In our projects, pull‑request reviews are often delayed by the comment, “CI failures are not related to my changes!” A newly published blog post, “We Don't Merge into a Broken Master Branch,” tackles this issue. The advice is simple: if the CI jobs don't pass, don't open a pull request. What if the team takes too long to fix the build? What if it never turns green? You still wait. Do not submit a pull request to a repository with a broken build.
Где вы читаете айти новости в Телеграм? Разыскивается канал, где можно было бы несколько раз в день увидеть самое важное из мира компьютеров: новые технологии, продукты, законы, инвестиции, назначения, увольнения, книги, научные статьи, конференции, мнения, доклады, аналитика, статистика. Без мемов и рекламы, само собой.
p.s. Есть среди вас журналист, готовый вести такой канал (за деньги)?
p.s. Есть среди вас журналист, готовый вести такой канал (за деньги)?
According to Joel Spolsky, “the only person who can close a bug is the person who opened it in the first place.” But, why? Why shouldn’t the bug fixer close it, when it’s obvious that the bug is fixed? There are at least three reasons. Read the blog post I’ve just published: Let the Bug Reporter Have the Last Word (discuss on reddit).
In a few of our GitHub repositories, we've started using issue-title-ai GitHub Actions plugin — recently created by @horw — and it works great. Every time a new issue is submitted, it automatically improves the title using the OpenAI API. As a result, our backlog looks cleaner and is easier to navigate. Definitely recommend!
This blog post explains it in more details: Good Title — Good Bug Report.
This blog post explains it in more details: Good Title — Good Bug Report.
"Настоящее несчастье — это эгоизм. Если любить только себя, то с приходом тяжелых жизненных испытаний человек проклинает свою судьбу и переживает страшные муки. А где есть любовь и забота о других, там нет отчаяния." — Феликс Дзержинский.
С праздником созидательного труда и справедливого распределения его плодов, товарищи!
С праздником созидательного труда и справедливого распределения его плодов, товарищи!
Разыскивается архитектор для создания нового телеграм бота, с ИИ функционалом. Стек не важен. Работа полностью удаленная. Частичная занятость. Бюджета хватит и на архитектора и на команду из пяти человек, все удаленно. Присылайте сразу свой GitHub аккаунт, резюме не нужно.
Опубликовали интервью с Еленой Владимировной Андреевой, к.ф-м.н, заведующей кафедрой информатики в школе-интернате им. А. Н. Колмогорова при МГУ. Под ее руководством школьники многие годы побеждают на всероссийских олимпиадах по информатике. Смотрите на YouTube и VK (75 минут, with English and Russian subtitles).
Have you heard about Takes, a Java web framework? Unlike Spring and others, it is object-oriented by design, which makes it easy to configure and test. We need a similar framework for Ruby. If you're interested in building one, I'd be happy to contribute as well.
Опубликовали интервью с Вадимом Петроченковым, одним из ключевых контрибьютеров в компилятор языка Rust — maintainer-ом команды Rust Compiler Team. Вадим рассказал, как удается совмещать full-time работу в Huawei и open source хобби, как работает Rust community, как развивается язык, за что Rust любят и за что не очень. Смотрите на YouTube и VK—51 минута на русском, with English subtitles.
GitHub recently introduced issue types. In my opinion, this is a methodological mistake—every ticket in a bug tracker should be a bug, not a task or a feature request. Mozilla team even published a paper about this, in 2002 (we discussed it in the OSBP course, last year, in this lecture). However, if Microsoft insists on categorizing issues, where’s the "question" type? Just one more reason to loathe Microsoft :)
Nine years ago, I gave a talk at the DevTernity conference (yes, that one) about XDSD and @zerocracy. If you're wondering what Zerocracy is and why its "pay-by-result" methodology matters, you might want to check out this 48-minute talk on YouTube—it's in English with Russian subtitles.
Now it's time to submit your open source project to KaiCode'25 competition (two weeks left until the deadline). You may win one of these prizes. Just fill out this form, it's free to participate. You may also want to join the group where we discuss the festival: @kaicode_org
A bug report is not a discussion board. It's a contract between its author and the team. Just published a new blog post about this: No BTW in Bug Reports. (image by Sora)
We released a new MCP server: aibolit-mcp-server (written in TypeScript). Add it to Claude Code (for example), then ask: “Find the most critical design issue in my Java class and fix it.” Claude Code will forward the request to the MCP server. The server will invoke Aibolit—a command-line tool (written in Python) that you must install beforehand. Aibolit will analyze your Java file, identify the most important design flaw needing refactoring, and return the result to the MCP server. The server will then respond to Claude Code, which will fix the issue accordingly.
Long story short: we need your help with Aibolit—a Python command-line tool that uses machine learning to detect issues in Java code. The backlog is full of ideas, and new contributions are always welcome.
Long story short: we need your help with Aibolit—a Python command-line tool that uses machine learning to detect issues in Java code. The backlog is full of ideas, and new contributions are always welcome.
Requirements engineering is the discipline almost killed by Agilists: we don't manage requirements anymore, we talk to customers instead, trying to become their friends. Don't be like them, read this #book instead: Software Requirements by Karl Wiegers.
How do you test your Bash scripts? Good question, right? I found an xUnit framework for that: shunit2. It works like a charm on both Linux and macOS. I'm now planning to migrate a hundred tests in the yegor256/cam repository to this framework. Want to help? Jump in.
Update: I don't like it, because it doesn't isolate tests — while this is one of the most important principles of unit testing.
Update: I don't like it, because it doesn't isolate tests — while this is one of the most important principles of unit testing.