We Tried That, Didn't Work | Prime Reacts 

That's true it can be frustrated but there is no clear yes/no answer here. People with no experience have a tendency to think they know better then someone who has experience, but most often they are wrong and only sometimes they are right. But on the other side no one should just use one argument “We tried that, it didn’t work” it should be much more clear than that: what didn't work and why.

industry standard ≠ good look at javascript

I usually spend more time trying to come up with how I'm going to respond to potential "We've tried that" and "That's impossible" responses than on an original idea. And yeah... sometimes I'm missing a crucial hidden requirement that kills an idea but sometimes, it's just a lot of "Nope, nope, nope... [not my idea]"

@@amotriuc Experience cuts both ways. I have enough experience now (~6 years working professionally) that I understand the ability to spot potential problems ahead of time-things a new person likely isn't considering, etc. This is not the person I'm talking about. The person I'm talking about is someone who has 'experience,' but has also spent 15-20 years giving up as soon as they encountered resistance on a problem. Aim assist, capsule/camera separation, and more-very standard game features that are ignored because they tried and failed at it once. Experience can inform you and give you a good ability to anticipate problems, but if you're not engaging with it critically or thoughtfully, it can also keep you stuck in your ways and refusing to react to the landscape changing beneath you.

not a smart move at all since he can't prove his way is objectively the good way, the guy just have a big ego and want to be polarizing.

are trolling? Where did he say his way is the objectively good way. Or are you also in an “intellectual failure state”?

@@samuelodan2376 you have to choose either who use ts does it because has "intelectual failure state" or the article is only about his personal taste and he doesn't want to demonstrate that dropping TS is neither a good or a bad thing.

@@Lemmy4555 you sound you have a grudge from before this article, and that’s understandable. But to respond to this article in a way that suggests you have an issue with the piece itself is misleading, to say the least.

Believe 👏 all 👏 attempts 👏

Yeah in my opinion, that phrase is just apathetic indifference. “We tried.” Okay, but how *hard* did you try? Are we talking like a tad, a lot, did you try different strategies orthodox and unorthodox? It is a cop out to me. My first lizard brain response is to reply back “Well, TRY again!”

Follow up question to people saying this: but what if you tried it wrong?

​​@@twenty-fifth420you are a manager in the making Because alot of the times, when you are not in the team, you can keep screaming that, doesnt mean you understand how THAT implementation happened What the FUCK does "Try harder" mean? Try harder in what? Try harder in that thing that DIDNT FUCKING WORK????

nah, the actually follow up question is: "what went wrong?", because once they answer that, it becomes so much more obvious what needs to be done to get past it.

Care to explain like im a junior noob? I wish to learn more about jsdocs and zod.

@DontFollowZim I wonder if this would work /** @type {z.infer} */

​@@SRG-Learn-Code jsdocs add annotations to your code. You can write special comments that have a syntax that tell you, your fellow devs, and your IDE stuff like "this function returns a number". It's a way to add type information about your objects and whatnot without writing typescript. Imagine you have an http request handler in express, and you're working with `req.body` or whatever. req.body is the object uploaded by a user and could be anything. You could write a jsdoc comment that annotates the body with type information. It's basically like saying "this request body should be an object with a username and password properties, both of which are strings". It's helpful to describe your expectations for that object, especially since your IDE can autocomplete the properties it knows about while you're working with it. You can even have a build step that actually verifies you aren't doing anything risky. But how do you ACTUALLY know it's what you expect? Remember the user can technically send you anything at runtime. That's where zod can come in. You use zod to define a schema for something and then validate it matches that schema _at runtime_. That way you KNOW that you're working with exactly what you expect. The coolest part of zod is that you can just define the schema, and doing the zod validation automatically adds the type information you'd get with jsdoc or typescript.

this man uses zod

Does z.infer not work in (typescript powered) jsdoc?

Wow!! And also program in Rust btw (this comment is not sponsored by the Rust foundation™)

He thinks C++ sucks. I agree

I wonder if he gets a free Netflix subscription

I still don't know he works for Netflix

TRUE

Meh... every post anyone writes is triggered by something

I see where you’re going with this but my initial thought was html was the declarative dom api 😂 it’s just that diffing changes and injecting new/same elements is slow

Makes you wonder about his work hours doesn’t it?

​@@airman122469probably remote

@@airman122469 he does livestreams and breaks them up into separate videos. he has a schedule that he has agreed to with netflix and they're ok with it

Plot twist: Prime is the guy behind anonymous email from few weeks ago stating that he works hours per week.

this is from his livestream. and he doesn't edit his own videos so he spends his time wisely

I tried that, didn't work

@@DonAlonzo skill issue

This really weirds me out too, especially juxtaposed to c# dev environments. I've got a complex js full stack using hmr that shows my changes faster than visual studio can tell I clicked on something. I don't even have to debug a .csproj file to avoid invoking a complicated web of build dependencies just to run a simple unit test.

I even handrolled my own bundler back in the day with the google closure compiler and it wasn't what I would describe as difficult. I even had caching working with very little effort. Honestly web devs these days have it easy compared to the dark ages, and if your biggest pain point is bundling, then I feel like you don't really have significant pain points.

Sounds like the use of bigger teams is the problem to solve...

While this is the way to break past things that people are stuck with, it is also not particularly diplomatic, and grinding the wheels of those with seniority at a new company might not make you particularly popular, which in turn can make it harder to get changes through. The last part is especially bad, because you got it to the point of "we just didn't know about it", which is a clear signal that you removed most of their resistance to change, and beating them up after that is just pissing on their dignity.

@@sorcdk2880 I'm not looking to win any popularity contests, but then I'm a DevOps engineer so 🤷, I have had to fix a lot of things here and change people's outlook in my team and then from my team start to change the wider engineering's focus which has finally started to get through and filter up. The last part was just an Internet search away, which they would have had to do anyway. As it happens, we've now built two working systems, one a proof of concept in the legacy platform, which was tried before and failed, the other in a new green field project. My current resistance I'm trying to overcome is getting people on board with blue/green and designing it into our platforms so that we don't have to have maintenance windows and downtime. Again, I'm getting a lot of resistance but I'm going to try to get it into a new greenfield project that we're building as a proof of concept so that we have have a template we can point to and say look, this is how we should do everywhere. My boss gives me quite a lot of latitude, I present ideas, he appreciates them, schedules them in and we get things done and we're probably the one engineering department that's actually achieving a lot. We have a lot to do, but many of my changes are only now going to get deployed a year or so later.

Prime frequently jokes about various languages/frameworks and some don’t get that he’s joking. I’ve noticed that a lot of his live chat think he’s being serious.

"transforming the web into what its not" - pinterest is the first thing that comes to mind, but the same anti-design could be made with no-build js too

It feels like your brain is copying and pasting snippets of code at random times. Its also a good indicator on whether the language you are using is sorta garbage or not when it comes to syntax. Languages like C and Go have it really good where you can much more easier read it. Languages like Java, Rust and Typescript (not a language, just a documentation tool we decided was one, but whatever) just feels like you're reading chicken scratch.

difficult is the answer, try hard i must

@@ThePrimeTimeagen proceed you can

I share this sentiment with DHH in general, so many of his takes annoy me. His recent keynote puts most of it into context though and now I see that the vision of DHH is not all that different than what Prime and his audience (us) are interested in. Look for: Rails World 2023 Opening Keynote - David Heinemeier Hansson He start his talk around 10min and the better parts start around the 20min mark. The TLDR is that: - We have too much bloat - Server rendering is cool - Solutions like HTMX/Turbo/Unpolly mixed with interactivity solutions like Alpine/Stimuls/JS are great at giving us what React/Vue/Angular/Svelt give us with a fraction of the JS/TS DHH as a person still annoys me but many of his ideas have a solid foundation.

The TLDR seems to be "import maps exist now, so with a bunch of effort you don't need a build to get good performance" Yeah, but with the same effort you get even better performance.

Backend dev = being exposed to a fraction of what is required to deliver solutions to users.

@@-Jason-L of the 2 (main) fractions. Argument works both ways. And still does not make the JS dramas less enjoyable

Interesting perspective: I've seen the other side: a full ETL setup that takes over an hour for processing only gigabytes of data at most (and often megabytes!). It should have been entirely IO bound, and finished in seconds.

thats pretty fun

@@ThePrimeTimeagen The letters FUN are not used in spelling status meetings. Love your content, mostly on YT but sometimes on Twitch when the time is right. Even my wife laughs at the jokes. Mostly entertaining, frequently educational. Have you ever thought of applying for an internship at Neflix?

Well, yeah, check your bloat, but that doesn't mean compression and bundling somehow don't make it better. Why use an import map to optimize loading 20 carefully handcrafted JS files, when you could just load one JS file that's half the size? You need to build an import map and/or handle cache-busting anyway, so it's not like you're actually skipping a build step.

I don't think you understood. They want to write js and not have any compiler or build. Totally the opposite of Sveltkit which is a DSL and a compiler.

@@pxkqdYeah I got the first point, it's just, why give yourself so much trouble to write dynamic apps when you have such good DX with no cost.

How's that working for you? I'm skeptical but open minded. Every time I find a good premade solution that saves me from having to reinvent it myself, I feel a spurt of joy knowing how much laborious time I don't have to spend making a lesser solution. ex: I know how to debounce input for a no-build js design, but I use react and someone else made a hook that is simple and works very well. The client code I make to use it is elegant and minimal, plus it's very sensible within the react design I'm already using. For your case, I'm wondering what it's like for more complex things like retrying an api endpoint smoothly. (I like how tRPC solves this, and I'm glad to have it in my project)

@@travispulley5288 For the majority of my professional work, I haven't needed to do work that was so complex either upfront or became complex over time that warranted explicit build systems like webpack, migration to Typescript, etc. even if it was still attached to CI/CD or something because of corporate policy or whatever, but splitting into multi-tiered environments (minimum dev/prod, preferably dev/staging/prod) is a must regardless. A lot of it has to do with design docs and spec'ing the projects out ahead of time to get resources ready, not just for allocation and budget reasons but also to get in a mindset where if you think XYZ piece of project has a "small" function relative to the project then it should expect a "small" amount of work relative to the rest of the project and likely not _need_ something like an external dependency to work if someone on the team has the knowledge and skill to implement it. Most of the time it was less about having a "robust" solution that a third-party that theoretically had spent more time and effort making and putting through the ringer and more about having a "good enough" solution that worked for what we needed and gracefully handled errors if we missed a blind spot. As for stuff like API testing, I only recall having to be hardcore about that at $_MAJOR_US_BANK, and even then there was explicit test data in the local database to confirm formats and such and I usually used a browser extension version of one of the policy-approved apps for checking results combined with the browser's dev tools. Pretty sure it was a commercial program, so I never had a copy of it for my own personal use.

IDEs can parse those comments

The amount of hours of debugging that strongly typed languages ​​save you is impressive (typescript doesn't count, its types are hallucinations). a backend can't be performant and reliable without types

BTW this strategy will lose compared to a well-cached bundle over enough time. Keep in mind Lighthouse always measures a fresh load which gives a bit of an unfair advantage to inlining (assuming you want users to look at more than one page on your site).

STOP

@@ThePrimeTimeagen WHY?

He is just someone who is very opinionated and let's the world know it

​@@LiveErrorsJust like most programmers

@@LiveErrors Nothing wrong with either of those things. Is that really it?

He's very opinionated, and backs it up like every single essay I wrote in high school. It comes across as arrogance and a lack of self-reflection, even though he is sometimes right and has objectively created a productive framework and multiple successful products.

It's really not hard to look this up. If you are out of the loop, start with "Basecamp diversity" and follow the red thread from there. Don't just read his own articles, though. The anti-type debacle and now this "no build" discussions is peanuts compared to what he did against Rails core maintainers and at Basecamp.