Frontend developers are officially cooked

10+ years of coding and I just learned today that `cd -` will take you back to the previous working directory... makes you wonder what other hundreds of shortcuts I haven't learned

Y'all have been doing sql queries wrong...

Is this just what I should be expecting in the javascript ecosystem on the backend at this point? I refactored my project to hono + bun which starts off around 50mb, but after a few days it's somehow up to 300mb memory usage after a few days. Like when exactly does the GC kick in? Is this a memory leak somewhere? This is for wdcstarterkit.com, it's literally one page serving some static assets and one api endpoint for subscriptions. Also using drizzle to connect to a turso database to insert those subscription records. There is no way a simple rest API serving some static images should be needing 300mb of memory to host a single page. I'm now testing out two identical sites, one using hono + bun and one with hono + node to verify if the issue is bun or node. After that, I'm about to refactor this into Go and see if I run into similar issues. Am I maybe overlooking something else? Could it be railway? Could it be something inside my container? I'm using a node-slim container which should be pretty small to begin with and not running much.

Let's breakdown his peak engineering response. His app, which does background removal in the browser, using wasm, takes about 9 seconds to process 3 images on my M3 MacBook Pro, 11 cores, 16gb ram, probably decent GPU. He obviously runs them in some type of task queue (as you can see in his debug console logs) because the processing is probably a little resource intensive. So, now let's assume the business requirements (which he didn't care to ask about btw) states a user wants to run the removal process over ALL photos in their album after clicking a button. Let's say they have a couple thousand photos. Btw, their photos are already stored in S3, and the average size is around 3mb+ in size each. Btw, they only run this process once or so, so it makes sense to just save the output (oh, so that means like back into S3??). 1274 photos * 3 seconds each = 63.7 minutes to process one user's album, sequentially. ouch ok, so how can we speed this up? We could rent a large VPS which might be able to process a lot of these fast. Just spend a bunch of time tweaking it so you don't allow the removal code to brick your machine. Maybe we can even dynamically spin up some larger g4dn (the ones with GPU access) EC2 instances and churn through these fast! Let me just spend a lot of engineering time to figure that all out and find out it still fails to get the job done fast. or... here me out, what if we just sent 1274 events to an SQS queue, and process ALL those events using lambda in a matter of seconds. Like yes, this literally takes seconds. Each spins up it's own container, downloads the images from s3, processes it, stores it back to s3, now the user can always access his processed image in the future. And I don't even need to worry about cpu contention or running out of memory because that shit just scales. ahh you didn't think of that... why don't you stick to tinkering with your little meme board buddy.

I work 9-5, have a family, manage to publish videos consistently, work on side projects, run 2-3 miles daily, but you can’t even finish watching one 30 minute coding tutorial. You don’t need guidance. you don’t need a mentor. You need discipline.

Remember when mongodb came out and everyone got so hyped and started using it just to end up switching back to sql? Remember when microservices got hyped but now the industry recommends use a monolith first? Remember when graphql came out and everyone started hyping it up until they ran into tons of short comings? Give new tech 5 years and reevaluate. Adopt that mindset and become a senior engineer.

The dumbest take I've heard is when people get upset that people fork MIT license code, build something new on top, and make money. If you don't want people to "steal" your code, don't release it with a license thats says "do whatever you want with it.".

Bro, just roll your own auth, it's easy! Besides, all you need to do is hash a password right?

My wife wants to learn to code. Rust or C++ sounds like a good starting point right?

The amount of people saying “cd ..” is the same as “cd -“ with such confidence beautifully highlights the bell curve of this industry.

Nextjs might be the only framework I’ve used where clicking on a link on localhost often takes more than 1 second to actual load the next route. This should be instantaneous, especially since my database is local.

Remote work is individual productivity at the expensive of group comradery.

y'all, don't be sleep'n on this new css library...

I've been working remote since 2015, and the best advice I can give a company or team trying to do fully remote is to track the response time for slack message. If you ask a question and it takes hours for someone to respond, you're NGMI. 5 minutes max.

stop living in denial

not a single server was used

bro, try laravel it's so simple compared to next. this is a todo list application:

The rise of slop development. If the language is loosely typed and dynamic, you’re doing yourself a disservice. Even typescript is a joke. Please just use Go or Java like a professional.

Me too

PM: "How many story points to add drag n drop?" Dev Team: "at least a 8 or 13" Dev using Cursor:

I’m happy to say I’ve been married 8 years to my beautiful wife! Thanks for all the support you provide to me and the channel babe! I wouldn’t be where I am today without you 😘. Happy anniversary.

You just don't understand Next.js and React, they are all about composition! oh wait, you can't import a server action defined in your top level page into your form client component, you actually need to split that into an actions.ts file or pass it in as props, duh oh wait, you can't export anything other than async components inside of 'use server' files. Want to define a shared zod schema const in your action file? Not so fast buddy, create another validation.ts file and put it there. oh wait, you need to call useFormStatus inside your form? easy, just create a separate 'use client' component called loading button and call the hook there because hooks don't run on RSC dummy oh wait, you tried to import a helper utility file in your client component which imported another function from another file which touched a different file which imported 'server-only'? you should have known better. No worries, just make some new files probably and move shit around oh wait, you know hooks can be conditional - please go learn the rules of leaky abstractions, I mean rules of hooks oh wait, you can't import that third party library because they didn't add 'use client' to the top of any of their files so now you have to wrap them all with your own 'use client' components to use them I loved react SPA because I could easily move functions around. Next.js kind of destroys that if you ask me. btw checkout my 8 minute speed run tutorial about next.js 👇

I got a Next.js 15 application with next-auth. I add a logout button. It calls a logout action which invokes the signOut method from next-auth. This all works great locally. I deploy to AWS using SST open-next, logout button never actually deletes the session and just keeps redirecting the user back to the same page. I manually delete the session cookie and refresh my browser, and my session cookie gets added right back. We've reached a point of abstraction on top of abstraction which makes you waste days debugging on production. Is this what winning feels like?

even with the abundance of ai tooling, you still haven't shipped anything. It never was a skill issue; it was just your lack of drive.

you know it's going to be a bad coding day when vs-code can't even recognize div

let's just deploy a simple django app to aws. - deploy to ecs fargate, $0.60 a day - need a load balancer to use fargate, $0.54 a day - need a small database on rds, $0.47 a day - need a vpc for security, $0.36 a day - need a ec2 jump box to access the rds instance to run migrations, $0.28 a day - need cloudwatch metrics for health checks, $0.37 a day - need a nat gateway, $1.12 a day - secrets manager, $.03 a day - small charges for ecr, secrets manager, s3, dynamo $113.10 a month let's make Bezos rich baby

the world ain't ready...

“But ai can’t understand complex legacy code bases with hard to reason about business logics” Bro, you and your entire team can’t do that either, relax.

I’ve been on a project at work for 5+ years now, and I’d say some of the biggest technical pain points have included dynamodb, serverless, api gateway. Might be a skill issue, but if I did it all over I’d say just always use postgres and deploy containers to a managed service until there is a reason not to for most projects. Dynamo is great when you know exactly what you’re building from the start. It’s also good if from the start you know you’ll be dealing with a lot of data that can’t work well in postgres (guess what sql has been handling lots of data for a long time). Dynamo becomes a pain when you’re doing agile development. SQL is a lot more forgiving when requirements change. Dynamo takes forever to loop over all your entries and update them. Updating 10 million records takes almost an hour, and that’s including doing parallel scans. “Bro, just increase your provisioned WCU! Sure, but you know it takes at least 20-30 minutes for that to finish updating your instance”? Doing the same update using sql takes 5 min on a 2 cpu machine with 8gb memory. Your inability to easily query for data in dynamo is bad. “Bro, just use GSI!” Ok, now you’re cost for writes are doubled, and each gsi is async updated so again when you need to update all entries, it takes time for your GSI to update fully. Accidentally picked a bad partition, sort key? Have fun writing a bunch of code just to migrate your data to a new table. The dynamo docs say “know your access patterns before you make your single table pattern”… most product owners can’t even describe what they want, you expect us to design our access patterns correct from the start? Lambda is a great when you have a specific need to quickly scale from 0 to 1000s of isolated workers. For example, we have a use case where we need to loop over hundreds of data entries and generate unique pdfs for each one. Lambda shines with this, but now you’ll basically need to use sqs or a queue system to orchestrate it all. Btw sqs has its own set of gotchyas, such as events might be delivered twice so you better write your code to make sure you don’t double process the same event. Luckily lambda supports running containers now, but previously it was a huge pain when you installed a package that requires a node-gyp binary which means now you need to build that inside the correct docker image that is compatible with the lambda runtime and then create a lambda layer containing those binaries. Save yourself the hassle and just always use containers for all running code. Probably just stop using node or JavaScript on the backend if possible, it’s pretty awful. 100% don’t use a mono lambda for your api. Api gateway is a pain and is typically used for putting a rest api in front of your lambdas if you want to make an api using a mono lambda. Works great until your lambda takes more than 30 seconds; api gateway will time out your requests. That means you need to instead go async events and figure out another solution to notify your users (websockets, sse) when the request is done. Have fun getting either to work on lambda. You’ll end up using api gateway v2 websockets that has more gotchyas. Connections auto timeout after 15 minutes, so you need to add ping pong logic, max connection of 2 hours, so again have fun writing more logic for those limits. Cold starts are a real issue as your code grows; which makes you find ways to lazy import functions if deploying a mono lambda api. Don’t forget deploying your lambda has a 250mb limit which is the biggest pain in the ass. Again, just run containers on lambda if you must use them. Add on top you’ll end up using terraform or another IaC tool just to get all this stuff deployed. SST is great, but if you think about it, they created it because we all admit deploying stuff to aws is a nightmare, especially lambda. Idk I’m just burned out on this entire ecosystem. Just let me deploy a single go server that renders html at this point.

having too much pride in the code you write isn’t the flex you think it is; in fact, it makes you a liability. fixing your system’s tech debt becomes so much easier when everyone on your team agrees their code isn’t a direct reflection of their intelligence, but merely a reflection of the best solution they could think up at the time.

Admit it, you use react because you’re not skilled enough to write vanilla js

I feel bad for those who haven't tried agentic coding yet. I've once watched a developer spend 10 minutes trying to write reduce statement. claude code can modify 20 files to implement an entire feature spanning frontend, backend, db schema with one prompt in a few minutes...

It’s actually such a simple renaming with less characters that people annoyed by the change should do some introspection as to why it bothers them so much 🧐

The longer I play around with Go or HTMX, the more I want to just go back to Next.js 😵‍💫

but wait there's more

I hope the next person who says “skill issues” spends the next 3 months learning react 19 with next RSC just to get hired for their first dev job maintaining legacy angular js code 💪

T3 stack was the first time I tried something and felt the productivity gains within a few hours using it. It was an obvious choice over making a separate rest api in node and then some random SPA in react. It simplified enough for me that I switched to using it. Then nextjs app router came out and I started using it. Again, I felt more productivity gains with RSC and server actions. It basically removed my need for tRPC since I could get my end to end type safety directly in next. Sure I complained about some rough edges, but overall it was a net positive for my needs. I still had to use random things like sqs and lambda to do async processing, and find solutions to notify the ui when those jobs finished. Then I started using Convex for my backend and again felt productivity gains within a few hours or less. I no longer needed to think about where to store files, or where to kick off long running tasks, nor had to worry about hosting a database, or implementing update logic in my front end when async jobs finished. With the amount of complexity I’ve seen on my day job, I’m always eager to find a simpler way to build an application as a one man team. I know people will chime in with “try nuxt, try svelte, try sqlite, try X, Y, Z”. I’ve tried enough things where if I don’t see an immediate productivity gain within a few hours, I’m not interested. I’m not looking for different implementation details, I’m looking for innovation.

For the longest time I was looking for a better abstraction for creating applications, or at least a way to become more productive. I switched between Vue, Svelte, Nuxt, Next, Go, Tailwind, Chakra, Material UI, ShadCN, etc. Trust me, I've tried a lot if stuff. Just admit it, once you know how to code, it becomes tedious. Developers are solving the same problems over and over again; especially ones others have already solved. Unless you're going for your PhD or doing legit research at a university or company IR&D at the forefront of discovery, you're probably just a code plumber. Now I've seen building a perfect abstraction is usually not possible. We've been trying it forever, low code tools, no code tools, different frameworks, component libraries. We never got THAT big of a win with any of it. It's obvious to me now that AI is the next logical leap. It's abstracting the implementation detail entirely. I give it my vision or idea, and it gives me working results. The stuff the coding nerds love to debate about: semi colons, spaces, typing, languages, etc, it'll all be dissolved or abstracted away at some point, and I kind of love that idea. Need a feature that shows a card with an action dropdown menu? Write a prompt. Need it to be accessible? Write a prompt Need it to be responsive? Write a prompt Need to change it's look? Write a prompt. Does it matter how it's implemented? No, we just need an LLM to understand my prompt well enough to implement it for me, and modify it when it gets it wrong. I don't care how it's implemented, just make it work. The more I keep playing around with @cursor_ai and now @v0, the more the lightbulb in my empty head is turning on. I'm seeing posts of one man building feature rich applications in 3 weeks using cursor that has more features than many larger teams of highly paid developers ship in months. It's inspiring, but the bar is being lowered even more in this industry, in ever industry, which is scary but a long overdue progression for humanity. I will say, it does kind of suck feeling like the past 10 years of my career seems a bit like wasted effort... obviously my skillset can help me write better prompting compared to others and fix minor logical bugs as needed, but this approach to building is 100x better than me needing to invest all this time actually coding myself when something else can do it for me. No joke I'm adding features faster now with AI than I ever did before (on my side projects) coding with one hand, drinking a beer or two, and rubbing my wife's back with my left hand. 🤷‍♂️ maybe I'm just a doom scroller, or maybe I'm a visionary.

Y’all know I used to make Vue tutorials on my channel right? I also never used to like typescript. Here is the breakdown of my recent journey on my channel. 6 years ago in 2017 I started making vanilla js and html tutorial on my own channel and also made content on @freeCodeCamp yt channel. Did you know I have a Vue tutorial on that channel with over 500k views? I even have a tutorial on adonisjs with vue. But after having my first kid I basically took a full year break and made a couple new tutorials in 2018, but again took another break until the start of 2020 basically. At this point I was making random yt videos about node, JavaScript, etc. I was also doing Vue tutorials as well. I loved vue, I won’t lie. But then at some point in 2020 I decided to switch my content because I was moved to a project at work that was doing react with a node api and I figured it would be a better investment for myself to make react content as well at the time. I hit 3k subs in July 2020 and was excited about that! Then we learned wife went into heart failure around Sep 2020, so started doing more videos to get my mind distracted. At the end of 2020, I remember being on anxiety meds and made a bunch of css battle videos. Covid, wife in heart failure, we were renting a small ass house, I was recording from a closet. Shit was wild. At one point I bought a green screen and started doing videos because I was inspired by coding with cj (@coding_garden ). Y’all think Im just a stupid influencer, but coding and this channel got me through some hard shit. Off and on through 2021-2022 I was making videos. Nothing special, I wasn’t on camera most of the time. Just doing random JavaScript videos, react, simple html stuff. I got into web3 and nft at the start of 2022. Basically didn’t have good direction. The big turning point in my channel was July 2022 when I got a new M1 air and was able to crank out videos much faster because my old laptop sucked. I started making videos and JavaScript shorts daily. I also decided to buy one of those colored lights and Edison bulbs. That’s the real secret to yt success. I went a full 60 days where I published a short daily. Quantity over quality, baby. Also around July 2022 I started seeing @t3dotgg channel blow up. his evangelism of typescript and trpc made me see what true productivity looked like. I’m pretty sure a big reason for my growth is because I started also doing some t3 stack content and switched over to typescript. Love him or not, Theo is often a spearhead of this react community and he is often correct on a lot of his opinions. From July 2022 till today I basically publishes videos as much as possible. Sometimes daily for weeks on end, sometimes I take breaks. I started documenting my journey as I built my icon generator ai application at the start of 2023 Shared things I’ve learned along the way. I was still big into t3 stack at the time. I got into @SST_dev a bit at this time, great library. I started getting more sponsored videos, jun 2023 I learned about @convex and did a sponsored video for them. I was honestly so impressed that I started using convex on my real projects because it makes me feel more productive than I have using nextjs. I changed my webcam to match Theo and got more views. Imagine that. I hosted a hackathon for convex which was the most enjoyable thing I’ve done with my community. @eraserlabs asked for a sponsored video and I ended up loving their service and started copying Theo’s approach diagramming code. I did a lot of high level diagrams back in 2023. 2024 I started doing custom thumbnails. They actually got me a lot of views, did kubernetes stuff, learned about go, got ddosed, transitioned into self hosting. now I’m trying to get my starter kit finished and working on project planner ai, but I’m hoping to get back to more community projects, code refactoring, etc. That’s my journey, hope you join me for more.

Damn, y’all broke Theo with all the vercel and nextjs hate. Now we get to endure 90 second integrated videos that everyone skips over anyway. There is nothing wrong with companies sponsoring a channel, especially when the creator legit loves the tech he promotes. Now Theo is probably going to have to create paid $129 Next.js course and make $500k off his fans because y’all can’t stop hating on the idea of content creators getting paid for the hard work that goes into growing a channel and brand. I hope you haters are forced to participate in agile retrospectives the rest of your careers.

That’s how you do it bois.

I’ve been training hard these past years to finally code professionally using next.js. I get moved to a project using django.

If my posts are starting to feel a little unhinged, it's because this industry has made me unhinged: I'm frustrated from wasting hours watching ci/cd builds fail and just "run it from failed" because my runner decided to have an aneurysm and I can't reproduce the error. I'm tired of needing to learn yet another new thing that achieves the exact same thing as it did before but it has a readme with cool emojis this time. I'm angry because I forgot to set an env variable on my environment but I wasted hours trying to add console logs and redeployed multiple times just to see I forgot the variable. I'm confused why my google oauth keeps failing because I set my callback url to api/auth/callbacks/google instead of api/auth/callback/google. I'm pissed that docker desktop ate up all the disk space on my machine without me noticing and now I can't even delete files without loading up my terminal, googling where my docker.raw file lives on macos, and deleting it manually. I'm sad that my cloudformation deployment took over 10 minutes JUST TO END UP FAILING and now I wait another 10 minutes for it to rollback. Oh wait, the ROLLBACK_FAILED as well FML I'm no longer patient waiting for 10,000 node_modules to finish downloading just so I can show a single page with a form. I'm so close to re-installing LAMP and FileZilla so I can just SFTP my files to a GoDaddy and call it a day.

tailwind haters: "tailwind is so ugly" twitter: "hold my beer"

I'm tired of these ugly tailwind classes going off my monitor; I'm cook'n up the next big tailwind alternative

So much debate about programming languages when 95% of your web dev work will always be writing an api endpoint to talk with a database.

Technically isn’t server side rendering always gonna be more secure than any other approach, such as a rest API? it’s hard to accidentally leak data that users aren’t supposed to see if you never leak it when you render out your templates, right?

When working on side projects, there are two type of devs: The devs who spend time designing a system for large scale even though they have 0 users. The devs who just get something working and ship it knowing they can improve it later when it starts to fall apart. What type of dev are you?

I legit feel depressed knowing an LLM can churn out more code in 5 seconds than I could in 30 minutes. We are no longer limited by the constraints of our typing speed which is amazing, but also makes the past 10 years of trying to get efficient kind of pointless. If you can describe the problem in fine details at a high level, the only key you need to get good at hitting fast is tab. Even the prompt I used speech to text when possible.

I’m amazed how many developers are actually living in denial about the power AI is giving us. Before you’d tell a backend engineer to make a page responsive and their eyes would go crossed trying to figure it out. Now they can do it by typing one sentence. The scary part is that this is just the most basic example I recorded. Since then I’ve asked cursor and claude to do much more complex things and it modifies multiple files and I save the files and everything works perfectly fine when I look at the UI I have a whole live stream showing me adding buttons that show dialogues and confirm buttons when I click on a delete button which calls a backend endpoint which it implements all the back and logic, and it worked flawlessly. Stop living in denial. We have crossed a certain threshold at this point.

anyone who says they don't use Cursor (or at least something similar) while coding is probably a D-tier dev. I'm dead serious. I literally use cursor add features while I cook steak and edit videos. it's 12am and I'm too tired to think straight? prompt, auto approve, test, reprompt, auto approve, ship.

Did y'all know you can do this in typescript? it checks to make sure the string type passed in matches a certain pattern 🤯

I just wasted an hour... this is why a framework should never overwrite a global provided by the runtime. Since @nextjs overwrites node's global fetch, any third party library that depends on fetch also has built in caching, so unless that third party library has support for providing your own fetch override, you'll run into caching issues which you can't fix without forking the library. Why didn't they just provide a custom fetch import we should be using in our code like any sane framework writer would have done? github.com/replicate/replica…

Please, stop using starter kits and just use Laravel. If you need a starter kit you probably don’t want to be in this JavaScript ecosystem to begin with.

I love how people hate on Next.js but ignore that fact @codewithantonio is literally making complete clones of every single SaaS using it and making 12+ hour tutorial videos teaching it. Want people to use your "awesome" library, framework, or stack? show up. 🦗

If you’re ever feeling down on yourself, remember there is some developer out there who designed the concept of __init__.py and actually pushed it to main. The bar to greatness is low.

No joke, this is the third time I’ve used an app and I’m like wow this app feels really nice and then I looked at what is made with and it says vue and nuxt. I’m seeing a pattern.

For those using AI to code, be careful. You're buying into an illusion that you're moving fast when in reality it's shoveling code into your IDE that will eventually harden into the system like freshly poured concrete. Don't add code on top of a software system which is already poorly structured, lacking decoupling, using magic numbers, hard to maintain, undocumented, etc. My advice: YOU MUST UNDERSTAND WHAT YOU ACCEPT. After you TAB, TAB, TAB your way to a working solution, analyze every single line of code you just added and understand WHAT IT IS DOING and WHY IT IS DOING THAT. If you can't re-write all the code by yourself, then you must do one of two things: 1. Prompt AI again if you do not understand. Let it guide you to how it came up with it's solution. Ask it more questions so you can understand any new lingo. 2. If you still don't understand: git stash, plan it yourself, solve it yourself, test it yourself, and write it from scratch. 3. If you get a working solution by yourself, ask AI to analyze your solution for feedback: "Why is your solution better?", "Is this optimized code?", "Is this code secure? why?". I have promoted AI and I always will; it's an obvious productivity multiplier. But remember, I'm a mediocre engineer and I understand 95% of the code it's shoveling into my IDE because it's the exact same code I write: garbage.

Just let me invest your money in the stock market because my predictions are always on point.

Just think in Next.js. Just be patient, you'll eventually find the page you're looking for.

If you’ve been coding for a while professionally, please go make content. Go teach others things you’ve learned. Learning how to communicate more efficiently while helping thousands along the way is an obvious win win.

🌶️ hot take: Dependency injection is an overused pattern in this industry which ultimately introduces more complexity in your projects. “But you might need to switch implementation in the future!” Thank you for giving the example of premature optimization. Now you have hundreads of interfaces with only single implementations for all of them, great job! “But it helps with mocking!” Use a more flexible language that allows mocking imported modules dynamically during your tests. The one good use case is when you need to toggle functionality based on some application or request context. Use the pattern when you actually need it.

My wife @VictoriaSeiber6 is having surgery tomorrow. All prayers and good vibes are welcome! Love you babe!

We now have a nextjs starter kit! including... ▵ @nextjs app router 🌧️ @DrizzleORM 💰 @stripe subscriptions with webhooks 🔐 @nextauthjs 📀 @PostgreSQL 📈 Event Tracking 🗣️ User Feedback 🪵 Changelog 🗑️ Account Deletion ✋ Subscription Management 🏛️ Clean code structure Save time when starting your SaaS project by using our starter kit (btw it's integrated directly into our project management service to make life easier).

I just coded for two hours on a stream using neovim. How do y’all live like this? Coming from cursor, it’s like going back to the stone ages… Do y’all at least hook in copilot when using neovim?

wow, Laravel even has stripe subscription integration out of the box... y'all must just be bad at marketing if this is the first time I stumbled upon this

Honest 📖 I’m pushing ai and ai tooling because I’ve seen me and other devs throughout my career waste literal hours on simple bugs that LLM fixes in seconds. Sure, it’s not perfect yet and you still need to understand code pretty well to even use an LLM for coding, but don’t be stuck in the mindset of “I tried chat gpt 3.5 a year ago when it came out and it sucked”. Like seriously, allocate an hour or two every month, try another model, try Claude, try cursor during a brown bag at work, try throwing diagrams into an LLM to have it explain it all for you in baby steps, try using diagram tooling that generates your high level diagrams you can talk about at a meeting at work. You’re smart, I know. You’ve made it apparent with your angry replies. But you’re not that smart. Using an LLM can help fill in those gaps you obviously have. Just acknowledge it, lose your ego, and get good. Times are changing. I mean come on, even Dax uses gpt to write his IaC, I’ve seen him on stream do it, and he’s your aws savior.

Typescript, vscode, next dev server. At this rate we might need a supercomputer to render a static html page in a few years

“We need you to update the footer to say 2025 instead of 2024” No lie, that’s probably a 3 story pointer on someone’s team and will take 6 months to get to production.

Learn IaC, learn AWS, get hired It’s that simple.

please, I beg you, talk me down from what I'm about to prototype

fine, not impressed? here is another example for the people in the back

i'm no longer going to promote cursor. after I made a couple of those posts, i'm finding it's timing out more and getting stuck on generating results. please, just go back to peak engineering - typing 50wpm and hitting backspace every other time because you keep mistyping your left parathesis.

What burnout looks like

Just flex’n on y’all 🎉

I've earned $87,818 doing content creation this year. I remember back when my videos used to get 100 views and I would celebrate with my wife. I also remember multiple times talking about giving up because no one watched my stuff, but the drive to teach others how to code kept me motived. You would be surprised with how powerful a single comment from a stranger online saying "this video helped me landing my first job" or "you helped me finally understand recursion". The reward of teaching beats any full time job I've ever had. If you're a consumer of online content, remember to give back by posting comments with kind words and gratitude; it goes a long way. If you're trying to break into content creation, just remember this shit is hard and being consistent is the main skill you'll need to master. Cheers to a great 2023, let's hope 2024 is even better.

I would legit pay money to be able to watch 100 expert software engineers just code for a full hour with no audio just so I can see if their idea of "productivity" leaves me speechless or chuckle.

Someone mentioned having a blog can help SEO? 30 blog posts with 1,000 words in less than a minute, let's go.

Me holding onto Next.js after learning Laravel has everything you need out of the box.

Like I said, if your remote engineers don’t respond in 5 minutes your company is cooked.

you know what's actually impressive? developers who can work on the same project for more than 2 weeks before switching to another framework, library, or programming language. shiny object syndrome isn't a flex.

I remember back in 2009 when I was learning Java, I'd have 100% confidence that renaming files in netbeans or eclipse would also update it everywhere else in my project automatically. today, almost 15 years later using VSCode: "good luck" we have regressed in this industry.

cursor is pay to win now... this sucks

Instead of putting in a ton of extra work for a company who will fire you without warning, use that extra energy to build up your own ways to achieve self sustainability. Now, I’m not saying don’t do a great job where you work, but always be growing a safety net for yourself when you ultimately understand you’re just a name on an HR spreadsheet.

There is always pushback when I make content around agentic coding, but at this point I stopped caring. I'm going all with claude code and other tools, not because of hype (I'm getting dislikes on my videos and losing views actually), but because I truly believe it's the future.

Let’s go! MacbookPro M3 Pro. 18gb ram. I was tired of my M1 air overheating while recording and streaming which caused it to throttle. Hopefully this does better.

There... I learned Laravel, Livewire, and Volt. Happy now?

People who deploy their software to bare metal or self managed virtual machines: do you all hate life? why choose manually managing a machine over using one of many platforms that let you deploy and auto scale with a single click? It’s an ego thing for you isn’t it.

I won't lie, writing Go is a breathe of fresh air compared to typescript and the js ecosystem. Granted I haven't used it enough to start uncovering the cruft, but so far it's an obvious DX upgrade.

We used to memorize phone numbers, but then contact lists were invented. We used to memorize directions, but then GPS systems were invented. We used to be good at coding, but then LLMs were invented.

You work from home. I work from home. We are not the same. Yes I’m sitting in a massage chair.

here you go 🙌

deploying to a vps just like dev ops gods always intended

Now we are getting somewhere... same application, one using Go + Fiber and the other using Bun + Hono. Now time to let it cook for a few days and see what happens

I won't lie, deploying a single go executable inside a scratch container is a beautiful feeling... at work we need to bundle our express api using esbuild to get the code to fit onto a lambda which is also using a layer containing puppeteer because of arbitrary 250mb limits. We also had to turn off source maps because when we turn them on, the lambda uses 2gb of memory. I'm 100% over javascript and lambda at this point. 3/10 do not recommend.

You’ll eventually reach a point where you, as a solo engineer, can only accomplish so much in a day. When you reach your peak, you’ll understand why working with a team will take you 10x further than working by yourself.

I've used lambda and serverless to host my applications for a while now... but I'm having a change of heart. I like the idea of serverless, but often the tooling surrounding deploying your serverless applications is a giant time sink. Tools like Terrraform, CDK, SST, Serverless Framework help, BUT the moment you hit an edge case in either aws or with this tooling, you end up wasting days debugging shit, waiting on stuff to roll back, stuff not working on prod after deploying but it works locally, going cross eyed reading yaml files and cloudformation templates, sub par type safety with terraform, no great offline support which requires you wasting tons of time mocking and building your own local solutions, etc. I'm slowly starting to convince myself just running your api in a container is the most maintainable solution for all software projects and has always been the case since containerization became a thing. The only reason serverless feels good is because you're using a library that abstracts it all away; the moment that abstraction fails you, welcome to hell.

how long would it take you to code this? give your best estimate like your scrum team depended on it.

I just spent 5 minutes trying to figure out why my changes were not applying after saving a file... I was looking at my prod url and not localhost