- cross-posted to:
- programmerhumor@lemmy.ml
- cross-posted to:
- programmerhumor@lemmy.ml
You must log in or # to comment.
It’s like Moore’s law. The number of bytes for a basic app doubles every 2.5 years.
When I was young, we’d get a few different games games on a single 1.4 Mb floppy disk. The games were simpler, sure, but exactly the same games now would be far bigger in bytes.
Games is the one example that actually makes sense though. The game code size hasn’t really increased tremendously, but the uncompressed assets have only gotten more detailed and more numerous.
At least games make sense, as the graphics get better. Though in some cases, the compression is also better. Like PS5 games are smaller on average than their PS4 versions, even though they have higher resolution textures in most cases, just because the PS5 has better compression/decompression tech.
Compression is mostly done in software.
uh, bad news for you.
All programming is done in software my guy.
Great! Your point?
Better than that, the lack of reliance on spinning disks means that asset duplication and data read order is less of a requirement to reduce load times. It can still be argued that there’s just too many polygons, since simply scaling things back would be plenty effective in reducing storage usage and load times.
The other problem for bigger GB games is texture resolution. Games don’t always need 8K or 4K textures. 2K is good enough.
My shitty eyes can’t detect any difference past 720p
Is this the appropriate point to reference the suckless community? I mean, that’s THE point of the movement…
Lazy devs not removing old non functional commented code and background code additions ?
Though I do get it if they don’t want to remove the old code if their employer is an asshole
That’s not why. It’s the dependency trees that run a dozen layers deep and end up importing “isEven”. If you’re building a react app odds are good you’ll import way more code than you ever write yourself.
And no one should be leaving commented-out code in their app, that’s what source control is for.
Did my husband made this meme? Because he is constantly saying this 😂😂😂😂
isn’t it a combination of younger developers not learning to programme under the restrictions of limited memory and cpu speed, on top of employers demanding code as soon as possible rather than code that is elegant or resource efficient or even slightly planned out
Much the latter.
Plus everything better work perfecly out of the box on any hardware, and there is a lot of different hardware. Compatibility layers are often built into the package.
Java, for instance, recommenda that you package the whole (albeit slimmed down) JVM inside the package for the target platform, rather than relying on the java runtime installed already.
The users arent expected to know any of that anymore.
yep, a lot of apps are just repackaged chrome running a web page.
which begs the question to companies that require use of the app instead of just having a working website i can use on my copy of chrome/firefox that’s already on my phone…
why do you need hardware access to my device?
1 reason is that they want as much data as possible. They sell the user data. Or they use the user data to improve their targeted advertising. They want more ad clicks.
Re app versus site, many know how to block ads on browsers. With an app, the firm is hoping they can show you ads. There’s a way to remove ads from certain apps but the layperson doesn’t know.
Mostly the latter. We don’t do any optimizations on our product whatsoever. Most important thing is to say yes to all the customers and add every single feature they want. Every sprint is spent adding and adding and adding to the code as much as we can and as quickly as we can. Not a single second is allotted to any discussion about performance or efficiency. Maybe when something breaks, but otherwise we keep piling on more crap at full speed non-stop. I have repeatedly been told “the fast way is the right way” followed by laughter. I was told to “merge this now” on multiple occasions even when I knew that the code was shit, and told the team as much. I am expected to write code now and think about it later.
As you can expect, the codebase is a bloated nightmare. Slow as shit, bugs galore, ugly inconsistent UI, ENORMOUS memory use, waaaaaay too frequent DB access with a shit ton of duplicate requests that are each rather inefficient themselves. It is a rather complex piece of lab management software, but not so complex that it should be struggling to run on dedicated servers with 8 gigs of RAM. Yet it does.
Generally maybe but apps specifically, it’s the default choice of IDE, Android Studio, bundling tons of libraries for added functionality bound to Play Services by default.
Which would probably be illegal in EU now, if any judge had the tech see-through for it.
Tap for spoiler
Get electroned
I’d rather have the Rickroll, please 😐
AHHH, please trigger warning
The hp printer app says it needs your location to connect to WiFi. It says it needs your location all the time when not using the app, again to connect to WiFi
I think that’s to do with how permissions work.
Having wi-fi access can technically tell the app where you’re located so you need to give it location access
Which is stupid because it then also gets GPS access.
Can confirm.
Now, I just need to know why my calculator needs access to my contacts.
That’s because it gets lonely and needs someone to text.
It’s because one is the loneliest number.
You just reminded me of one particular calculator app which put “+” button behind a paywall
programming as a profession is full of incompetent failchildren, like all highly paid positions
Don’t sell yourself short, your low paid profession is also full of incompetent fail children
touched a nerve did i?
No
#include “the_entire_fucking_internet.h”
You know we’ve reached peak bloat and stupidity when JavaScript web apps have a compilation step, and I don’t mean JIT.
If the goal is to not have apps be too large, you probably don’t want to send the full variable and function names and all of the comments over the wire every time someone loads a webpage. That would be a very inefficient use of bandwidth, wouldn’t it?
Don’t we have compression built into http already?
I’d rather take a compile step than having no type safety in JS, even as a user.
Except… the compilation step doesn’t add type safety to JS.
As an aside, type safety hasn’t been something I truly miss in JS, despite how often it’s mentioned.
I think they are talking about typescript which is compiled into javascript
Ok, that could be true. I assumed they meant the “building” phase that some frameworks go through.
There’s lots of valid reasons for this.
Imo the biggest one people don’t account for is this: Dev salaries are incredibly high. if you want fast performance the most optimal way would be to target the platform and use low level native code, so C++ or Swift.
It would cost you like 20x more than just using electron and it will cost you bigly if you have multiple platforms to maintain.
So it turns out having 1 team crunching out an app on electron with hundreds of dependencies is cheaper, naturally that’s what most companies will do.
Don’t want to use electron ? Then it’s kind of the same issue except this time you’re using Java and C# and you have to handle platform specific things on your own (think audio libraries for example). It’s definitely doable but will be more costly than using a cross platform chromium app.
Technically there is no “most optimal”. Optimal is basically best.
Simple reason - dependencies.
Modern devs dump any dependency and sub-dependency under the sun into their project and don’t bother about optimizing it. That’s how you end up with absurdly large applications. Especially electron is a problem in this regard.
You can still write optimized and small software. However, for most businesses, it’s just not worth their time. Rather using an additional couple hundred megabytes of dependencies on the client system.
In terms of programming, absolutely some bloat there.
But I would wager a majority (or plurality) would actually be high(er) res media assets, embedded animations and video etc.
The only sector where that is applicable is games.
Tell that to my banking apps!
It’s the ads.
I’d wager it’s the multilevel dependency of countless prebuilt components when devs are only going to use a small fraction of their capabilities.
I don’t get paid to optimize, I get paid to implement features.
uh, please do ask, why does opening a fucking glorified text and image processing app require 1 gigabyte of ram.
Who wrote this software? The guy from the bible who was the model for greed and gluttony? Jesus christ.
I don’t remember those being particularly emphasised traits of his.
it’s going to be ret conned if you give it long enough, just like every part of the bible.
Stares at Mullvad…
Is their app big? fwiw on desktop, I just use their config with wireguard app, and that works quite well for me.
Usually, instead of having 8-bit art, you have epic songs and very high definition textures. That is a good deal of why.
I think the epic songs and 4K textures are missing in my MS Office.
Yeah but they made xlookup, that’s worth a few hundred megabytes
All hail xlookup
Textures and songs are not a thing on most apps right? For android, using Kotlin has created much bigger appsize than old java
Kotlin doesn’t have much impact on binary size.
For smaller apps maybe. I’ve seen apps that should take less than 1mb rise to 15mb or so
That’s not due to kotlin.
Why else?
Backed devs: sweats
Fucking Chrome/Electron is why.
I honestly wouldn’t mind that if they could all use the exact same runtime so the apps could be a few MB each, but nooooo.
See: Webview2
Unfortunately, it is extremely painful to work with😔 Enjoy rolling your own script versioning and update systems instead of using squirrel et al
Edit: I think Tauri works by targeting this and webkitgtk via their wrapper library, unfortunately I can’t get my coworkers to write rust
Isn’t that just the same pig, just wearing different makeup? I’m not a fan of msedgewebview2.exe allocating 500+ MB RAM just because Teams is open, but maybe that’s Teams fault…
Not quite. instead of a bundled pinned version of electron, it is an arbitrary version of edge’s WebKit fork shared across all programs using it. That means you don’t need to keep multiple copies of the webkit libraries loaded into memory.
That’s not to say that building things on web technology is an efficient use of resources. Even if multiple programs are sharing the webview2 library, they’re still dealing with the fundamental performance and memory problems caused by building an app in JavaScript.
As for why teams is so memory hungry? I would blame Teams.
Discord manages to make a half decent, highly responsive webview app, and that’s with the overhead of having its own separate instance of electron.
EDIT: the original poster was also talking about application binary size, not runtime memory consumption. Application binary size should actually be significantly helped by linking webview to instead of bundling electron.
Is there any alternatives to electron ? And why people’s doesn’t move on to alternatives if electron is huge & heavy resources ?
I mean, Object Pascal was doing the “write once, run anywhere” thing decades ago. Java, too. The former, especially, can make very small programs with big features.
Java (and Object Pascal, I’m assuming) have very old-looking UIs. Discord’s gonna have trouble attracting users if their client looks like a billing system from 2005. Also, what do you do about the web client? Implement the UI once in HTML/CSS/JS, and again in JForms?
So if you’re picking one UI to make cross-platform, and you need a web client, do you pick JForms and make it work on the web? or React and make it work on desktop?
I think maybe you’re confused. Java drives a significant percentage of Android apps. It absolutely can do modern UI. I can almost guarantee you’ve interacted with a Java program this year that you never considered.
Pascal is more niche, but it can do modern, too.
Java was doing web clients before the web could and still can. I don’t know much about Delphi’s web stuff, but I know they’ve targeted it for years now.
WASM and transpiling blur the lines, too. LVGL can provide beautiful interfaces on the web as well as platforms Electron could never target, and works with any language compatible with the C ABI.
I’m not saying these strategies are without their own warts, but there are other ways to deliver good experiences across platforms with a ~single codebase in a smaller payload. But mostly nobody bothers because they just reach for Electron. It’s this era’s “nobody ever got fired for picking Intel”.
We need more people working with and on alternatives, not just for efficiency but also for the health of the software ecosystem. Google’s browser hegemony is feasting. Complexity has become their moat, preventing a fork from being viable without significant resources. Mozilla is off in a corner consuming itself in desperation.
A US-based company holds a monopoly over the free web and a hell of a lot of our non-web software. So maybe let’s look for ways to avoid feeding the beast, yes? And we can get more efficient software in the process.
Yea, electron has flaws, but it’s basically the only way to make a truly cross platform native and web app. I would rather take a larger installed size and actually have apps that are available everywhere.
The sad truth is there aren’t enough developers to go around to make sleek native apps for every platform, so something that significantly frees dev time is a great real world solution for that.
The alternative is “just serve it as a regular website”. It doesn’t need to be an app to do its job. Name a functionality which only exists in electron but not in the standard browser API.
400mb iphone banking app entered the chat
