Jamie Matthews (@j4mie) 's Twitter Profile
Jamie Matthews

@j4mie

Web engineer. Python, Django. Sustainability, clean tech, fuzz pedals and dynamic languages. Co-Founder @dabapps, opinions my own

ID: 641903

linkhttps://www.mtth.org calendar_today16-01-2007 00:28:56

6,6K Tweet

1,1K Takipçi

1,1K Takip Edilen

Jamie Matthews (@j4mie) 's Twitter Profile Photo

Released django-readers 2.4.0 with support for Django 5.2 and a nice quality of life improvement for anyone using the DRF integration with schema generation: django-readers.org/community/chan…

Eleanor Berger (@intellectronica) 's Twitter Profile Photo

I've been programming for most of my life - over 30 years. During this time I have created a large number of bugs, and fixed almost as many. Of these bugs, maybe 1% have been typing-related. The typing obsession is cargo cult. Fix the damn logic.

Jamie Matthews (@j4mie) 's Twitter Profile Photo

Twitter: the population is crashing! Everyone must have lots of kids! Also Twitter: the most important thing everyone can do to live to 500 is minimise stress and get plenty of sleep

National Energy System Operator (@neso_energy) 's Twitter Profile Photo

Great Britain achieved a new maximum solar record on 8 July between 12:00 – 12.30pm. During this time, solar generated 14GW, which was nearly 40% of the electricity mix. ⚡ Download our app to see real time generation data and records. 👇

Flowers (@flowersslop) 's Twitter Profile Photo

When you use local models, you realize even more how bizarre and magical it all is. My GPU is thinking about how to code doodle jump using human thoughts. That would literally have been like black magic just a few years ago.

When you use local models, you realize even more how bizarre and magical it all is. My GPU is thinking about how to code doodle jump using human thoughts. That would literally have been like black magic just a few years ago.
Jamie Matthews (@j4mie) 's Twitter Profile Photo

ChatGPT needs “voice note mode”. I don’t use them at all with actual humans but the asynchronous nature of seamlessly sending some audio and then getting a reply (also audio) sometime later feels like it would work perfectly for LLMs. Like advanced voice mode but without the

Jamie Matthews (@j4mie) 's Twitter Profile Photo

Looks like OpenAI are preparing a public Skills repo at github.com/openai/skills (currently 404) that Codex fetches and caches locally: github.com/openai/codex/c… cc Simon Willison Eleanor Berger

Jamie Matthews (@j4mie) 's Twitter Profile Photo

Don’t agree with DHH on everything, but this post nails it, and I will always be firmly in camp dynamic. As an aside: to everyone arguing “LLMs work better with statically typed languages”. First, I haven’t found this to be remotely true, but furthermore I think we’re

Jamie Matthews (@j4mie) 's Twitter Profile Photo

“We are at an inflection point where a simpler approach is now the best approach.” Spend your entire career following these inflection points.

Jamie Matthews (@j4mie) 's Twitter Profile Photo

Why don’t any other browsers support container tabs like Firefox? Seems like the only sensible primitive for working with multiple sets of cookies for the same site. And no other browser supports it. How do people who don’t use Firefox get anything done? support.mozilla.org/en-US/kb/how-u…