Emma 🐼 retweeted
Just realized T3 Code can automatically update remote servers, too. If you’ve ever had to handle those updates yourself, you’ll understand why that matters.
We just shipped an updated T3 Code Stable release with over 1,400 changes 🤯
Quick highlight reel of my favorites:
- Massive perf wins on desktop and mobile
- Add a project to another machine from the "Run on" menu
- EAS Simulator remote hosts for iOS dev
- PR diffs now show code vs tests vs docs at a glance
- Idle Claude threads compact early so the cache stays warm (beta)
- Mobile got a big native pass. It finally feels native
- `t3 doctor` can reclaim database space
- `t3 connect` now checks your machine is reachable, and errors say how to fix them
- Webhook tasks got filters and dedupe
- Skill discovery is much more reliable
- Tons of Windows, terminal, and scroll fixes
Least favorite: I caved, LaTeX renders on web and mobile
Also heads up, we're getting a bit more experimental with how we treat Nightly, so now would be a good time to move back to stable :)
Emma 🐼 retweeted
This is the real Theo btw. If you stop for a second and actually engage with what the says and does regardless of the stupid YouTube persona - sorry dude - he’s actually a decent human being and will go out of his way to actually bring value and help
Can I be real for a sec? I'm super nervous about this launch. It's something I never thought I would do, and I want to be as transparent as possible.
This course is $1,300. You should not buy it. Let me explain.
I've done YouTube for over 5 years now. I've never done a paid course of any form before. I have worked hard to keep all of my content freely available.
I'm lucky to have found so many awesome companies that will sponsor short sections of my videos, enabling me to employ an incredible team without compromising my content.
I have always hated the idea of charging my viewers. If I can't convince a company to cover our costs, that's my failure, and I don't like putting that failure on my audience.
On October 26th, my first-ever paid course with @megadevhq goes live. It's $1300 a seat. That's really expensive.
I keep imagining a scenario and it makes me feel sick. I think of a young, excited dev who's recently gotten into my content. They're scared of falling behind, so they cut their expenses and save up the money, because they believe my course might be the key to getting a real dev job.
Even now as I type this, I feel my heart rate increasing and my stomach flipping over itself. The thought of anyone putting themselves in a harder financial position, all because of me?
This might actually be my biggest fear.
I don't want to sell promises. I don't want people paying me for hope.
So why did I do this course with Mega?
Two reasons: the people and the content.
I was hit up by Joel about this course late last year. I trust him with my life. Even then, it wouldn't have been enough to convince me.
Then I learned the other instructors who would be involved. Kent, Angie and John. All three were fundamental in my growth as an engineer. Getting to work with all of them was far too exciting. Made it really hard to say no.
If I had nothing unique to say, I still would have said no. I'm incredibly busy as-is, and I'm running low on space to even show you guys all the stuff I do.
What pushed me over the line was the realization that there is so much I want to say that doesn't fit in a YouTube video.
I just did a video about "prompting better". It's a deep dive on my real world workflows and how I think about building with AI. The only way I could get people to watch it was clickbaiting the title, "your prompts suck".
What if I wasn't bound by clickability? What if I didn't have to focus on being entertaining for the whole video runtime? What if I was freed from the constrains of making content for the 500,000+ people who watch my videos?
I love YouTube. I love my audience. I love the videos I make. But I'd also love to do a multi-part deep dive on "building with AI in real world code bases" that I know would bomb on YouTube.
That's what I filmed for Mega. I'm so proud of it. It's fundamentally different from any videos I've done before, but still feels like me just doing my thing.
Is it worth $1,300, though? If that $1,300 comes out of the same bank account that you pay rent with, probably not.
Is it worth $1,300 per person to a company trying to help their eng teams use AI properly at work? In my opinion, this is a bargain.
If you're paying your engineers $100k a year, and this course makes them 5% more effective at their jobs, that's a 4x ROI.
What I'm trying to say is simple: Don't buy this course.
Get your boss to buy it.
(If you have an education budget at your job that works too. Just please don't pay for this yourself. Seriously.)
Sorry for the rant. If you can't tell, I'm genuinely quite nervous about this 🙃
Emma 🐼 retweeted
Retweet aggressively.
Many people don't know the date for the elections.
APC is banking on this point.
Voter apathy = Successful Rigging.
Emma 🐼 retweeted
I recently built a handy little toolbox for my API work in Laravel, so thought I'd write about it a little so you can see how it works and why you should use it for your next API.
juststeveking.com/articles/m…
Emma 🐼 retweeted
T3 Code is remote-env-first experience.
I love this and don't ever plan to change it. That said, it has a bad habit of turning basic feature additions into complex jigsaw puzzles
Emma 🐼 retweeted
Just realized something.
Because I pursue accuracy above all else, I tend to make enemies with those who benefit from people being misinformed.
I did not tell him to hide anything.
I hit him up to ask about why T3 Code scored low. Learned he was on a 3+ month old version. He tried the modern version with our new orchestrator, realized his scores were not good, and adjusted.
He mentioned wanting to make a real benchmark after, so I offered to fund it because there’s no good research like this going on right now. He actually found a few things I’ve been rabbit hole-ing on all weekend.
I have no idea what Cloudroom is. I definitely did not tell him to remove it or anything else.
Only commentary I had on other solutions was that I was curious why a shitty clone of our app scored higher than us initially. Nothing to do with cloudroom.
Emma 🐼 retweeted
It’s never been easier to ship code, but testing has still been a massive bottleneck for us… so I decided to change that 🫡
With @Cap, we’re currently making a full transition to GPUI, and a screen recorder is a strange thing to test reliably, because you can export a perfectly playable video that has the wrong frames in it. Everything appears to work until you look closer!
I don’t want to ship slop, and any regressions are my worst nightmare. The faster we’re shipping code (and generously from contributors!) the more time I find myself spending trying to 100% prove that we haven’t broken something
So I’ve spent the last couple of days building an automated testing harness for Cap… 🥁
It runs our actual media code through repeatable scenarios whenever a relevant PR changes it and runs simultaneously across macOS, Windows and Linux. Also made it so that a draft gets a smaller suite so feedback comes back quicker (this is great for agents!)
The interesting part though is how we know whether the video is correct…
To do this, we generate test videos with an ID encoded into every frame, alongside audio markers placed at exact moments. These are our video fixtures. When we put one through Cap’s exporter, we can decode the result and check which frames actually made it through
So a cut would change which frame should appear next, or maybe a speed change alters when it should appear. We calculate that independently of Cap’s implementation, so the test has something to check the output against
For example, if frame 125 appears where frame 126 should be, we will know. We don’t need someone to sit there watching for it!
Performance gets measured too. The PR and its merge base are built on the same machine, then benchmarked in alternating runs after warming up. That gives us a comparison of CPU time before and after the change, while reducing the noise you’d get from comparing unrelated CI runs
We’ve moved most of this onto @useblacksmith . One Windows test batch went from 9m 52s on GitHub-hosted runners to 2m 59s, about 3.3x faster (!!) Another went from 5m 08s to 45s. Nearly 7x faster for that test step (!!!)
What I hadn’t quite expected though was how much we’d fix while building the actual harness itself
In one crash recovery test, the footage lost went from 11.03 seconds to 1.03 seconds. We also fixed a decoder bug that could show a frame from 89 frames earlier after seeking, without meaningfully increasing CPU usage
There’s something almost self healing about having this in a loop. We get a reproducible failure, fix the cause, and the test stays behind to catch it if it ever comes back. Each fix leaves us less dependent on remembering everything that’s gone wrong before
I think now that we have this in place, the logical next step is to scale it up and run performance/regression testing autonomously from things like Sentry logs and user feedback. Definitely going to commit a decent amount of $ to it. I probably need at least another 4x Claude subs!
Really excited to ship the improvements we’ve already made just from building this. With the harness mostly in place, I can move much faster on my own changes and contributor PRs. Being able to make that progress without the constant worry of breaking someone’s recording feels pretty fucking good I can’t lie 🤩
Emma 🐼 retweeted
Building Brimble is the hardest thing i've done in my life and to be honest I am kind of suprised to see it's still standing, it's so crazy because we don't support local builders enough and if it's in S.F it's the other way round
1) You get constantly compared to vercel/railway and it's sad because if you want to get people in this space to use it to you need to ship features to the quality they have already set (either same quality or you go higher)
2) After shipping almost the same quality, local users here still force you to make decisions by cutting cost of your product, just because you want them to use it and at the end of the day, you still get compared and you even have no choice.
3) I and abraham have been constantly bootstrapping the product for the last 4 years, so we only scale on demand. We don't have the luxury to throw money into compute unncessarily.
4) There were different iterations of the product just to get PMF, the directions changed about 2-3 times between the years.
5) oh fine, "enterprise" has the money. But at the same time, you need a good amount of money to get the licenses to support enterprise and at the same time, some enterprise will look down on you especially when it's just 2 engineers managing the product, ever since we got the idea in 2022.
You do performance and security audits yourself, you need to be extremely stable (just with a team of 2 and after all that, you still get compared. It's very much fine, but nobody gives a damn)
6) Brimble literally has everything (databases, sandboxes, fullstack deployments, autoscaling, static outbound ips e.t.c). Strived to succeed
7) You are also trying to pitch to investors, in a case where some of them don't have a clue or the nuances involved in what you are building. The best everyone has to do is support local products, supporting local products even gives you as a user a ton of advantage. You control the type of features you want, the prices is very affordable for you, you'd get the maximum support you need without having to send emails or wait for 3 business days to get feedbacks.
A few angels really believes in the vision and fingers crossed, really love it.
But at the end of the day, it still seems you aren't good enough for people to use you.
I'm extremely grateful for people that took all of these out of their mind to actually support us in terms of using the product end to end. Sharing reasonable feedbacks and also sticking around even when it had issues.
If i have the opportunity to do this again, i would peacefully decline.
Emma 🐼 retweeted
Thanks to an awesome skill I found, I thought adding explainers to pages would be pretty cool for visual learners.
What do y'all think: apiguide.dev/guides/idempote…
Emma 🐼 retweeted
Just got my first ever GitHub sponsor - and it's from @cap!
I've been using Cap for ages and genuinely love it, so getting sponsored to make it faster is kinda crazy. 11 PRs in, 5 merged, not stopping now :D
Massive thanks @richiemcilroy 🫡
Emma 🐼 retweeted
We’re scaling up automated testing for @Cap and need some Mac minis.
On AWS, five M4s would cost $4,490/month. Commit to a year and pay upfront: $45,945.
I could buy five refurbished M4 minis with the same 24GB RAM for ~$4,645 total, stick them in a rack and run tests around the clock.
There’s power, internet and some maintenance, sure. But that’s roughly $41k left over, and we’d actually own the machines.
Why shouldn’t we do this?
Emma 🐼 retweeted
We've sponsored 7 contributors so far today already with 15 different PR's merged!
All PR's directly making the GPUI version of Cap much more reliable and stable ahead of launch soon 👀
