Step 2, the part where you actually build your critical thinking and experience, is at risk of vanishing entirely, replaced by instant code generation. The future is a blur because devs literally don't know what skills to bet on anymore to guarantee they'll still be employable in 5 or 10 years.
In the end, we're just watching tech go through the exact same loop as other sectors. It’s normalizing. It’s becoming a classic industrial sector: like mining, the auto industry, textiles, canneries, fishing, whatever. A normalization complete with its own automation, the hunt for extreme profitability, and now, this new "Taylorism" settling in.
#SoftwareEngineering #TechIndustry #FutureOfWork #AgenticCoding #VibeCoding #CodeReview #AI #TechCrisis #SociologyOfWork
-
CW: Loooooooooooong read on the tech identity crisis, AI, and developer careers
Late-night thoughts... sparked by an engineer on Mastodon asking how, with all the "vibe coders" out there, the demand for actual, competent code reviews is ever going to be met: https://mastodon.world/@paulschoe/116894894916264050
So I just started writing...
The end of the tech exception
For almost thirty years, IT had this sort of "immunity." It was (yeah, was...) the ultimate safe-haven sector. Almost non-stop growth, incredibly comfortable (sometimes even indecent) conditions, and companies always begging for more people than were actually available. Cutthroat, sure, but the kind where you could just hop from gig to gig in Silicon Valley to always land a better deal.
But we have to look at two things. First, historical instability. The way we absorbed past major crises (oil shocks, the dot-com crash, subprime)... those were mostly economic or geopolitical. They hit the global economy, and tech felt the shockwaves, but never as hard as other sectors like real estate or the auto industry.
The second point is the current instability. Today, the crisis is homegrown. It’s coming from inside the tech industry itself (whether that’s Fintech, space tech, or even gov tech). Generative AI and agentic tools are changing the very core of how value is produced. (And the money and physical resources are taking a hit because of it, by the way).
This isn't just an economic dip or a structural pivot. It’s a full-blown identity crisis for a massive, multi-layered profession that honestly thought it was untouchable.
This existential crisis about an individual's actual usefulness in a sector points back to a really old concept... one theorized by Émile Durkheim in The Division of Labour in Society (he called it anomie) back in the late 1800s during the Industrial Revolution.
The human need to look ahead
No drama here, we’re just talking about the need for stability. Not economic stability pushed by some brand-new liberal market, no, but purely personal stability. The ability to see your own future. A human being needs a marked path. Until very recently, a developer’s career path was pretty linear, predictable, and reassuring (with the usual life bumps we all go through). But looking a bit deeper, the sector had these very rigid, schematic rules for the social escalator:
Learn the syntax and the basics of engineering.
Practice like crazy (the famous "eating lines of code") to actually become autonomous.
Move up to Senior, Tech Lead, or Architect roles (for the people who want that).
Today, the fog comes from the fact that Step 2, the part where you actually build your critical thinking and experience, is at risk of vanishing entirely, replaced by instant code generation. The future is a blur because devs literally don't know what skills to bet on anymore to guarantee they'll still be employable in 5 or 10 years.
In the end, we're just watching tech go through the exact same loop as other sectors. It’s normalizing. It’s becoming a classic industrial sector: like mining, the auto industry, textiles, canneries, fishing, whatever. A normalization complete with its own automation, the hunt for extreme profitability, and now, this new "Taylorism" settling in.
#SoftwareEngineering #TechIndustry #FutureOfWork #AgenticCoding #VibeCoding #CodeReview #AI #TechCrisis #SociologyOfWork
-
AI agents can recreate the old rockstar developer problem: impressive code that only one mind can explain.
The fix is shared design, local repo rules, small diffs, tests, and review that asks whether the change belongs.
https://www.the-main-thread.com/p/ai-agents-rockstar-developer-problem
-
AI agents can recreate the old rockstar developer problem: impressive code that only one mind can explain.
The fix is shared design, local repo rules, small diffs, tests, and review that asks whether the change belongs.
https://www.the-main-thread.com/p/ai-agents-rockstar-developer-problem
-
Great turnout at #nuedigital in #nuernberg! We talked about our research on #codereview (presented by my fabulous colleague @andreasbauer), tax compliance in software engineering, and our ridiculous scientific publication system. Three topics nobody warned us would generate quite so many questions, so many, in fact, that we ended up going quite overtime. 😂
Apparently, we picked the right topics for the right audience. Thanks to everyone who joined us!
-
Java‑код скомпилировался — это ещё не значит, что он работает
Есть целый класс Java‑багов, которые компилятор пропускает, а тесты на счастливом пути не ловят: код синтаксически корректен, но делает не то, что вы имели в виду. В статье разбираем пять таких самых частых ситуаций, которые встречаются у начинающих разработчиков.
https://habr.com/ru/companies/otus/articles/1047040/
#Java #staticanalysis #codereview #trywithresources #equalshashCode #SonarQube #NullPointerException #обучение
-
Coding agents do not need trust. Repositories need a rejection path.
I wrote about isolated workspaces, fast local hooks, merge-boundary enforcement, and why AI-generated diffs need explicit tool roles instead of prompt etiquette.
-
Coding agents do not need better manners. They need a change budget.
I wrote about scope limits, isolated workspaces, review cost, and why "human in the loop" is too weak if an agent can already turn a one-class task into a 14-file diff.
https://www.the-main-thread.com/p/coding-agents-change-budget
-
Free software isn't any more secure just because it's free!
When the Windows source code was leaked around the turn of the millennium, there were several attacks because there were flaws in how graphic formats were handled. So Windows was only secure as long as the source code remained secret.
Linux is secure even though its source code is public, which is the key difference from proprietary software.
Of course, that doesn’t mean that errors never occur in Linux or that security vulnerabilities can’t go unnoticed for a short time.
The security concept is based on the fact that, thanks to the open-source code, anyone can review the code, allowing bugs to be detected and fixed more quickly. While proprietary software keeps bugs secret and covers them up, Linux makes them transparent and public, and fixes them promptly. You can make fun of this, but it only shows that you haven’t understood the security concept.
Free software that is not tested or reviewed by anyone is not secure simply because it is free.
#meme #linux #arch #aur #update #cybersecurity #malware #security #fail #software #problem #hack #hacker #os #floss #foss #opensource #news #attack #check #control #qa #development #review #codereview #gnu #gpl #error #bug
-
On top of guaranteed uptime, you'll get better review capabilities: interdiffs between revisions, per-region commenting on images and docs, and built-in support for stacked changes.
Stay unblocked with Review Board.
https://www.reviewboard.org/get/
#GitHub #Microsoft #CodeReview #SoftwareDevelopment #DevTools
-
Creationist: Living things were clearly designed by a higher intelligence!
Scientists: Junk DNA and the recurrent laryngeal nerve are strong evidence against intelligent design.
God at the DNA code review: "I didn't check the code. It's mostly slop, but it works."
-
Am 02.07.2026 beim #nuedigital Festival in #nuernberg ?
Dann schaut gerne bei uns am Institut für angewandte Informatik der #ohm vorbei und diskutiert mit uns über das kaputte Publikationssystem in der Wissenschaft und wie wir es reparieren können – oder, für Software Engineers (und alle, die es werden wollen), über #codereview oder das 30-Milliarden-Dollar-Steuerproblem von Microsoft und warum Tax Compliance im Softwareentwicklung so kompliziert ist. Ich freue mich auf euch!
-
Anthropic's Mythos Preview Bolsters Vulnerability Discovery
Anthropic's Mythos Preview is delivering impressive results in vulnerability discovery, with one tester saying it's the closest thing yet to a straightforward find-something solution. Early trials show Mythos Preview excelling in source-code audits and tackling complex tasks like native-code and reverse-engineering…
#VulnerabilityDiscovery #ArtificialIntelligence #MachineLearning #CodeReview #SourceCodeAudit
-
🚀🤖 Oh, the #irony of trusting #AI with your code, only to spend eons automating doubt 🤔! The article is a riveting tale (or snooze-fest?) of someone rediscovering skepticism—like watching a toddler learn to walk, except with zeros and ones. Apparently, the solution to AI's flaws is to create more work for yourself—to scrutinize what you paid the AI to do in the first place! 🙄💻
https://www.alexself.dev/blog/automated-doubt #Automation #Skepticism #CodeReview #TechHumor #HackerNews #ngated