I really wish people gave a damn about the “gui over the network” problem x11 solves. Wayland drops this use case entirely so we’re pretty much universally stuck with vnc. Microsoft rdp is a great solution for this in windows land.
They drop this use-case, but still use sockets for IPC, so I still have to pretend I'm doing network programming, serialize my messages over the "network" and "flush the stream" (insanity) but don't actually get any of the benefits of this model.
I genuinely wonder if they stopped to think why X11 has sockets or just blindly copied it over. Or are they unaware other forms of IPC exist, that don't require you to go through the kernel 13 times to send a byte to the other process?
What would you rather they use to communicate with the server? Message passing via shared memory?
UNIX sockets are perfectly fine for IPC with small amounts of data, and is how everything in UNIX has always done it, network transparency or not. They provide a simple, efficient and reliable communication channel between two processes.
Wayland uses UNIX socket for message passing, but then offload most of the work to shared memory when the real work begins (GPU rendering). I just wanted to add some nuances that it's not as black and white as this comment made it seem.
How about making the standard client library's API the interface, and have it hide whatever the system is actually using?
A long time ago when I looked at designing a X11 replacement, that was my approach. AFAIK, only special X utilities used anything but Xlib anyway.
And later I think this is what early revisions of Canonical's Mir did.
If I had to guess, because then at least -hypothetically- easier to optimize for the 'local' case (depending on implementation, at least from my understanding of the POSIX/*nix paradigm, bending and breaking a bunch of rules possibly) by dropping in a different implementation.
Optimise how? There already is only the local case in Wayland, pixel data is shared using shared memory which only works locally. Only small communication messages are sent via the socket. And the Wayland protocol also uses native endianness because, again, it only cares about the local case. It even sends file descriptors over the socket.
So what would you do differently in an alternative client library?
> So what would you do differently in an alternative client library?
I should have better disclaimed my comment.... to be clear I don't know much about the graphics subject, I probably should have prefaced it with,
"I don't know anything about Wayland but as someone totally naieve on the subject but assuming someone else's assumption".
At least to me, even if it breaks the X11 model (Which is a shame, that was fun to play with back in the day) if they're doing it the way they are I'm guessing Chesterton's fence will come into play at one point or another.
Yes, command buffers over shared memory are the correct way to do this.
1. You don't need to convert your discrete messages into a stream with size metadata, only for them to immediately be converted to a message on the other side.
2. You don't need to jump into the kernel to copy over 20 bytes, only for the other side to jump into the kernel to copy it back.
3. You don't need to deal with the "oh but what if my read returns half a message because this is a stream"
4. You don't need to pretend you're doing network programming.
Regardless, it's not that big of a deal - this is like my 73th biggest gripe with Wayland, I only mentioned it since GP was talking about network transparency.
It's pretty representative of the project though - "We're doing things the way we've always done them, but slightly different. Now rewrite all your software to work with our thing. No, you cannot do global keyboard shortcuts or set window position. You don't like it? We're doing this for free, you cannot critique it."
You don’t have to hit the kernel for 20 bytes. Buffer up all your commands and send them to the kernel with a single write(). The other side can then read them all (or however many fit in its receive buffer) with a single read(). The only real difference is that the memcpy happens in the kernel instead of the receiver and that the kernel provides a useful blocking mechanism by default so you don’t have to manage that in userspace code.
You need some kind of serialisation either way. It can be as simple as “this message has the shape of this C struct”, but that’s the case whether you’re talking shared memory command buffers or sending data over a socket (and there are good arguments for and against in both cases).
You’re right that you don’t need to deal with “oh I received half a message” when using shared memory command buffers, but that’s more a code complexity thing someone solves once in wayland-client and then nobody has to really think about it again. It’s not really a performance concern (because hopefully the rx buffer is large enough for it to happen rarely) or application code complexity concern.
Sure. But imagine some piece of exotic hardware, e.g. computer mouse, that reports its movement at 1000Hz.
If the compositor wants to notify the client as soon as possible, it has to send 1000 messages per second. If you buffer them, you're wasting the hardware's potential, if you don't buffer, them you're doing 1000 write()s per second, which is... ugh.
If you're literally going to design the protocol from scratch and require all existing software to deal with it, why not pick the IPC model that doesn't have this issue.
Wait how do you solve that with shared memory command buffers? Don’t you need to involve the kernel to notify the receiver that you’ve written stuff to the command buffer?
Or would the receiver spin in a tight loop on a memory load from some byte in shared memory which indicates a new buffer is submitted, so that it gets notified without involving the kernel? Or is there some fancy mechanism I’m not aware of?
You most likely already have a tight loop lying around - the compositor needs to composite the screen each frame, you can probably poll in there. The client likely has one too, if not, you can involve the kernel & scheduler. If you need super high precision you probably busy wait, I don't know what the Linux scheduler's resolution is.
I would probably expose a poll() and let the client deal with it, I don't know if there's a one-size-fits-all signaling mechanism. But you have control over it, which is probably another plus.
Those loops aren’t tight, they sleep for like 10-16ms for live windows or much longer for background windows. You were talking about receiving updates at 1000Hz.
Just to let you know, 1000 messages per second over a UNIX socket between two processes is very much well within the capabilities of any recent machine running, say, Linux. That’s not many at all.
> But imagine some piece of exotic hardware, e.g. computer mouse, that reports its movement at 1000Hz.
Speaking of... it looks like common Wayland compositors [0] still kill clients that can't keep up with "high speed" event generators like 1kHz mice. [1] So, that's nice.
(For people who plan to retort with "just handle events in a timely manner", check out the comment here [2]. OSX, Windows, and X11 all cope just fine with programs that go unresponsive for multiple seconds. If the statements in this bug report (and the reports I've read elsewhere) are accurate, Wayland doesn't... and that is inexcusable.)
[0] ...or whatever the Wayland terminology is for the thing that does the work of the X11 compositor + window manager...
They’re right on this one, shared memory isn’t some scary dangerous thing. Both processes will just have some region of their respective virtual address space which are mapped to the same physical memory, which they can use to share data. Wayland already uses this for pixel data.
Not really, you can have one command buffer per client or process, and map each one in the virtual space of the process that's supposed to write to it.
A fundamental part of the Wayland protocol is passing file descriptors through the sockets so this doesn't generalize to network sockets. It also can't be done with shared memory.
I might be mistaken, but both need a physical monitor turned on. If your monitor goes to sleep you can't connect/see a black screen.
The kde one doesn't support remote user login, and while the gnome one does on paper, I never got it to work. The remote connection situation in Wayland is a major regression.
Yeah that's fair, the KDE people have remote login as a goal but it's not there yet. Waypipe is definitely the more mature option at the moment if you need a headless workflow.
> I really wish people gave a damn about the “gui over the network” problem x11 solves.
Security-wise there are concerns but...
Early dial-up Internet days (early for me), 28.8k modem, I was already running Linux, probably on a 486. I also had a very old PC laptop (I think a friend of my parents gifted it to me after he got a new one from work), a 386 I think (with the horrible slow display/refresh rate: a TFT IIRC). I used a parallel cable and PLIP (Parallel Line IP) and X11 networking to send a window manager+browser from the desktop (the 486) to the laptop.
So my brother and I could both go on the Internet at the same time.
It felt like the future and, honestly, we've kinda seriously regressed when it comes to "GUI over the network".
Your story here also illustrates why GUI over the network used to be a much more important use case than it currently is.
These days it’s unbelievably niche, as opposed to more controlled screen sharing scenarios. I think it’s understandable that it doesn’t get a high priority.
Your AGI seems to have the capability of super AI; can a human do any of that? No, is there any reason to believe super intelligence is a possible outcome? No, so how is this not fear mongering?
> The Federal Reserve Bank of New York recently placed unemployment for recent CS graduates in the United States at 6.1 percent, with computer engineering graduates at 7.5 percent. Compared to philosophy majors at 3.2 percent and art history graduates at 3.0 percent, those figures look alarming.
Alarming doesn’t begin to describe it. This is an existential crises for our industry. The situation for entry level has been dire for some time. Those of us who have decades experience have nothing to worry about; the companies who replace juniors with AI are doomed. It takes years to gain proficiency with art of software engineering. Who will replace us? Or what am I missing?
Yes, this has unemployment computer engineering at #2 with 7.8% and computer science at #5 at 7.0%.
Philosophy is at 5.1% unemployment.
The next column is also important to look at - the underemployment rate. Is the graduate in a profession that requires the degree.
The underemployment rate is defined as the share of graduates working in jobs that typically do not require a college degree. A job is classified as a college job if 50 percent or more of the people working in that job indicate that at least a bachelor's degree is necessary; otherwise, the job is classified as a non-college job.
Philosophy has a 47.1% underemployment rate. Half of the graduates with a philosophy degree aren't employed in a job that requires a college degree.
Underemployment for computer engineering is at 15.8% (3rd lowest) and computer science is at 19.1% (9th lowest).
If you want a unemployment rate for computer science that matches philosophy the answer is easy - hold your nose and take the front desk receptionist job.
Also... sort by "median wage early career." Computer engineering and computer science are #1 and #2 at $90k and $87k. There's something important there too - most college graduates are not getting $100k/year jobs. That expectation of Big Tech wages out of college and turning one's nose up at a job that offers the median claiming that "it isn't competitive" may be contributing to the unemployment rate.
There isn't an existential crisis there. Most college graduates are finding jobs in the profession and computer science and engineering (from that data) are the highest paying college majors.
I'm glad you pointed this out because I think the difference is due to philosophy grads being ready and willing to enter the workforce as a welder or an au pair or a restaurant manager, whereas a CS grad is gonna hold out for a CS job.
Source: all the B.A. Philosophy grads I know who entered basically any job they could get, often including the trades, and knew during their degree that that would be their path. But wow are they more interesting to talk with and more well rounded than a tech-head who turned up their nose at their humanities prereqs during university and as a result know nothing about the world outside of their narrow field.
Isn't a philosophy also a popular pre-law degree? A lot of these sorts of articles neglect to account for the actual reality of how these degrees are used.
Software tends to be complex enough that you need a lot of people and thus a tech company. It rarely makes sense for a company to make their own software that they only use to internally. Many non tech companies makes their own software but it is shipped to customers as part of the product
>It rarely makes sense for a company to make their own software that they only use to internally.
From my understanding China operates this way. They supposedly have such an oversupply of software engineers that every company just build all the software they need internally. Now with AI they have supposedly been super aggressive in adopting it that its probably even more of the case that everyone is building most of what they need internally.
Eh it depends. I’ve worked at / with a lot of more traditional non-tech companies and you’d be amazed at how a lot of the software looks like Excel circa 1995.
I guess they could be using third party software but it seems like often they are just using an ancient thing they built themselves.
Sounds like you are picturing WinForms in your mind (Was so awesome to create forms and ship really customized usable software quickly). Does business software really need to be super pretty?
No definitely doesn’t need to be pretty. My point is more that building and managing this stuff often requires a programmer. It’s not “cool” or cutting edge but it’s a job.
There is an image crisis. Yes, it's not a badly paid profession. But the perception that it's a dead end will lead to a sharp drop off in the student numbers.
> Alarming doesn’t begin to describe it. This is an existential crises for our industry.
In my (admittedly vibes-based) opinion, this is just a result of there being a huge supply of CS grads in this country due to it being popularized as a path to a stable, high-paying job. Those degrees are now more often than ever held by people who aren't necessarily passionate about, or good at, the field.
The signal-to-noise ratio in hiring, therefore, is worse than ever. AI exacerbates the problem, of course. But I don't think this is an existential crisis; I think the market will sort itself out, as those less-qualified entrants leave.
I think we're going to see a big scramble to pick up the pieces in a few years when a bunch of vibe-slopped houses of cards come crashing down. I imagine it will be like the demand for COBOL developers but on a much more massive scale.
A few major failures will scare the risk mitigating bejesus out of some kinds of businesses, but maybe AI will be better than us at fixing those kinds of problems by then.
COBOL was mostly outsourced to India, and it's a terrible professional path for anyone in the EU or US, and has been since the Y2K bugs got fixed at the last minute.
(And probably a bad path in India, too, but I have no data one way or the other. It's just that all the excellent Indian devs I know use almost exactly the same tech stacks I do.)
Then why did it take Anthropic over a year just to fix the flickering issue in one of their main products when they have internal access to the latest and greatest models?
In the case of vibe-coded slop like OpenClaw it's not a question of some vague notion of "code quality", it's a case of the software shitting the bed and not working anymore, with no recourse of fixing it. (Neither humans nor LLMs have the context window to analyse and fix tens of millions of lines of code slop.)
> and Anthropic banned OpenClaw usage
If OpenClaw wasn't broken it would just use a standard token API.
But see above - as software it is fundamentally broken and unfixable.
Such a correction was always going to happen. Coders always were the blue-collar workers of the 21st century, and capital ruthlessly optimises for profit. Where you once needed thousands of workers to run an assembly line, you now have dozens; where you once needed hundreds of programmers to run a big SaaS, you will now have a handful. It was always inevitable.
That doesn't mean we're all dead or anything - factory workers still exist, developer jobs will still exist. They'll just be far fewer than they used to be.
I do tend to agree. Though at the current pace of change I don't know if we can take it for granted.
As a recent example, I was on a chat with the two most experienced technical people in our company and the original developer of a feature trying to work out why we were getting a null pointer exception in a very specific case. Of course we had a fix, just a guard against the null pointer, but I'm always uncomfortable with not knowing the underlying cause.
I kept digging while someone promoted the fix. Eventually ruling out two of our original theories as to why it happened. Until eventually someone just asked Cursor which spit out a theory which matched the symptoms perfectly and which we quickly reproduced locally.
I still think we'll need some kind of human who lives in that wide space between the 95% of the population who couldn't get Excel to sum a list of numbers and the machines but the industry will be unrecognisable.
In your example you knew the issues with the original fix, had some ideas to the cause, even if they were wrong, and generally knew where to look.
In my experience the LLM when given the ticket would have done the original null pointer guard fix given the bug. Only under direction does it ever dig deeper and for me it'll often go down some wrong paths unless I tell it to go somewhere else. It's great when it gets it right the first time. But that is rarely the case and usually you just get good enough if you don't care to go further.
I think that figure (haven’t verified it but assuming it’s true) isn’t complete. It hides who and where those people are - for example, I imagine art history skews towards higher ranked schools in the first place.
Unemployment is based on the amount looking. I gotta say, how many philosophy students do you know actively looking for jobs? Now ask yourself why you think it's zero.
I had a frank conversation with a hiring manager about it.
What he said was even if we hire juniors, juniors using AI are never going to rise to the level of our current seniors who built decades of experience without AI.
So basically, today’s juniors are not worth investing in. Until society really sorts itself out with responsibile AI usage in a way that still develops independent professional skills, there is no point in hiring juniors. They will just be a more expensive version of whatever AI agent they use, which can be used directly by seniors anyway.
Companies today do not have to really worry about who replaces the seniors, that will be a problem for newer companies in 20 years or so. In time a solution will arrive naturally.
thats interesting, the HMs where I work love hiring juniors (who pass the bar) because they are so AI-native
the more experienced engineers can help with setting guardrails and mentorship, but the juniors come unconstrained by priors on how to use ai in creative ways to solve all sorts of business problems.
> I was able to ask it questions about what a particular tube or group of components did,
Was it correct or hallucinating? Do you have the knowledge to tell the difference? I’ve been burned too many times to take what they say as the truth without checking; especially in a subject I’m not an expert in.
Corporate "hackatons" are often a mess. We have anything from a truly novel ideas clearly invented right at the hackaton (as was originally intended) which are rough and very unpolished and buggy or even broken, to teams brazenly bringing up developments which clearly were ongoing for months, not even hiding that and showing weeks or months of test and dev results in the eventual presentation. The latter teams won of course every time. I kinda get the business benefits, but the spirit of the contest goes out of the window.
Oh, and don't get me started on the fact that a lot of developers get two relaxed fun days while with catering, networking and basically paid for self-improvement workdays, while QA, supports and other teams are expected to work as usual AND cheer for those participating and watch presentations (thankfully that last is optional).
> but the spirit of the contest goes out of the window.
"-thons" aren't contests, at least not in the traditional sense, they are activities where participants test the limits of their endurance. A marathon, for example, allows runners to see how far they can push themselves running. Likewise, a hackathon gives a place for one to see how far they can push themselves to create something over a short amount of time, beyond what would normally be possible, and beyond what would be sustainable in an everyday setting. I suppose you could argue that it is contest with yourself, but calling that a contest is atypical.
You are right to call out that few people want to push themselves to their limits for a corporate event, so corporate entities have turned to contests instead to try and find something that does appeal, but a contest, no matter how it is conducted, is outside of the spirit of a hackathon.
It is viable for a race to take place at the same time as a marathon. From a logistics point of view, there is benefit in trying to draw in both those interested in marathons and those interested in racing. But there is a difference in motivation. Choose at random one of the 70,000 registrants of the NYC Marathon and ask them if they plan to win: the answer will almost certainly be no. They are there to see how far they can push themselves only. Whereas in an event that is strictly conducted as a race, generally every participant will tell you that they are seeking the win.
It is also viable for a sales pitch competition to take place at the same time as a hackathon. When we look past the early hackathon days, we can observe a trend towards events hosting both. Similar to marathon/races, appealing to a wider audience helps with logistics. Maybe that is where things get confusing?
Perhaps this is easier for you to reason about with a telethon? Rarely do we find a competitive element show up as being an objective of the event in that setting. Only the endurance component, where telephone operators push themselves to answer phones for periods of time beyond what would be considered normal.
Hackatons are commonly used as a way to take credit for & reap the benefits of another person or team's work, without attribution or compensation. And oftentimes, a promising hackathon idea will be "improved" by management & added to the creator's workload with tight deadlines (because the hard part is already done!) -- even if they don't necessarily agree with the "improvements".
Yep, a way to get free work by pretending its fun, most corpos immediately turn around and do this.
It's also funny to me because of how they try to show its a treat to the engineering staff, and then railroad them as soon as they can to implement the half baked idea.
The truth is that most management don't ever get beyond half baked ideas and so trying to push you to make low quality crap is often their only move.
I dunno, I think I got more out of it than I put in, but it was mostly due to happenstance and knock on events, and my name sounding more familiar to some important people, but this wasn't really in the cards.
Imo if you don't do stuff others dont you wont end up in places others dont, which might be good or bad.
I mean, one of the biggest raises I got was when I brought my dog to the company cookout, and it turned out my boss's boss was a huge dog person, and we bonded over that, and he decided I was a good guy, which was kinda ironic as I was working my arse off to little benefit at that time.
Same scummy move as streamers / online personalities having a "contest" for a t-shirt design or logo design, which the streamer will sell for their personal profit while offering nothing but "exposure" to the creator.
I did a 3-month bootcamp back in the day and the top assessed final presentations were a powerpoint and some bs JS "game" that indeed was a bunch of nice graphics being manipulated with code. This was a Rails bootcamp, not a powerpoint one, not a js one...
I did a hackathon many years ago (before LLMs) where I spent a serious amount of effort training a conventional machine learning algorithm and integrating it into a react native app. I had a genuinely impressive team to make this possible given it was 2016.
The winning team bought a bootstrap theme for $35 and made a landing page for a nonexistent app.
When did this happen? I remember judges at hackathons used to be very forgiving about lackluster UIs as long as the idea was cool and at least functional by the presentation time
It depends a lot on the hackathon/what the judges are looking for. A few are run by technical people who pick the coolest technical architectures, a few are run by casual users who pick the best looking result.
The majority of the ones I’ve been in have tended to be run by people who judge based on their notion of how useful the app will be societally, with the tie breaking factor being the UI/architectural design.
10+ years ago, when most "grassroots" (and some of the better startup) hackathons were displaced by enterprise-sponsored hackathons. I can mostly talk about the Berlin hackathon scene, but as far as I understand it the same thing happened in SF/London as well around the same time.
Presentation-first judging has been a thing for a long time, and unless there is a organizing party that explicitly makes code reviews a part of the scoring, and the organizers ensure attendance quotas for different personas (engineer vs. product vs. designer) it will always drift that way.
Reminds me of the inconsistency of take home interview tests where you have no idea if the person reviewing cares that the UI is shiny or not or if they want you to write a novel in the readme and make it look like a real project.
This has pretty much always been the case. You've never been able to build production-level software in 2 days (not even in the age of AI, no), so it's always been about having a UI with mocked data.
I did a lot of hackathons when I was in school more than 10 years ago and that's how they all were and what every team did.
I had a similar experience in my undergrad software engineering course. One group literally took a JS facial recognition library and wrapped a halfway decent UI around it. That’s it, that was their entire project. But the grad student teaching the class and a lot of the other students were very impressed.
You'd be surprised about how much production level software was built in two days — the fact that organizations are usually unable to do that is what gave rise to the agile movement, though it falls apart as soon as management asks for agile coaches and for agile teams to document their processes for others to use.
Llms don’t learn after they’re trained. At the most it could maintain a profile or some such. Which may be the equivalent of what you mean, but not as provocative.
> Quite frankly, I'm kind of excited to see what OpenAI can build.
I’d go further and claim if controlling the hardware is a existential threat for these companies, their ability to deliver or not should be seen as an indictment of the entire “llm formulation of ai” era we’re currently in. If llms have the potential they claim, then empowering them to design a best of breed phone should be table stakes.
> If you want the nuclear option, can you move away from the current environment?
The common refrain here is usually they can find a bad environment anywhere so just moving won’t solve the problem. Unless it’s really remote, and home schooling. Which would not be healthy for anyone imo.
reply