Hacker Newsnew | past | comments | ask | show | jobs | submit | accurrent's commentslogin

A huge pain of desktop development is managing dependencies particularly if you are coming from C++/Native code. Imagine you wanted to make a simple app in the 2010s that accesses the webcam for instance: If you use chrome you got the full html5 runtime, whereas with other tools you'd need bespoke libraries which your end user may not have.

The only way to ship the same code between linux, macOS, and Windows would have been to either use something like Qt or the JVM back in the 2010s. The JVM desktop stuff was super ugly and Java itself had become notorious (also if you wanted anything remotely interesting like webcam/microphone or multimedia access you'd still have to ship dynamic libraries). Qt required special packaging skills to ship to the end user especially if third party libraries were involved. Electron killed that by bundling chrome (and chrome has a huge api surface).

Today, I would argue the package management situation for a language like rust is a lot different. You dont have to cobble different DLLs together, you can just statically compile dependencies and cargo/rust do a pretty good job of making sure it's bundled with your app. Other languages also now have similarly useful tools.


But then the C++ libraries you mentioned also had the ability to statically link into a binary, which means you could statically link the webcam access in, in your example. So I don't see the advantage you're arguing for?

It comes down to package distribution.

Chrome had already solved the hard problem of "how do we support webcams on linux, windows, and mac" and you as the dev just needed to use the Javascript webcam api. ( https://developer.mozilla.org/en-US/docs/Web/API/MediaDevice... )

In C++ land, you are ultimately going to either need to cook up your own webcam library or look for a 3rd party (which, in 2010, likely didn't exist) to provide it.

Today, there are a few easy to discover rust crates which provide cross platform webcam apis. But further, rust does cross platform better than C/C++ IMO. It's really easy to change how code is compiled based on the target. That was all well thought out because rust was supposed to be a language for browsers.


A cross platform webcam library isn't that hard to build at C++. I mean, it's tedious, and having to support multiple different operating systems makes it harder, so it certainly is going to cost a small pile of money, but this is not hard. The pile of money needed won't be all that large. Mind you, I prefer to use something that's off the shelf rather than writing my own, but it would not be hard to write my own.

The hard part of all webcam libraries is all the different webcams running all their different proprietary protocols.


Write the code or build the code? That's always the problem with C++ libraries.

It's not that difficult to write cross platform C++ code. Consuming the dependency is always painful. What build system does it use? What compilers does it support? How do you get the code? There is 0 standardization for these things.

C++ code isn't always that portable either. Old code in a new compiler is very likely to break on warnings, especially if the old code made the high quality decision to hard code -Werror in their build system.

From personal experience Rust as a package and ecosystem is far more portable than C++.


Qt is easily obtainable via vcpkg or conan.

Was that true in 2012?

Depends, it was easily obtainable via OS package managers, and installers in 2016.

However that has been true since 2016, and here we are, 10 years later.


For Linux, but not for Windows and Mac

I mentioned "and installers" for a reason, I know Qt since KDE exists.

Very yes.

Neither existed in 2012, so of course not. On the other hand, QT has always been really easy to download from QT. I don't remember who owned QT in 2012. They've changed hands a couple times over the years, but in 2012 I was downloading QT from whoever on them. QT was large, so you needed a decent sized computer to build it, but in 2012 this was not a big deal.

People get too hung up on package managers. When you're building your own proprietary you don't really need a package manager, you just need a central place to download the tarballs from. Or you could do a monorepo and just build everything locally - in 2012, that means you start your build before you leave Friday afternoon, and when you come back Monday, everything's built.


It is 2026 and regardless it has been quite a while that cmake fetch content had been usable.

The only reason why a cross platform well supported C++ package/module isn't available is because nobody has bothered to build it.

If it really is something that bugs you then be the one to do it, throw your clanker of choice at it and integrate it into a project you run in production.

"Chromium ahips with it" isn't a valid excuse these days, that is likely an hour or two away with a modern LLM, building and packaging electron is more work.


Dynamic vs static linking is mostly a red herring for desktop GUI application development unless you want to integrate into something like Linux distro package managers, as dynamic libraries on Mac and Windows can be self-contained within the application bundle/directory with no disadvantage beyond marginally longer startup time, and you can do similar things with Flatpak and friends on Linux.

The real difference is that Rust provides a package repository with a unified build system that's as simple to use as NPM or NuGet.

vcpkg which is similar exists today, but it didn't in 2015 when Discord started, still isn't anywhere near as comprehensive as package repositories for languages where package management is part of the standard toolchain, and the underlying build systems are far messier under the hood.

It can also be helpful in development to use libraries built from source as part of the application's build process, at least as an easily-accessible option, which wasn't common practice in most C/C++ ecosystems historically where different libraries used different build systems and if you wanted to integrate them into something like a Visual Studio solution for easy reference to source code during development and debugging you were often on your own.


you can do that easily on linux too

IIRC Qt is LQPL, so if you linked it statically you'd have to license your code as LGPL too. Shipping Qt as a separate dll didn't trigger that issue.

Java desktop was only ugly if you chose for it to be. It was pretty minimal effort to get native look and feel. Nothing built in electron looks native. If we're accepting custom interface using electron, that was doable with Java too.

JetBrains IDEs arguably look/behave better than VSCode as far as a native app goes. I wouldn't say either look great on MacOS but a truly "native feeling" Mac app is usually quite different than a Linux/Windows equivalent.

Long time ago, Apple had a proper OS X Swing L&F, when you could do Cocoa in Java, and there was still some uncertainty if Objective-C would be embraced by Mac OS development community.

That look and feel still exists last time I checked. There is also a similar one on windows.

I love JetBrains IDEs and have been using them for a decade, but I’d say their general behaviour as user interface is among their worst qualities. Chief of which being how any sort of dialogue or additional window is fully modal. Opening any of the settings windows for example completely locks you out of interacting with the main window.

I find them janky as hell, but their ability to wrangle code is top class.


I much prefer vscode to intellij

Why is it no one ever seems to do it then? I have never seen a Java app look or act native on Windows.

A Swing application looks ugly and not native by default, but the platform allows to plug a different look'n'feel. With small effort, you can make it look nice with something like flatlaf [1]. Pretty sure there are others, I haven't done Swing UI in a while.

[1] https://www.formdev.com/flatlaf/


It takes one line of code to get native look and feel in swing. Not sure why so many didn't do it.

Eclipse

I haven't used Eclipse in a decade, so maybe it's changed, but back then it was very much a Java UI, not a native feeling one.

I remember the Java UI, but at some point they switched to a version of Swing that made it look more native, at least on Linux.

Eclipse uses SWT which uses real native controls.

Of course Linux doesn't have native controls - I think it uses GTK on Linux.


It used SWT, which was a thin wrapper over native controls.

Ahh yes that's right.

Eclipsed used SWT which produced actual native controls. Swing controls are not native but look and feel native.

They had native looking skins for Swing that still felt slightly off and behaved differently. Netbeans IDE looked nice, but it was weird in subtle ways.

Discord ships plenty of native code, though. Actually most popular Electron apps do

>Qt required special packaging skills to ship to the end user especially if third party libraries were involved

How outrageous to demand that a software developer possesses the 'special skills' to ship a Qt application. It's not surprising that people think AI is killing programmers when this is the level the average 'programmer' has been operating at for 15+ years.

Imagine going to a chef and he tells you he can only warm up microwave lasagna because cooking the steak correctly requires special skills. The special skills are your job


Probably an economical thing. If you have a much higher effort with packaging and testing vs coding the app logic.

If developers don't want to be replaced by AI they need to bring something to the table because employment is not a charity.

Usually it is hard for those coming from scripting languages.

Strangely enough we managed to write portable code across platforms during the last century, and there were more than only three major desktop OSes.


> Strangely enough we managed to write portable code across platforms during the last century,

Yep, that argument always seems strange too me. Do people fear that much to design their own interface and hook to each platform ecosystem? You look at something like the BSD which runs brought one single kernel concept and various subsystems across a variety of hardware. And somethine like libao[0], which provides a common interface to various sound system.

Whatever the app, if you care enough, you can abstract the core features away from the GUI and other various dependencies.

[0]: https://xiph.org/ao/


I never really understand this attitude. It isn't really hard it took one or two evenings to understand how native libraries work and you close this problem forever.

Then why not static link library ? And please don't mention about the app size because you are already fine with shipping the entire browser with your app.

> The only way to ship the same code between linux, macOS, and Windows would have been to either use something like Qt or the JVM back in the 2010s.

Correction: you didn't know how to do it. That doesn't mean nobody could.

Hence, electron came on the scene and enabled people without that particular experience or skills to create desktop apps.

Grumpier people like me rolled their eyes and said people should just write desktop apps similar to how it was done in the 1990s, to which the answer was generally "that's too much work". If this is changing, I would guess it's because AI is now doing that "too much work" for them.


I think theres no need to get personal here. Ive been dealing with Debian packaging of C++ apps for at least 10 years and my $day job is/has been C++ (I personally never have written an electron app). It's not untrue that people with the skill to correctly bundle C++ code were and still are far and few between. What Im saying id I can see the appeal for a large number of programmers.

IMO senior software engineer is someone who drives consensus in a large organization. I dont think LLMs are going to solve that problem any time soon.

I also see a bunch of skills emerging. First if you're doing frontend type dev, there is a lot of value in good aesthetics and UX intuition. You can build hundreds of prototypes and test concepts much better.

On the backend, while LLMs can trace race conditions, the need to reason against spaghetti and figure out what your context window looks like is still very much there. Designing tests that ensure seamless migrations still requires a degree of human oversight.

Then there are frontiers like robotics where even with LLMs coding, there are so many cross functional problems that you dont feel like you are moving faster.

I work in a fairly niche role. I use llms for coding all day but there are times for the type of operations research work I do I simply find it easier to write algorithms in almost correct code.


> someone who drives consensus in a large organization. I don't think LLMs are going to solve that problem any time soon.

This statement resonated with me. Thank you.


A lot of things. The British empire is a master class in financial engineering (largely by luck imo). They started by raising capital via IPOs to create trading companies. The first successful ones were the ones that established the 13 colonies that became the USA and Jamaica.

While they maintained control over those, they engaged in war and looting against the established Spanish empire. They managed to entice the Indian elite by being extremely effective at paying back loans using the loot they got during the Spanish era. Internal division among elite in India meant that the bankers in bengal thought they could use the British for deposing a king they did not like and at the same time handle the raids that the Marathas were performing. The British essentially used the money to bribe the king's army and took control of Bengal (which was the ricebasket for one of the largest empires of the time and the largest cloth producer). Unfortunately for said bankers, once the British took control of bengal they betrayed the bankers who lent them money. They raised an army of local sepoys and went on to take the rest of the subcontinent (although this was a slow painful process).

Even as the US revolted and gained independence, the British found themselves with a growing colony in India where civil wars and divisions between kings made it easy for the British to move in. One thing to remember is that when we talk of a British army at this point it really is an army where the lower ranks are made of Indians who were trying to make a living (as it turns out British were just better paymasters than their local counterparts).

Once they conquered India, they moved east and use Indian soldiers to take on Chinese while waging the opium war (and making a tonne of money from the trade of drugs).

Ive glossed over a bunch of things but honestly the success of the British in the 1800s is built over many many years of multiple investments some of which started in the 1500s (for instance their acquisition of Mumbai was entirely by sheer dumb luck as it was part of a dowry given by the Portuguese, who in turn received it as part of a peace treaty with a local ruler)... Many of the things that brought them success was a result of fairly immoral behaviour. And while their governing class in Europe is fairly small they had massive local administrations run by locals reporting to them.


It's still important to support research in this direction cause us meatbags cost a LOT less energy to train than GPTs even if you assume that it takes 30 years to train a PhD. Our current training methods honestly leave a lot to be desired. We almost certainly dont do full backpropagation.


I’m not sure that’s true. We have literal billions of years of training.


Human "pre-training" is vastly underappreciated.

Hell, animals are an even better example. Many animals pop out of the womb and start walking and eating and acting just like an adult!


That is only surprising because for simplicity age is counted from birth not from conception.


One could argue that humans are more adaptable to novel environments


If you wanna get philosophical then the decades of research that humans did is also AI "pre-training". And all of that is built on knowledge that humans have spent our whole existence amassing so maybe all of human "pre-training" is also AI pre-training.

A fun but silly exercise.


You need to factor in the cost of training all those PhDs that never end up producing much of interest though you don't get to pick the best afterwards and claim all it took was to train him/her. Plus, once trained, it's productive only for 6 subjective hours per day (including weekends and holidays). Might have costed gazillions to train a frontier LLM but it works for millions of hours/day.


I'm sure we don't literally do backpropagation, but differential equations that settle toward stationary states show up a lot in biology


I think it's fun back-of-the-napkin sometimes to compare meat to matmuls but ultimately fallacious to its core, making it an intellectual tarpit.

It's much harder to argue with the math and empirical results.


Energy cost may be a pointless comparison since we don't eat electricity, but it's very reasonable to look at sample efficiency and ask what we have there that our theories miss.


PHD inference doesn't scale, though.


One thing that concerns me from all this is "understanding" is very important to human progress. The fact it took 400 years to crack Fermat's theorem resulted in a lot of "Side Quests". These side quests helped grow other fields (for instance elliptical cryptography). Im concerned with AI that we will loose these side quests.


True.

Having to ask the village elder about how to do things meant learning various other incidental lessons. It was reduced when books started to become available.

Having to search in the library also often lead to serendipitously seeing a book on the next shelf and falling in love with a topic you didn't even know about otherwise. Or you had a chat with a librarian asking for advice which books to look in for a topic. Google search eliminated that. With Google search and reading websites, you still had to read or skim the page and may see some other info or click to read the author's About page. Now with AI we get straight to the answer.


Don't worry, in its final reply the agent will always present one or two things "worth flagging" to keep you engaged


I do neither maths nor science with AI but in my experience most models are perfectly willing to burn tokens on a ton of sidequests at the earliest opportunity.


Yeah but do you read through and find if one of those side quests is useful?


They're often directly applicable: minor bugs, inconsistencies, missing tests, commonly also stuff I already now (like some infra config details).


Minor bug and develop whole new field of cryptography are very different scales. I do agree alms are really good with a lot of common bug fix related tasks to the point you have to split commits out. The fact is you recognise it's a minor bug. Who is sitting through the proof of Navier Stokes and going through it and finding connections between itself and other fields? Im not saying LLMs are bad, but I do think understanding is important. Heck, the fact you identified the minor bug suggests you understand the output. Im not so sure the same can be said of a gajillion line lean dump.


Yes, why not?


We'll get innovations in AI as a side effect.


Its complicated... You can't function without feedback. Imagine you have a pothole, a 0 feedback society would tell you that the contractor the government selected is perfect and there are no pot holes in ba sing se. If you actually go to china you would find that roads are genuinely pothole-free. Obviously there are feedback channels. You can't run a country with zero feedback. China knows this and they monitor a lot of things. There are channels (like they have online petition platforms, 12345 hotline - organised demonstrations are not allowed). Theres also a whole industry around gathering metrics from citizen surveys and online studies. Every government program is measured carefully and methodically. Is it democratic in the western sense? no. Will you get killed for reporting a pothole or flaw? Absolutely not.


Government ban is part of the reason, but cultural differences are also important. There are other countries where Google is not the defacto search engine: South Korea, Japan, Russia, etc.

Similarly if your were to look into food delivery companies/ridehailing you would also find many countries where Uber has completely failed.


Google results have gotten so bad, I have to wonder if it is still the defacto search engine anywhere or whether the numbers are simply made up.


At least speaking for myself I haven't interacted with traditional google results for a while because their AI is similarly free on the front page and does a remarkably good job of filtering and summarizing the search results I would otherwise have had to spend 30 minutes trawling through.


This. The circus that is academic publishing is just not worth the effort. Ultimately, companies do have to make money.


I really like the diagrams! This is a great way to show what happens deep down. A lot of engineers I know have very poor understanding of how software actually works. However a couple of things: - Theres a lot of text and not all of it is valuable - Hijacking the browsers "back" button is kind of annoying.


Hi, thanks for your feedback. Regarding the back button issue, that was an oversight from my part. I'll fix that.

I am glad you liked the diagrams and simulations. I understand that it would have been better if the article was much shorter. I actually didn't want to leave any gaps in understanding or leave any important historical context. But i'll try to improve it.


Awesome article


The intelligence is only one part of the story. People need to actually go out there and do experiments. Science is not only theory, but also experimental. The best science happens when experiments show that a previously held assumption was not true. Eg the Michaelson Morley experiment where the assumption of ether was disproved. While AI is an incredibly powerful tool, it does not replace the act of observation. Thus Im sure we will still have scientists in the future and some degree of open science. There are experiments out there that by nature of their complexity need massive public coordination (CERN for instance) which in turn benefits from openness.

What is going to suck though is the ladder for juniors. We dont start out by working on big ticket problems, but usually early career researchers solve really tiny problems in a cheap way. The lowest bar for a cash strapped PhD student would be to contribute to some new theory in some way even if the student doesnt have access to equipment.


> People need to actually go out there and do experiments. Science is not only theory, but also experimental.

For biosciences and physics, sure. For mathematics? I am skeptical that your assertion applies.


Ah I see your point, I took the word science to mean bio, physics, chemistry, etc. IMO mathematics is a different discipline and not one Id consider a science, rather I see it as the language with which we express science.

If AI is somehow able to prove everything wouldn't it bypass Godel's incompleteness theorems?


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: