Software Engineer | Independent Researcher | Edge Computing | AI Agents
Madrid, Spain
- Tweets 15,126
- Following 1,166
- Followers 8,433
- Likes 31,540
Hey, jovem! Estou cada vez menos utilizando redes sociais para publicar coisas e ficando mais passivo no rolê. Porém eu compartilho os artigos que leio e outros links no meu LinkedIn (sem postagem de motivação tosca)
Me segue lá
linkedin.com/in/william-oliv…
1
9
1,603
William Oliveira (bsky.app/profile/1ilhas) retweeted
He discovered Rails is slow today
Been running a beta version of Campfire rewritten in Rust today! The code itself is still ugly as sin, and 6x as verbose, but the conversion was free*, and I never had to look at the Rust directly, so this is awesome 🤩 github.com/basecamp/once-cam…
39
69
1,795
133,418
Big tech sendo Big tech
appleinsider.com/articles/26…
1
261
William Oliveira (bsky.app/profile/1ilhas) retweeted
🚨 BREAKING: vibe-coders have announced that databases are no longer needed:
237
67
995
107,761
William Oliveira (bsky.app/profile/1ilhas) retweeted
There’s a rank specifically for Google now? 😭😭
482
491
10,668
7,240,171
William Oliveira (bsky.app/profile/1ilhas) retweeted
De preferência sem uma próxima
21
628
4,343
41,453
William Oliveira (bsky.app/profile/1ilhas) retweeted
Faz todo o sentido usar NixOS.
Bora copiar, Brasil!
The Netherlands is moving government PCs from Windows to Linux.
The Dutch government is building DAWO, a new workplace based on NixOS and open-source software. Eight municipalities are already testing it.
The goal is to reduce dependence on Microsoft and other non-European tech providers for operating systems, productivity software and cloud infrastructure.
It’s still in the pilot stage, so this isn’t a nationwide Windows replacement yet.
But the Netherlands now joins a growing list of European governments actively looking beyond Windows and Microsoft 365.
Europe’s push for digital sovereignty is starting to reach the desktop.
11
12
193
5,604
William Oliveira (bsky.app/profile/1ilhas) retweeted
Sorry, son.
Dad is busy burning tokens.
62
143
1,849
93,139
Desculpe comunidade Pythonica (uma das primeiras comunidades que me abraçou). Mas, depois de ter trabalhado um tempo com Golang, fico triste de voltar a trabalhar com Python 😂
1
5
284
William Oliveira (bsky.app/profile/1ilhas) retweeted
There is no winning in this economy. 😭
42
6
119
2,510
William Oliveira (bsky.app/profile/1ilhas) retweeted
Se filtra el mando de PlayStation 6.
136
4,119
44,012
1,030,791
Always remember that you can try this solution
corsarr.woliveiras.com/en/
207
William Oliveira (bsky.app/profile/1ilhas) retweeted
Agents wrote terrible unit tests in every OO language I have used, with excessive use of mocks most times.
I am not surprised to be honest. To quote Joe Armstrong:
> The problem with object-oriented languages is they've got all this implicit environment that they carry around with them. You wanted a banana but what you got was a gorilla holding the banana and the entire jungle
So unit testing objects either rely on good design (yay) or mocks (boo), otherwise they cosplay as integration tests anyway.
This is not a “funcional is better” spiel. But we have to be aware the death of unit tests is a failure in design and AI automating the bad practices human have been doing for decades.
I agree with @thorstenball, I think unit tests are dead in the water.
The ones the models write are terrible, at best just doubling total LOC.
Inverting the testing approach - heavy e2e/black box/golden master, reaching for lower levels only if necessary, works better for me
38
43
557
48,526
William Oliveira (bsky.app/profile/1ilhas) retweeted
"while retaining code quality"
Just one engineer and a team of agents ported the GitHub Copilot agent runtime to Rust, shipping 800,000 lines of production code while retaining code quality. 🚀
How it worked 👇
github.blog/ai-and-ml/genera…
66
190
6,246
164,748
William Oliveira (bsky.app/profile/1ilhas) retweeted
There are a lot of people dunking on this. I have been critical of many things the Claude Code team has said in the last 6 months but this is not one of them.
Folks should use the excitement to teach and guide those incoming. Discuss pitfalls and why you can't formally verify all software. Show what is possible, explain what isn't.
I understand it can be overwhelming but that's the best outcome for everyone. This trend is much more valuable to us as an industry than "markdown files will describe how my system behave".
I used Opus 5.5 to formally verify the Claude Agent SDK using Lean. A couple short prompts = 16 PRs fixing various bugs and race conditions. Video attached.
TLA+ also works well. I sometimes combine Lean and TLA+ to look for issues around data flow, concurrency, and state mgmt.
I don't know either language well, but Claude is excellent at both. This approach is super useful for formally modeling your code and finding bugs that a human probably wouldn't have spotted.
Is formal verification the future of coding (or at least, bug finding)?
20
18
239
42,207