Christian Findlay (@CFDevelop) header
Christian Findlay (@CFDevelop) profile picture

Working on dev tools

Melbourne, Australia
Joined January 2019
Photo by Christian Findlay (@CFDevelop)Photo by Christian Findlay (@CFDevelop)Photo by Christian Findlay (@CFDevelop)Photo by Christian Findlay (@CFDevelop)Photo by Christian Findlay (@CFDevelop)Photo by Christian Findlay (@CFDevelop)Photo by Christian Findlay (@CFDevelop)Photo by Christian Findlay (@CFDevelop)Photo by Christian Findlay (@CFDevelop)Photo by Christian Findlay (@CFDevelop)
Christian Findlay retweeted
It’s almost 2027 and developers still doubt coding agents’ usefulness and practicality. I’m sorry but you’re out of your mind if you believe handwriting code, manually moving files, and checking logs yourself is better than having an agent one-shot it.
8
4
1
30
1,703
Christian Findlay retweeted
Change is the only constant. Adapt or die. AI won’t substitute us, but it will substitute those who don’t adapt. They’ll see the change from the benches.
It’s almost 2027 and developers still doubt coding agents’ usefulness and practicality. I’m sorry but you’re out of your mind if you believe handwriting code, manually moving files, and checking logs yourself is better than having an agent one-shot it.
1
2
8
357
Christian Findlay retweeted
Jensens "Hung" Huang details his past week
116
1,302
175
7,913
955,608
Real-time procedural drum synthesis in #Flutter with flutter_soloud! Using the new RealtimeStreamHandle.startFloat32Feeder API, DSP runs in a background Dart Isolate with low-latency renderAheadFrames & devicePeriodFrames, routed through a Bus with freeverbFilter.
5
4
41
1,837
I have looked at Deslop from the point of view of someone who is writing a complementary FOSS Slop Audit and I agree. A multilingual deduplicator is NOT trivial. This is hard work but there is a crying need for it.
Pay attention here My contention is that code deduplication is a hard computer science problem and we’re only about 60% of the way into the journey Deslop is on the journey. Please check it out and the research behind it.
1
1
4
244
Christian Findlay retweeted
Nice, just asked Opus 5.5 to run it against my Typescript/Swift project: „What I think … It's useful as an occasional review: fast, and the top production hits are real …“ 😎🎉
2
1
2
400
Christian Findlay retweeted
I spent this morning trying Deslop (deslop.live/), a duplicate-code finder built by my friend @CFDevelop . I ran it across something I'm build for my agent-practice monorepo: TypeScript server and web app, Python tooling, 61k lines. The plain scan is fast and useful so can be used on a CI step. Here's what it found: - 341 files in about 6 seconds, ranked worst first, with HTML, JSON and text reports. - 534 groups of duplicated code, 9.5% of lines. - Its top hit was a 2,000-line protocol file we copy on purpose. One exclude line fixes that. - The real finds were in production code. One response-handling block was copied across three API routes, and one data-polling hook was copied in two views. Both fold into a single function. It compares syntax trees, not lines, so it catches renamed copies and not just pasted ones. It compares syntax trees, not lines, so it catches renamed copies and not just pasted ones. It also has an MCP server, so a coding agent can check before it writes. I gave Claude Code a hook it was about to write, and find-similar pointed it straight at the existing copy. That's the "don't write a third one" loop I want in every agent session. The optional embedding layer looks for code that does the same thing but is written differently. - I pointed it at Ollama (nomic-embed-text) on my homelab GPU. - It found something the plain scan couldn't: two versions of the same git helper in two different reporters, written in different styles. - That costs about 20 minutes on a first run against 6 seconds, but its cache makes a second run's embeddings almost free. It take a far while to run with embeddings so I'll run it as a review, not as a CI gate. Definitely sorth a run on your own repo: github.com/Nimblesite/Deslop
5
2
2
9
1,449
Analog Indie devs.
1
3
312
Christian Findlay retweeted
I just keep thinking about how AI is really exposing how fundamentally broken traditional education and learning systems always were. Education failed me because school expected me to learn in this rigid straight line and treated the way my brain worked as the problem. But now with AI I have this endless non linear curiosity tool that moves at the same speed as my brain. As a neurodivergent person did I actually lack focus or was I just starved of an environment that didn't punish me for how I naturally like to learn...
21
14
3
114
3,519
The punchline to this is his own product is now turning against him to make memes 🫡🤣🤪
1
3
2
13
6,881
I laughed so hard I cried. God damn this was well done.
311
2,419
379
14,590
1,936,058
Christian Findlay retweeted
I know I can. I analyzed 200 established, community-recognized production applications (50 each in Python, Java, TypeScript and C#), every one with at least 100 GitHub stars, at least 10,000 lines of production code, a year or more of active history and a real test suite. 99% of typical enterprise systems (outside of React front-ends) turned out to be structurally incapable of being fully verified through testing. try.slopaudit.org
Can we now admit that human slop is so insecure that nobody should use it? Can we admit that AI, despite its flaws, offers the only path forward for gradually detecting and fixing software that is actually just a pyramid of swiss cheese?
4
3
1
10
1,432
Christian Findlay retweeted
Likes and YouTube subscriptions appreciated This is a tutorial for writing C# and F# in VS Code youtu.be/wUbS4UYQbMQ
1
2
157
It amazes me that more people aren't solving this with automation. It can be done. I'm doing it. So is @CFDevelop.
Anthropic’s report is basically saying humans are the bottleneck. “Amid this model-driven acceleration, human attention is a constant bottleneck. Claude sessions amount to around 80% idly waiting for human input. As one engineer put it, people are “bottlenecking the agents.”
3
1
2
389
Christian Findlay retweeted
Koru now has an experimental(!) and partial C#/.NET-backend. Our effect-branches/generators emit to static calls so the performance is good, as expected. Test coverage nowhere near Zig, but a fun start. Check it out here: korulang.org/blog/cs-backend
3
3
55
2,919
Christian Findlay retweeted
How do you build a car app that works for everyone? You talk to everyone! 🏎️💬 Learn how Toyota is leveraging Flutter to rebuild its infotainment system and the UX research process that helped refine their software.
1
6
3
101
10,306
Christian Findlay retweeted
Had to do imagine out of it..
🤖 Made with AI
AI is literally slot machines for devs 🎰 You put your tokens in and sit around waiting for the dopamine hit of software tumbling out the hole
1
1
1
2
179
Christian Findlay retweeted
The claude folks should just own this 🤣
221
600
263
4,018
704,313
Christian Findlay retweeted
Claude has discovered a previously unknown enzyme system hidden in the DNA of bacteriophages. Beside the enzyme’s gene sits a long array of repeating DNA—a structure that looks somewhat similar to CRISPR. We don’t yet understand what this system does, but only a handful of known systems share its features, and all of them are able to cut, copy, and paste DNA. Historically, the discovery of such programmable systems has helped revolutionize medicine. CRISPR, for instance, is now the foundation of genetic medicines. But it will take much more work to learn what this system does, and whether it can be put to similar use. Read more: anthropic.com/news/claude-di…
1,545
5,274
2,638
40,711
25,750,420