The Changelog: Software Development, Open Source - Why your framework doesn't matter (News)

Episode Date: May 6, 2024

Bahaa Zidan says your web framework doesn't matter, DHH writes about magic machines, Dylan Huang reviews thousands of opinions on HTMX, Tim Ottinger says programming is thinking & Tim Spann says small language models (SLM) for the win.

What up, nerds? I'm Jared, and this is Changelog News for the week of Monday, May 6th, 2024. Mere days before Google proudly announced that over 400 million accounts have used Passkeys, William Brown, who's the developer behind WebAuthnRS, wrote, Passkeys, A Shattered Dream, in which he describes how corporate greed from Apple and Google destroyed our Passkeys-based future. How's that for some 2024-style tech dystopia juxtaposition? Okay, let's get into the news. Why your framework doesn't matter. Baha Zaidan takes us on a brief history of web development.

Do you remember websites? Then reminds us why the framework we choose doesn't really matter. Quote, want to keep using React? Want to switch to something better like Svelte? Want to avoid JavaScript like the plague and use HTMX? It doesn't matter to the end user. As long as you're providing value to people and or having fun doing it, you're good. Don't feel bad about your technical choices because someone on the internet wants you to. End quote.

I wholeheartedly agree with the sentiment, but would temper Baja's premise just a bit. It's not that your framework of choice doesn't matter. It's that it doesn't matter. It's that it doesn't matter nearly as much as you've likely been led to believe. Magic Machines. DHH writes about a phenomenon that I've both noticed and inhabited in my software career. Quote, an interesting psychological phenomenon where programmers tend to ascribe more trust to computers run by anyone but themselves. Perhaps it's a corollary to imposter syndrome, which leads programmers to believe that if a computer is operated by AWS or SAS or literally anyone else,

it must be more secure, better managed, less buggy, and ultimately purer. End quote. The logic I've employed when coming to this conclusion goes something like this. They have more X resources, more X knowledge, and arguably more to lose than I do if X fails. So they've got to be better at managing X than me, where X is a variable, not a platform, formerly known as Twitter. But David's not buying. What I'm selling, quote, there's no magic class of computers and no magic class of computing clerics.

It works on my computer is just the midwit version of it works on that computer. It's all just computers. You can figure them out. You can make them dance, end quote. Are you buying what DHH is selling? Reviewing thousands of opinions on HTMX. Speaking of frameworks, here's Dylan Huang on the newish hotness. Quote, HTMX has brought an

absolute whirlwind of controversy with its radically different approach to building user interfaces. Some folks are skeptical, others are excited, and others are just curious. To analyze how developers truly feel about HTMX, I went to where developers live, Reddit, Twitter, Hacker News, and YouTube. I parsed thousands of discussions and synthesized my findings in this article, striving to present only thought-provoking opinions. End quote. What resulted was a radically diverse set of opinions ranging everywhere from, HTMX is just hype, to, HTMX makes you productive. You can click through and read some of the spiciest opinions if you're interested, but here's Dylan's big takeaway, which is quite a bit less spicy, maybe even a bit bland. Quote,

competition is good, HTMX is thought-provoking. But I think it's great because it forces developers to entertain new and novel ideas. Developers can often be wary and cautious of new technologies. Since it might not be solving a personal problem they are already facing. But for the developers that are resonating with HTMX, there is an enthusiastic group of developers who are starting to use it in production. It's now time for Sponsored News. Three questions to ask of any DevOps tool in 2024.

Fire Hydrant CEO and recent Changelog guest Robert Ross says, Is your DevOps tool stack out of control? I feel like every day I talk to someone who feels this pain. The technological golden age of the past few years created a lot of niche tools, but now that CFOs and boards alike are demanding budget restraint, many of these tools are being scrutinized. End quote. If you're one of the many tech workers currently being asked to evaluate your tools, Robert gives you three questions to ask. One, will this tool help me reclaim time through automation?

Two, will this tool help me ensure accuracy? And three, does this tool help me identify efficiency and value? Read the entire article. It's linked up in the newsletter where he fleshes out these questions with data, examples, and the why behind each. And thanks to FireHydrant for sponsoring this week's ChangeLog News. Programming is thinking. Tim Ottinger in 2014 made the assertion that programming is 11 12ths thinking. Then he goes on to show why that is, at least approximately, the case. Then, based on that fact, he makes this provocative conclusion. Quote, if programming is one twelfth motion and eleven

twelfths thinking, then we shouldn't push people to be typing eleven twelfths of the time. We should instead provide the materials, environment, and process necessary to ensure that the thinking we do is of high quality. Doing otherwise is optimizing the system for the wrong effect. What if we changed tactics and intentionally built systems for thinking together about software and making decisions easier to make? I think that productivity lies in this direction. Small language models for the win? Tim Spann shares a sentiment which I very much want to be true because it reduces our dependency on large cap model providers, but I honestly don't know if it's true or not. Quote, I don't need a model that knows a little bit about a lot of things up to last year. I need a model that knows everything about Apache or Python programming, or how Bitcoin

works. Not only are these trained on just the problem space, but they can run faster and on smaller hardware. We can usually run on smaller, cheaper machines with simple or no GPUs with just CPU. End quote. Maybe this large versus small model debate won't matter in the long run, as all models become open source commodities. Or, maybe the move will be to take an off-the-shelf open source large model and rag it, and or fine-tune it, to specific problem spaces. I don't know, but I like the idea of small. How do you think this will play out? Let me know in the comments. That's the news for now.

But also scan the companion newsletter for more stories, like the Pareto Principle applied to Mario Kart 8, woodworking as an escape from the absurdity of software, and the heat death of the internet. If you're not signed up for the newsletter, fix that bug at changelog.com slash news. We have some great episodes coming up this week. On Wednesday, we talked to Paul Orlando about his new book, Why Now? How Good Timing Makes Great Products.

And on Friday, we're joined by Annie Sexton, our old friend from Get Your Reset On. I'll leave you with this quote from Shopify's Toby Luecke. Not all fast software is world class, but all world class software is fast. Performance is the killer feature. Have a great week. Share changelog news with your friends who might dig it.

And I'll talk to you again real soon.