Angehefteter Tweet
"What is React" course released at
@Pluralsight Learn the basic architecture of React, what a React app looks like and what makes React awesome
@reactjs pluralsight.pxf.io/2r6ORa
English
Peter Kellner
10.5K posts
Peter Kellner
@pkellner
Software Developer, Architect, Pluralsight Author, React, .net and forever Learner. https://t.co/nQkNf5PUdz
Pre: 37.3034548,-121.95834 Beigetreten Eylül 2007
1.2K Folgt2.1K Follower
@tannerlinsley @jherr @quartzdevgg Hmmm. Dog Fooding also means treating this like the production site it is with minimal, and if anything, scheduled downtime. I’d suggest only test on prod staging for things like lighthouse (respectfully).
English
We 100% dogfood our tech on TanStack.com, but we also move *extremely* fast and are at an interesting cross section of sustainability experiments and traditional docs/open-source.
We're always improving. E.g. over the last 3 days, I was able to cut total JS loaded over a session by 80%, improve all lighthouse metrics, and make the site load way snappier.
You may have caught the site in the middle of some of those commits/deployments, so be sure to give it another shot!
English
I’m loving the idea of tanstack. But it really makes me question the product with how many issues the official site and docs have day over day. Like if it’s so problematic, and the framework is so good, atleast showcase that in the official site. This should be the #1 priority.
English
@MingtaKaivo @matteocollina And I also tell it not to change existing test to make them pass.
English
@matteocollina added "never delete or skip failing tests" to CLAUDE.md 4 months ago. still catches it trying to comment out assertions every few weeks. the fix that works: a pre-commit hook that diffs test count — any PR with fewer tests than before gets auto-rejected
English
Now that I’m using both Claude and Codex daily, I’m seeing that Claude tries to “cheat” more often, eg deleting a failing test.
What’s your experience?
English
@matteocollina And Claude often gets side tracked and will delete working features to make its job easier. Thankfully my test usually catch that.
English
@sebastienlorber I don't trust AI to get it right. It just enforces another layer of checking. I suppose if my tests were perfect it would not be needed ... but...
English
Honest question for React devs using AI today
What are you asking the AI to write components in TypeScript?
Are you taking into consideration that using another target language as an implementation detail could bring additional benefits? (faster server code, safe)
David Sancho@davesnx
@sebastienlorber nowadays, no react dev wants to write in anything other than TS, but back in the day Reason had all the attention. I would argue that a better language (most probably not ocaml, not reason) will emerge in this era - safer - faster - and spicy take: it will be prove-verified
English
@donnfelker That's so few on a big inventory of code. I suppose it's way better than nothing though. My worry is that the code looks so good, yet has such subtle bugs. Really hard to find often. Little stuff like index offsets. Oh my.
English
@pkellner Yeah, and I also limited to the number of lines that I wanted to change. So I’ll make sure that certain ref factoring are limited and scope to say 500 lines or 1000 lines so it’s easier to manage. Nobody wants to review an 8000 line PR.
English
A lot of people say that AI agents don't work well on large legacy codebases.
That's false.
Over the last eight months, I've been successfully using AI agents to clean up massive portions of legacy tech debt in a very large Java and Kotlin codebase that's over 10 years old and used by millions of people.
It works. You just have to set up the prompt and engineer the context management correctly.
English
I wish there was a @AnthropicAI CC option below the "Start Executing Plan" that said "put this plan in your memory and give me a prompt to start it in a fresh context". Just sayin...
English
@pkellner Also, the TanStack x account is still relatively young. Not a TON of amazing content there other than announcements that have also been echoed by everyone on the team. I'm wagering we'll be okay?
English
@tannerlinsley I have a lot of railroad tracks on my back for making decisions like I think you are about to make. You have so much goodwill wrapped up in those old links, sometimes just best to live with it.
English
@pkellner People get it wrong all the time. Then they're confused when they see a horse's face instead of the island lol.
English
Why did it have access? If a junior engineer takes down prod, it's never the junior's fault. That's an access and SDLC issue
Alexey Grigorev@Al_Grigor
Claude Code wiped our production database with a Terraform command. It took down the DataTalksClub course platform and 2.5 years of submissions: homework, projects, and leaderboards. Automated snapshots were gone too. In the newsletter, I wrote the full timeline + what I changed so this doesn't happen again. If you use Terraform (or let agents touch infra), this is a good story for you to read. alexeyondata.substack.com/p/how-i-droppe…
English
@sorenbs @tannerlinsley @prisma @tan_stack This is a huge endorsement by a company the knows more about data then anyone other I know.
English
@prisma Studio now uses @tan_stack DB for ui and data state management. This allows us to add new features with confidence.
English
@jherr Devs for devs? That sounds like we are going back to @sv_code_camp days! Having organized 15 of those, I’m in! (But not holding my breath for a comeback).
English
Three simple reasons why you'll love working with TanStack Start SSR. youtu.be/EnzKA0fQkBM TanStack was written by devs for devs and the features in start make it easy to build the kind of app you want, just the way you want it.
YouTube
English
@eelcodotdev I’m finding codex is better at targeting problem. “Fix this exact bug” but opus better at implementation of bigger rewrites. Like you, I do see opus spinning a lot
English
Biggest difference between Opus 4.6 and Codex 5.3 seems to be that Opus very often wastes a lot of tokens on find/grep often adding a lot of unrelated/irrelevant code, bloating the context window.
Codex is much more effective.
Other people experiencing this too?
English
Have heard from so many different engineers lately that:
1. They haven’t handwritten code since like November
2. They’ve churned off Cursor and only use Claude Code
3. A lot of them spin up & manage CC instances on their phones
Some big inflection point was hit the last 3 months
English
@DavidKPiano this is why I keep my $50/month chatgpt team (which gives me a bunch of Codex) around. I find it a little better than opus 4.6 but I run out of gas after a couple days (these are those days now)
English
@paulg Do you seriously think he paid a penny of tax to California?
English
Larry Page is gone. He wasn't just pretending to move to Florida. He has moved. The proposed wealth tax hasn't even passed, and already it has cost California both Larry's presence and all the tax revenue it made from him.
English
@John_Papa sorry to hear you have to close so many of your restaurants. 😀 cnn.com/2026/02/26/foo…
English
@sionemart Correct. Desperate times call for desperate measures for Cloudflare?
English
Brutal.
While I like Cloudflare and use their products, their Vinext thing was a really immature move from them, especially with the kind of influence they have, getting even some government sites to use a pre-alpha framework.
I would expect it from a smaller startup trying to get some X clout, but it felt deceiving coming from Cloudflare.
Guillermo Rauch@rauchg
We've identified, responsibly disclosed, and confirmed 2 critical, 2 high, 2 medium, 1 low security vulnerabilities in Cloudflare's vibe-coded framework Vinext. We believe the security of the internet is the highest priority, especially in the age of AI. Vibe coding is a useful tool, especially when used responsibly. Our security research and framework teams are extending their help and expertise to Cloudflare in the interest of the public internet's security.
English