I just finished the first iteration of the big refactoring I was planning for a long time at work.
A nightmare for the reviewer, but the reviewer is me =)
I'm wondering now what the relation is between the number of changed lines and the number of bugs I introduced...
#work #job #SoftwareDevelopment #Flutter #mobileDevelopment #codereview
LLM as a Judge: опыт оптимизации генератора описаний Pull Request
Меня зовут Дмитрий Успенский, я работаю в команде ML RnD Техплатформы Городских сервисов Яндекса, и в статье я расскажу, как мы применили подход LLM as a judge — когда сама языковая модель оценивает качество генераций и сравнивает между собой разные варианты описаний. Поделюсь опытом определения критериев качества, сбора валидационного датасета, подбора промптов и выбора модели. Результаты оказались обнадёживающими: метод действительно позволяет улучшить генеративную систему без участия ручной разметки и асессоров.
https://habr.com/ru/companies/yandex/articles/907646/
#llm #AutomaticEvaluation #PullRequest #CodeGeneration #PromptEngineering #codereview
Flutter App Running Slow? Let Dartitude Audit and Fix It
Don’t let performance issues hurt your Flutter app’s success. Dartitude provides detailed Flutter App Audits that uncover bugs, inefficient code, UI glitches, and backend flaws. With our strategic fixes and tech consulting, you can elevate your app’s quality, security, and speed—all in one go.
Read More: https://dartitude.com/services/flutter-app-audit
#flutterappaudit #appperformance #flutterdevelopment #codereview #mobileappoptimization #apptesting
I am so fucking tired of arguing with people on why PR/commit descriptions are important.
You saving the 30-90 seconds from writing one costs so much more down the road from reviewers to gaining historical context.
Ok, so you didn't include one. No worries, get it next time. Now then arguing that you don't need it makes my fucking infuriated.
Also don't use AI here. I don't care what the PR is. I want to know why you coded they way you did. Whats the context. 2 sentences goes a long fucking way
I've on many occasion said I hate some word or phrase, but I don't think I've ever truly meant it, until "vibe coding".
A large part of my work is code reviews. I'm now seeing an increasing amount of influence from AI generated tripe.
It might be speeding up someone else's job, but mine just became much harder, challenging and time consuming, since I now have to go through the code twice as thoroughly.
Fuck AI.
The new RBCommons is live!
This update features Dark Mode, multi-commit reviews, a new-and-improved review experience, dashboard improvements, better code safety, mobile support, image review in diffs, new integrations, and much more.
https://blog.beanbaginc.com/2025/04/26/welcome-to-the-new-rbcommons/
I just pushed a new CRASS grep-it.sh version out, check it out: https://github.com/floyd-fuh/crass/blob/master/grep-it.sh #pentesting #codereview
Code reviews boost team knowledge and zap bugs early. Use Visual Studio Team Services' pull requests and branch policies to embed reviews in your coding routine! #CodeReview #VSTS https://www.funkysi1701.com/posts/2018/codereviews/ #Git #VisualStudio
I ensure the maximum quality standards on my code review comments.
Working on some #code written by an #intern a few years ago and it's taking all of my energy to just add the feature and not refactor the whole thing. What's the point of having an intern and not reviewing their code as they go to prevent these issues? So verbose, so much duplication 😡😡
I guess I've been writing tools all by myself here with no review either, hopefully whoever takes them over eventually feels like I designed them reasonably 🤷
Even skilled in-house #Rails teams benefit from a fresh, unbiased Rails App review – like a specialist's opinion. #reinteractive's Comprehensive Rails App Review offers that crucial external lens, going beyond typical internal checks.
Learn more: bit.ly/4llSnlq
gör CR o snubben har skapat en (totalt onödig dessutom) variabel:
var10000
...?
det luktar chatGpt
sorry, blir decline på den
I recently did a live demo of a "reproducibility audit" for @us_rse. Check it out: https://www.youtube.com/watch?v=Q2ZsLbBkWrk
Sometimes I don't understand whether UX is an excuse or a real obstacle. In the endless thread on "Allow commenting on the commit message in merge requests" here is an answer (link in the image caption). I can imagine some solutions in line with the existing UX, or hacks, but anything is better than nothing. The perfect UX does not exist, much like the perfect code.
#Gitbutler now has it's own review system (compatible with GH PRs) from the creator of GH PRs.
https://blog.gitbutler.com/gitbutlers-new-patch-based-code-review/
So yet another external review system showing how broken #githubs PR system really is.
We now have #ReviewStack (does anyone actually use this), Graphite, and GitButler in the Github compatible alternate review system along with the client side systems (#jj, #ghstacks, #sapling, et al)
However I think GitButler totally misses the mark and continues the the sins of PRs into thinking that you need to groups PRs together into a logic unit.
Phabricator and Gerrit have both supported stacking diffs and the power of stacking was that a change could be related to a feature (i.e building up to implementing it) or just dependent on other features in the stack.
Forcing that a "branch" is has to be a single feature and your commits are reviewable but you still can't stack branches is STILL a hurdle to productivity which is the entire point.
Also no command line for gitbutler makes it my least favorite by far.
I built a Disjoint Set as a #programming exercise, and I'd appreciate if any #elixir experts out there could give me some #codereview feedback.
https://github.com/dragon0/disjointset
Our next workshop will be presented by SORTEE president Ed Ivimey-Cook:
"Reproducible data analysis in R"
Ed will introduce the basics of reproducible projects, data & code, and conduct a code review.
Register at https://events.humanitix.com/sortee-workshop-reproducible-data-analysis-in-r
