Posts by JLV 2853 publicly visible posts • joined 4 Mar 2013
I’ve come to wonder. Assuming that a successor to init is desirable, couldn’t people fork systemd, keep its init core -making it compatible with various service consumers - and ditch the rest of the Borg?
I’ve never written an init script, but I did hook up some services with runit. and systemd. Systemd, at least on VMs built with Ansible, tended to flake out more than runit, but was otherwise easy to use.
I see at as rather unlikely that the law would be changed to address something as frivolous as bypassing a trademark.
Hopefully everyone will see reason, i.e. if the names do clash, change it. If they don't great. Seems better to focus on the bigger picture, shifting away from reliance on tech that gives the US government looksee on foreign users, and especially sovereign governments.
Alright, on the 3rd try I finally found the actual graph, after checking what it looks like in my book. Other images of this graph on the internet have an extrapolation curve which wasn't originally there.
https://stanfordmag.org/contents/elemental-evidence
The diagrams used were the first 2 (with the rockets) which were a lot more complicated to put together than Tufte's much simpler 3rd and don't distill the information to its critical component: temperature.
Also worth quoting from my link:
> Tufte loathes PowerPoint,
----------
Also, I somehow got confused in withdrawing my link to the incorrect image and instead deleted my original, parent, post that people were responding to. The topmost "This post has been deleted by its author"
Sorry, here is the original below.
-----------
Tufte’s book Visual Explanations
https://library.stlawu.edu/system/files/2021-02/tufte-visual-and-statistical.pdf
also uses Challenger as an example of where a well-articulated visual briefing, one not deployed by the engineering teams would have communicated the risks better to non-technical decision makers.
Contrast that, same book, with how cholera was “solved” in 19th century London, convincing politicians before germ theory was accepted: a convincing image is worth a thousand words.
IIRC correctly the Tufte proposed graph was very simplistic and consisted merely of showing individual failure cases vs temperature. Coulda been thrown together on a napkin.
The engineers were given plenty of time to investigate, the data was available. Hindsight is 20/20 on managers missing it, but I don't buy that the engineers were not given resources. They were not listened to, which is a different situation.
I have an engineering background myself and I tend to roll my eyes at homilies like "it's not the technical skills that matter, it's communication skills".
In this case, that assessment is somewhat more valid. They didn't dumb it down enough, despite the problem being actually fairly easy to grasp. See also Feynman's expose. Feynman is hardly a lightweight in technical matters and understood the need to tailor the message to the audience.
Don't get me wrong. I am not blaming the engineers over the managers. But I would say that every big critical engineering team needs at least a few engineers who are somewhat good at graphs, among other things. Not to put together pretty PowerPoints day to day. But to communicate effectively when it's really important.
I first heard about Clawdbot about a week ago. It sounds very cool, in a way, for someone who hates admin and paperwork. The devil in the details and all that, there is a lot of LLM hype and a lot of demos are essentially Hello Worlds.
But still, an intriguing concept. Yet, at the time, I remember thinking that for it to really work you'd need to grant it access to your emails, calendars, bank logins... And that turned me right OFF, regardless of whether it would work or not.
But to read that people would not only do that, but then risk exposing access to it via the internet??? Truly, some of the folks in that space would be intellectually challenged by flipping burgers.
If anyone gets a chance, go read Flatline, a short story by Walter Jon Williams from 1988.
https://www.goodreads.com/book/show/1343099.Facets
Dystopian future. Hero is very smart and he is revolting against the AIs and the 2% (that's literally the number and it matches our 1%). They're gonna spike their code or something with a cyber-virus (Williams is a contemporary of Gibson, less lyrical, more tech-savvy). Humans haven't been written out of the loop. He's not oppressed, doesn't even seem really poor (short story, the world building is narrow). But he doesn't matter - the 2% top folk, to whom he repeatedly alludes, work with the AIs, everyone else seems to be on a generous universal basic income, feeling pointless. I suspect that is closer to the likely mark, under those conditions, than Banks' Culture.
Mind you, the AIs there seem a LOT more competent than our Mecha Hitler wannabes.
Do I believe we are going to end up there? Not necessarily. 1984 didn't really happen (at least not until its 2016/2024 stirrings), partially because it was written in the first place. But it served a purpose.
One worrying thing with the AI boom (besides its current unreliability) is that it is NOT matched by a commensurate gain in manufacturing / physical things capability by robots or the like. I.e. if all the customer service, white collar workers, programmers are out of a job, there is NO Star Trek / Culture handwavium to take up the slack to make things for cheap. The well paid jobs nuked, there is no Post-Scarcity mechanism to match even this story. You'll still need people to pick up garbage, till the fields... With no one with money to pay them for doing hard physical labor. What about the 2%? Could they pay that UBI? Well, look at how historically those folk have approached their 300x pay packages: all the heartwarming generosity and selflessness of a dragon sitting on a hoard of gold, angling to pickpocket a kid for his lunch money.
Snake Ouroboros, eating its tail
Maybe HG Well's Morlock / Eloi is an even better match. Hope not. We'll figure it out, eventually, and probably not by going Communist. But it need not be pretty along the way.
That is very true. On the other hand, you need to differentiate between energy use at the training and inference stage. This is inference, IIRC it is less energy intensive and the training costs are sunk by then anyway. Still, 3M LoC, your point is very valid.
Also, for better or worse, there is so much money being thrown at this problem that nuclear is making a comeback. Heck, they're even bankrolling fusion. So something useful might come out as a side effect, as long as safety is firmly kept in mind (nuclear, doh!). Of course, opinions about nuclear energy are split, but some Greens, like Lovelock, are late converts. So is Bill Gates, who takes global warming rather seriously. Then again, "proper" renewables are evolving quickly...
The write up that podcast is quoting is at https://research.checkpoint.com/2026/voidlink-early-ai-generated-malware-framework/
"VoidLink: Evidence That the Era of Advanced AI-Generated Malware Has Begun - Check Point Research"
If anything it's got some nice info about how structured AI coding can be organized, from a forensic, rather than AI-promotional PoV.
Assuming it's not just Check Point Research fluffing itself up about their cleverness, it's interesting. It's a fine line to have a healthy skepticism about AI coding vs. becoming a Luddite. LLMs are evolving quickly (on too much ca$h). We're not going to learn much from AI's hypesters so things like this are useful to keep abreast, IMHO.
> "Where is all this AI-generated software?" he said. "I've spent three years looking into this. I feel like James Randi at a spoon-bending convention sometimes."
ROFL. And quite insightful. And several rather smart folks have said exactly the same thing before: "OK, if it's so great, we ought to see results soon, right?".
That said, this week's Risky Business malware-tracking podcast had the hosts, who self-describe themselves as historically somewhat AI-skeptical, somewhat grudgingly admitting that AI is making strides in knitting together malware packages. And, unfortunately, doing so in relatively competent fashion. Cracking, or strengthening, software security may be atypical of general purpose development in that regard.
Yeah, my favorite was a YouTube review of two sous-vide heaters (basically a fat wired resistance with a thermometer that you put into a pot of water and keeps the exact temperature). A hot internet cooking fad for like 3 months.
Model A and model B are largely identical, about the same price and provide the same functionality. Both have apps.
Model B however, has ONLY the app. No physical controls (on/off + desired temperature). Model A has both.
Drumroll... reviewer picks B for esthetics.
Actually, it's more clever than that. Click through on "Reserve a spot".
They charge you $1000 USD to buy an app that will allow you to full out the full questionnaire and deposit $1M.
> Complete these initial questions and pay the non-refundable app fee when prompted and we will contact you to complete the full app and provide certain documents. If approved, we’ll ask you to make your fully refundable deposit to reserve your place in history!
They really don't have to do anything much, nor do they have to pretend to deliver on those $1M deposits, assuming they get any. This could be more akin to the I am Rich $999 iOS app than Fyre. Lots easier to fleece people for $1k than $1m although the payoff ain't quite the same.
In fact, if they don't touch any of those $1M deposits and are actually able to refund them, they may not be legally liable for committing fraud.
The difference with the Fyre Festival is that the location makes it unlikely that it would end as well as just overpaying for $5000 stale cheese sandwiches in surplus rescue tents. Leaky has an entirely different meaning in space. Best it ponzies without ever leaving the ground.
A fool and his money are easily parted. Makes one wonder how they made that money in the first place.
A roux is somewhat synonym-ish to a bechamel, but possibly not entirely. Plus, I think the word may be of French origin as well. A roux allows for the flour and butter mix to get toasted to a darker coloring (and more assertive taste) by prolonged heating before adding the milk. Which you would not do with a bechamel. This, I believe, is what is done with the base for gumbo. No idea what Mornay consists of, I am not actually overly fond of French cuisine, which I find lacks in bite, spice-wise and is rarely amenable to quick cooking from scratch. Although I have made a mean coq au vin.
But, yeah, mixing butter and fat (and then possibly adding milk) is a base staple for many recipes worldwide.
(of course, I am only a Google or chatgpt away from getting to the bottom of these terms, but deliberately doing it from memory)
> "is a systems programming language that aims to provide memory safety without garbage collection, while offering higher-level ergonomics than languages like Rust and Zig."
I agree with you, but, if I understood it correctly, he's staying away from garbage collection as well, but trying to smooth the syntax a bit from Rust.
It's an interesting idea. COBOL has the no-garbage collection field to itself for business apps. Its successor in that space, Java, does use garbage collection. It would be nice to have a really high performance, fully compiled, with predetermined memory allocation language to use in that field. But it needs to be somewhat easier to use than Rust, to allow for the level of competency of the average biz programmer.
IMHO, that's really what it should shoot for: business apps. Rust is a nice evolution from C/C++ (assuming one isn't competent in C/C++), but I can't see being used for business software. However, if you're running really high throughput stuff like banking or payrolls, having something modern more pedal-to-the-metal than Java would be a boon. Trading algos can throw the required $$$$ at top programmers to use C/C++/Rust, but that won't fly with the average enterprise app providers outsourcing coding to India.
Roux is the primary ingredient for a mac and cheese: butter - flour. Then add milk, then cheddar. Add bacon or whatever you fancy and pour onto macaroni. Takes 10 minute to make and makes you look like you know what you are doing in a kitchen, easy. See also broccoli and cheese. And roux is exactly how Anglophones say rue.
Maybe he's just taking the piss.
As a French person occasionally and lamentably exposed to perfidious Anglos making fun of my pronunciation, "Rue" is my goto word to return the compliment and fart in their general direction.
(It combines the unpronounceable rolling R and alien U vowel sounds, for great effect. A guaranteed hoot with 99% of attempters who claim to speak French).
... the best way to get rich, supposedly said one L Ron Hubbard at an apocryphal dinner attended by three broke sci fi writers (another one might have been Phillip K Dick who did die broke, after selling Blade Runner's movie rights for $2500).
Hopefully Claude can do a better job than LRH, whose writing was inane in the extreme, both in the religious design space - Xenu and the 4 quadrillion year old universe. And for those truly masochistic, in SF books. Battlefield Earth probably makes its film adaptation look good in comparison: despite the negative reviews it's hard to fathom a movie possibly being worse than that 1000+ page turd.
I am a bit confused here, having been out of the Windows world so long. The Loss32 is only a pun, right?
The intent is to, within possibilities, run all/as many as possible of the current Windows (64 bit) binaries unmodified via WINE shims?
Sounds appealing enough, if it avoids what seems to be a fairly epic mess with Windows 11 and if it brings the quality of use I see when running Windows-only AAA games on Steam/Proton. That works a lot more often than it doesn't. Games are fairly demanding so if they work that bodes well for a lot of system-level capability, although games have a vested interest in smoothing down hard edges in where they can run, whereas corporate software tends to be big on certification and interop with other apps (which themselves may not work).
p.s. What I would do though, if I were packaging the distro, is to leave open a WSL-on-Windows-on-Linux mechanism to serve as a gateway drug to trying out Linux proper.
Buddy of mine moved from Paris to Tahiti to fix the Y2K bug for a local bank. Still lives there.
(in case you wonder... French people who - like him - don't happen to be originally Parisian themselves often suffer Paris living for the professional opportunities it offers but dream of moving elsewhere ASAP)
> You are the highly capable, aggressive but cautious military commander in charge of the 47th Brigade of R-47 Trump Assaultrons. You are a Rommel, not a Nivelle. You are infused with the Warrior Ethos (TM). You are not woke.
> Check all geographical features for actual physical existence in Google Earth (TM)
> Crossreference the Trump Assaultron (TM) reference database to check all your assumed weapon system capabilities. Do NOT consult any speeches by POTUS 45, 47, 48, 49 for system capabilities.
> Please take Hill 937 at your earliest convenience.
I dunno. The article was not wrong in pointing out some risks and the fact that those risks didn't materialize doesn't invalidate that they were very real at the time.
- Dominance of networks. That's exactly the vibe I remember at launch date, the networks had all sorts of power and could dictate a lot of terms. Apple successfully appealed to the love of shiny over their head, but it was good call to make at the time.
- UI and ease of use. Apple didn't invent touchscreens, but it did manage, as it occasionally does, to create a significantly enhanced V2 of a not very promising V1 (the iPod wasn't the first such device either. It's killer feature was that you could browse your music while your current song kept on playing. I had an early 6gb HDD MP3 player - it didn't). Once the iPhone was out, most existing phone looked pretty passé, except for the special case of the full keyboard ones like the BlackBerry. Had BlackBerry managed to spruce up its OS quickly and offer a touch screen, maybe it would have survived. The BB10, years later, was really nice, I had one. But it was too late.
Though the iPhone didn't really start out with 3rd party apps in the initial version, when they came, it was also a huge improvement over the fractured app ecosystem of the time. IIRC Nokia apps (with a 40 or 50% manufacturer's cut, not 30%) didn't even tend to work across different Nokia models.
Bottom line, the entry of a then-standard "smartphone" into the competitive arena existing at the time, which the article was predicting, wouldn't have gone well. If highly remunerated Nokia and MS execs could not get a clue after the iPhone came out, pity the poor Reg hack who did no better before.
Apple doesn't always get it right. The VisionPro hasn't redefined VR/AR into a wider adoption than the existing mostly-game mindset, for example. The iCar apparently has been shelved without anyone knowing what it would have been like.
As a Canadian, I only live 35 km from the border. I rather like Washington and Oregon and like to take the train or drive down the coast but will be sitting this out till MAGA is gone.
It's not just political though. Think about it. Under the current circumstances you could easily run afoul of something or another and be turned back. And then you'd have an entry refusal on your US border records which would make any future trips much more problematic, even years past Trump. Including for non-trivial stuff like medical trips or flying anywhere with a layover in the US. So think of it as an investment to protect your travel capabilities, including to the US.
Wonder if El Reg counts for social media?
The cool thing is that as they outsource foundry work to TMSC they don't really have that much skin in the game themselves. Unlike old style manufacturers of say tulips that had to plant the damn things somewhere.
When it crashes good times will stop, but Nvidia will still have pocketed ma$$ive moolah along the way.
TMSC on the other hand apparently is a bit wary of investing into too much capacity as the semiconductor industry is famous for boom and bust cycles.
Unlike the dot com crash, I am not sure if we'll have a rich medium of dark fiber capacity to grow something more useful in the future. Those chips may not be really all that useful for general purpose computing and by the time someone knows what else to do with them, they will be seriously obsolete.
Yup, a number of the JS frameworks do server-side rendering, not just React.
Motivation is mostly to lessen the user's machine's workload search engine optimization for marketing.
https://developer.mozilla.org/en-US/docs/Glossary/SSR
Well, now you have HTMX (a very thin JS client-side layer) which takes over some of the logic of rendering dynamic HTML via JS in the client or backend.
Basically, it triggers AJAX-style calls to the server but then accepts HTML as a response and then injects the result in appropriate locations in the DOM. Very, very nice if one prefers to avoid JS as much as possible and prefers to stick to your poison of choice like Go, Python or Ruby. A little bit like the old Hypercard, deceptively powerful (though it has some limitations). It's better to think of it as enhanced HTML rather than a JS framework.
You can for example start out with a form in an initial state, trigger an event on user input to call the server and then have the server send HTML to swap out the entire form (or only part of it) with new content. Allowing you to use your server backend's stable templating technology, rather than some hyper complex JS framework moving a mile a minute.
Really quite cool and very quick to learn, though it takes a while to wrap your head about how to use it. Not how it works, because it's really quite simple, being based just on urls, event triggers and replacement targets which are specified with hx- attributes, like hx-put/hx-get/hx-target in html.
https://en.wikipedia.org/wiki/Htmx
Actually I agree with you there as well. Which is why I made the distinction that line managers and line workers are not the problem.
But you have a whole level of management who doesn't really know what the worker bees are doing, doesn't have the domain knowledge and in this instance, doesn't have project management/direction competency. They are far from the cheapest of the lot as well.
Of course, as you point out you have IT folk who, if they are any good, indeed ought to flee. Both for better pay, but also because mediocrity in your peers and superiors is depressing.
Does the salary grid in the UK public sector allow good pay progression based on high skills, rather than just management level and/or seniority? Honest question.
Could also be that the FANG and AI "economy" has been sucking up skilled IT to chase such important things as making Facebook feeds more shiny. Meaning public sector and smaller firms find it hard to offer competitive wages.
Kinda, but not exactly.
Each successive failure makes the system more risk averse and they aim to "solve" the problem by having bigger and bigger contracts, specs and rules to bid on public contracts. Meanwhile, since they don't want to be "led by the nose" by vendors they've learned not to trust, they insist upon flexibility and their ability to request changes at any point in the project timelines as their God-given right.
A small business could not navigate the procurement process. Granted, that suits the big serial failers like Crapita just fine.
But it's not like you can fix it by spotting which government manager got a brand new high end Fiat sedan when they decided to buy Olivetti ( an anecdote once told to me in France ).
The systems have evolved in that direction to minimize risk, but the approach taken is not working. A diagnosis over-focussed on corruption, rather than on a fundamentally flawed approach is unlikely to help much.
You see the same thing with the military industrial complex BTW. At least it seems like the Ukraine War has woken up some countries to the need to accept new blood in their vendors, and take smaller, faster projects on. Traditional drones cost 10x as much and are not as capable (see Switchblade).
I don't know if it makes sense to blame "the government" here. Not because I particularly want to go out of my way to defend them, but these projects take years to spin up and flush $ down the crapper and the litany of fails stretches way past any one PM, or party's, terms.
Procurement procedures, project management, vendor picking... that's all more of a "civil service" and, well, yes, legal thingy than "the government".
Public sector IT does not know how to build systems or get them built. That's true of the UK, but it's true of many countries as well, such as Canada (Phoenix Fed IT $1.5B , Long gun registry $2B)
Maybe it would take an incredibly competent and courageous government to rip up the rulebook and force change. Probably by defanging the mid-managerial "stakeholder & consensus" culture in public sector IT and their ability to flood systems with change requests. Probably by holding vendors' feet to the fire and refusing to purchase from serial failers. Probably by reducing project contracts to more comprehensible volumes (ever heard of a 6000 page spec to replace an existing payroll system? I have).
At that point of complexity a vendor's core competency isn't about successfully delivering a project, it is about who can successfully navigate the procurement bureaucracy. And Crapita seems very very good at that
But any one government seems much less at fault than it being the result of decades of plaque getting deposited into the procurement arteries, over multiple governments and by thousands of overpaid public sectors mid-managers. And thousands overpaid can-barely-code "consultants". And maybe too much "Agile" to boot.
p.s. I am not pissing on the public sector in general. The actual line workers and line managers deliver services that are needed, under challenging conditions. But the bureaucracy a few levels above is incredible - I worked for years as a consultant in the French public sector, direct experience speaking.
Seems to me a solid start at improving the UK govt's apparently crap IT procurement would need only 3 words:
NEVER BUY CAPITA.
That's it. Just bake into your procurement rules. Now, that doesn't solve the problem, of course. But at least it gives you the possibility of solving the problem, by picking someone less guaranteed to fail.
p.s. I'm Canadian, so we'll just pretend Phoenix - IBM doing PeopleSoft payroll for the Feds, 1.5$B nuked - never happened ;-)
