ULA is stuck with SLS, which had its high-level design micromanaged by Congress in a way that guaranteed it will fail (at everything except collecting government funding). It makes sense that Bruno is jumping ship shortly before the reckoning comes for SLS, to a company where success is possible.
Boeing has the contract for SLS, not ULA. Boeing owns 50% of ULA, with Lockheed Martin owning the other 50%. But SLS is a Boeing product not a ULA one. ULA's main rocket now is the Vulcan, with a few more Atlas V launches left.
If the only GPLed component used is the Linux kernel, you probably aren't entitled to any noteworthy source code. It's well established that using the kernel doesn't create a GPL requirement userspace software running on the same device, and the most likely arrangement here is a completely-uncustomized kernel paired with an open-source userspace program that does all the interesting bits.
It's trivial in terms that it will cost them nothing, because it's very likely there are no changes to the kernel, or nothing of value nor commercially-sensitive anyway.
It's not trivial in terms of big company bureaucracy - this request will have to go through so many levels of red tape that they (correctly) decided not complying to random people's requests is more profitable.
I'm sure if you actually sue them then they will comply right away, because at that point paying for some engineer's time to tar up the source tree and send it to you now becomes cheaper than lawyer time.
But their analysis is correct in that nobody will waste time/money suing to get what is effectively a stock kernel they can get from the official source anyway. Which is why these complaints are also a bit stupid - they're not asking for anything of value or using the GPL to advance software freedom by freeing up some valuable code, they're just wasting both theirs and others' time asking for something they can already download directly.
> because it's very likely there are no changes to the kernel
That is a gratuitous assumption. My experience is, as long as there is the smallest custom hardware, you will have to make some tweaks here and there.
> they're not asking for anything of value or using the GPL to advance software freedom by freeing up some valuable code, they're just wasting both theirs and others' time asking for something they can already download directly.
I'm sorry that the company which is making lots of money by using a copyrighted SW has to "waste" 200 dollars in some bureaucracy, printing and postage. But is the license of the SW they are using, and should abide by it.
> That is a gratuitous assumption. My experience is, as long as there is the smallest custom hardware, you will have to make some tweaks here and there.
How sure are you small tweaks create a derivative work? In your experience.
No the license does not hold! Fair use exists independently of the GPL license.
Using GPL software does not take away your rights under the law. As much as the angry FSF lawyers want you to believe - I can use any GPL software in any way I choose and have the affirmative defense of fair use for every single claim they throw at me.
Whether I lose or not depends on specific factors of my use - and I'm sorry but one comma isn't going to cut it.
Edit: A single comma will lose 100% of the time to a summary judgement via de minimis. That's just an incredible legal theory you have there and not something you should be spreading.
Edit2: The Beastie Boys famously won via de minimis at summary judgement for a 3 note sample. It matters not whether those 3 notes were GPL notes or not. It matters what the law is.
I'm a T1 diabetic, have worked on open source diabetes-tech (OpenAPS), and have used a number of different CGMs (though not this one specifically). This story... does not make very much sense.
CGMs (of any brand) are not, and have never been, reliable in the way that this story implies that people want them to be reliable. The physical biology of CGMs makes that sort of reliability infeasible. Where T1s are concerned, patient education has always included the need to check with fingerstick readings sometimes, and to be aware of mismatches between sensor readings and how you're feeling. If a brand of CGMs have an issue that sometimes causes false low readings, then fixing it if it's fixable is great, but that sort of thing was very much expected, and it doesn't seem reasonable to blame it for deaths. Moreover, there are two directions in which readings can be inaccurate (false low, false high) with very asymmetric risk profiles, and the report says that the errors were in the less-dangerous direction.
The FDA announcement doesn't say much about what the actual issue was, but given that it was linked to particular production batches, my bet is that it was a chemistry QC fail in one of the reagents used in the sensor wire. That's not something FOSS would be able to solve because it's not a software thing at all.
> CGMs (of any brand) are not, and have never been, reliable in the way that this story implies that people want them to be reliable
This has been my impression. I briefly used an Abbott Lingo to help me understand some health issues I was experiencing.
It's always been clear to me (including in the app and documentation) that CGMs are an extremely convenient tool as a first line - but struggle in extreme circumstances. And, let's be clear, if you would generally know if your body is in one of these extreme circumstances. You'd probably be feeling like shit.
That's not to mention the device in question, the Freestyle Libre, is (to my understanding) by far the most popular insulin-dependent diabetes CGM available.
This article is equivalent to calling the Boeing 737 unsafe because it's had the most Full Lost Events while completely ignoring it's flown 238.84M flights (which is basically more than the entire rest of the list combined).
100,000 people in the US alone die of diabetes per year (none of the 7 were in the US). Other reporting shows that 3 million glucose meters are being recalled. Diabetes can be tricky to manage as it is.
One could look at this story case-by-case and what happened to the affected individuals. Did the device directly lead them to harm? Of the wide coverage of this story, I see no testimonials from affected people.
We're left with the statistical perspective. I don't see the math supporting this story. I expect that as many or more people would have been harmed by diabetes during the time period without this bug.
I think it is more harmful to perpetuate the lack of context or analysis that brings crowds to look at statistical noise and agree that something must be done about it.
The deaths are associated but not necessarily caused by the incorrect readings, but as is often the case, medical interventions treat all observed conditions as side effects (this is similar in the case of drug and device trials), and the FDA has typically operated from an abundance of caution (though policies are somewhat erratic under the current administration).
I share the skepticism of the top-level comment by jimrandomh, in I understand that CGMs are used to guide treatment but not determine it, and that the consequence of spurious low blood-glucose readings is not likely to be immediately threatening (that is: the consequence of mistreating based on the mis-reading would be an actual high blood glucose event), though of course over the long term, high blood glucose levels are precisely the mechanism by which long-term and late-stage diabetes symptoms and conditions emerge.
Given the large number of devices (38% of US adults, or ~125 million), and millions of CGMs in use, seven associated deaths seems a relatively low number and correspondingly low risk.
TFA also would seem to misclassify the problem as one principally of software where the actual principle issue would be of potential patient noncompliance with protocols. That itself is complex, and isn't necessarily a matter of blame (the very young, otherwise ill, or cognitively-impaired might well be expected to comply poorly with instructions), but is a concern providers and dispensing pharmacists would have to be exceedingly cognizant of. As well as device manufacturers.
> This article is equivalent to calling the Boeing 737 unsafe because it's had the most Full Lost Events while completely ignoring it's flown 238.84M flights (which is basically more than the entire rest of the list combined).
You don’t get many people calling the MAX a good plane.
If you include in the count a new model which arguable should never have been allowed to be called the same plane, then yes, your prior good record looks ok. Over various generations the hull loss rate had come down to 0.18 per million flights while the MAX is at 1.48 per million flight.
I have two elderly relatives that use CGMs and both are at a stage in life now where they really cannot be expected to exercise common sense. I am pretty sure they've both been using CGMs exclusively and haven't been using finger sticks, at least not regularly, and one of them has a very hard time even understanding that apple pies are filled with sugar. No real intuition for which foods have or don't have sugar.
If CGMs are so unreliable and need double checking, I am quite confident that many patients don't understand this, even if it was carefully explained to them by their doctors.
Ok but then that would seem to absolve the manufacturer of liability. If you sell someone a hammer and they try to eat it the manufacturer isn’t liable for the damage.
Bro I'm not calling for anything to be banned, I'm responding to the sentiment that the story doesn't make sense because users of CGMs should know better.
That is odd. A too-low reading would result in less insulin and a high blood glucose, which can be extremely uncomfortable but is not immediately deadly.
If it had read too high, it could result in an insulin overdose, which can indeed bring coma followed by death in fairly short order.
Theoretical you can get a hyperglycemic coma but for that to happen you need continued and sustained high blood sugar in the way your toilet would smell like a sugar factory for quite a while.
This checks out with what a diabetic friends told me as to why he does not really uses tech: he preferred to take the time to learn "himself" and recognize the symptoms, because of such issues.
I suspected he was paranoid, but thanks for the rational explanation!
>and recognize the symptoms, because of such issues.
There are a few problems with this. I'm a T1D and your sugar level can change very rapidly and you can be near a critical situation before you feel it. Even worse is an issue after you fall asleep. Tell your friend you'd rather not find him dead in the morning.
One of the main uses of these technologies is precisely because some type 1 diabetics can become unaware of the symptoms over time either through chance adaption or over-exposure, for instance, hypo-unawareness.
Not only that, but the symptoms for hypoglycemia do change over your life, so that what is felt today (e.g. excessive sweating, blurred vision) may be totally different in the future (e.g. confusion, tingling thighs). Or you lose that sense of feeling entirely and never notice a problem until it's way too late to easily remedy.
Having high glucose levels won't kill you in the sort term, yes. But we cannot compare pre-diagnosis high blood sugar level (the body had that for months so it is accustomed to it) to the suddenty of it with cutting off insulin. In fact, things can spiral out quite quickly.
You see false low glucose figures, that last, you start reducing your slow acting insulin, you skip some fast acting insulin. Within 24h, ketoacidosis starts and you can start feeling nauseous. At some point, if you eat, you vomit. You are cornered: you don't have the carb intake to inject insulin, and you can't eat. Even worse, at some point, if you drink, you vomit, so you dehydrate, and it's a matter of hours to live. Shit happens fast, things can get critical is a few days.
Diabetes management is complicated, this is far from exact science, and having a good knowledge of everything is hard. I was already bitten by this cycle of nauseous feeling with slow acting skipped a few month after my diagnosis. I learnt to never ever skip slow acting insulin, even when blood sugar is through the floor. Prepare some apple juice and still go on.
I have Freestyle Libre 2, and it is quite a disappointing thing software-wise. I have to reverse engineer another app to get an API for my data, I have to go through Internet to get my blood sugar level (for a standalone display for example, so I can't make one that works "off grid", like... in my plane), they do sparse updates, they lag behind OS version by dizains of month for their apps, they have 10s of apps/websites, it is hard to understand. So I'm not surprised by poor bug management.
I wish some big names invest in a CGM device. Don't make it medical (even medical grade ones like Abbott & co say you have to check with a finger thingy device, so why bother), make it $500 one time plus $10-20/month, make it open about the data and you'll get everyone. Maybe no one want to invest because in 10/20 years Diabetes will be a thing of the past?
Muscle movement will cause different reading in mine. They are great for trend monitoring but not reliable for real values. ... Neither are finger measurements as in lower and higher regions they also differ quite a bit. But as usually more measurements by more different methods get you a better image.
I found having the monitor on my left arm results in a more reliable connection and consistent readings. This isn't just during the day either, so can't be explained by lifestyle patterns?
The FDA announcement make no statement in one way or an other about the cause, only that there is a problem with two monitor sensors under certain model numbers and serial numbers. It not a given that a single production batch include a multiple of model numbers and products. Assuming it is bad quality control of the chemistry is thus not supported by the FDA announcement.
It could be the software freedom conservancy assumed software bugs, with the same limited knowledge as the assumption being made here about chemistry quality control, so readers will have to decide which sounds more likely. The article do state later that "We also will probably never know whether this issue was in hardware or software... the public deserves to know the technical details ". We can make a favorable interpretation here that they acknowledge the possibility of it being software, hardware or QC. Making accident reports public information is a common step in other areas in order to allow people to learn from mistakes and produce better products.
I will add that blaming faults on human error has generally been shown to be a dangerous route when dealing with fatal accidents in all human endeavors. Correct training and behavior by patients can help to reduce fatal accidents, but one should always be careful to put blame here as a culture of blame generally produce more rather than less fatal accidents. Human-computer interaction is a complex subject and its very possible that the accident rate of those specific CGMs could have been reduced or prevented with better design, depending on what the issue actually was.
>the possibility of it being software, hardware or QC
There's a certain overlap here. It's not completely orthogonal. Having worked on safety critical systems before a lot of effort is put into detecting hardware errors in the software. E.g. random bit flips, ALU hardware issues, RAM writability issues, hash check of the loaded software being ok, plausibility check with (partually) redundant sensors.
You can detect a lot of hardware/QC issues on the software level. While it's still a hardware issue, better software can sometimes at least detect it
I agree with you that the Fundamental Attribution Error typically wins the day. If people are making a mistake, find the systemic solution. But, it's critical to include Education as one of the potential components. If false readings are always a possibility, the alternative to expecting people to double check results is that we don't allow devices like this on the market.
>the alternative to expecting people to double check results is that we don't allow devices like this on the market.
Excellent, to avoid killing a few people a year, you've killed thousands.
If you're not a diabetic or if you have no medical experience around this kind of device, kindly butt out and mind your own business. Low blood sugar in the middle of the night is an immediately deadly condition that needs treatment or the patient can end up with brain swelling. It's also not a condition that will wake the person experiencing it up. Having a CGM blare and alarm has saved countless people and given them a far better life from better sleep, less anxiety, and not randomly dying while resting.
Every CGM comes with directions telling you to calibrate the unit often and do blood stick tests to ensure the unit is working properly. Any diabetic should also be under the care of an endocrinologist as it's a complicated and deadly disease with lots of terrible ramifications.
>If a brand of CGMs have an issue that sometimes causes false low readings
Not sometimes. "Over an extended period".
"Abbott Diabetes Care stated that certain FreeStyle Libre 3 and FreeStyle Libre 3 Plus sensors provide incorrect low glucose readings. If undetected, incorrect low glucose readings over an extended period may lead to wrong treatment decisions for people living with diabetes, such as excessive carbohydrate intake or skipping or delaying insulin doses."
Months of high blood glucose level can worsen patient's condition or if high enough even put them into hyperglycemic coma in weeks(?).
While true, you would have to ignore all other indicators for quite an extensive period of time. Like excessive urination and hypersensitivity being obvious ones. Not impossible but I have the strong sense there is more to this story than reported in the FDA disclosure.
Not true, the effect of high blood levels can be very unpredictable. Especially above a certain number - and largely depending on the basal insulin strategy of the patient, for instance long acting shots versus constant micro doses of short acting via pumps. In the latter case, an untreated high blood sugar could escalate in a matter of hours to a fatal level.
>incorrect low glucose readings over an extended period
I use the G7 and the directions say to always use a finger stick to celebrate the unit, especially at high and low
readings.
Did these people also not see and endocrinologist to get things like A1C?
Diabetes is very unforgiving as you get older or are a fragile diabetic. If they were just dependent on the CGM alone then it's likely a lot of other mismanagement was already occurring.
"A new monk arrived at the monastery. He was assigned to help the other monks in copying the old texts by hand. He noticed, however, that they were copying copies, not the original books. The new monk went to the head monk to ask him about this. He pointed out that if there were an error in the first copy, that error would be continued in all of the other copies.
The head monk said, ‘We have been copying from the copies for centuries, but you make a good point, my son.’ The head monk went down into the cellar with one of the copies to check it against the original.
Hours later, nobody had seen him, so one of the monks went downstairs to look for him. He heard a sobbing coming from the back of the cellar and found the old monk leaning over one of the original books, crying.
He asked what was wrong.
‘The word is ‘celebrate,’ not ‘celibate’!’ sobbed the head monk."
I'm not a diabetic, but even I was skeptical of the title "Seven Diabetes Patients Die Due to Undisclosed Bug"; this draws a very direct 1-to-1 association when in reality, we know that a death would be the result of multiple failures/oversights.
I thought this article would try to sell us on the benefits of formal software verification or something... Though of course, you can't formally verify complex human biology.
Agree. The linked FDA recall said the 7 deaths are "associated", which could just mean contemporaneous. This article is written by a new diabetic who doesn't seem to understand the disease very well yet, and is sensationalist in its reporting (perhaps unintentionally). They are probably opening themselves up to a defamation lawsuit here and are certainly disseminating misinformation, sowing FUD in service of an agenda, however well intentioned.
I rarely do this, but I'm flagging the article in hopes of limiting its exposure to new readers.
The OP is hardly anywhere near as sensational as the latest AI generated github something-or-another typically posted here. I found the article extremely useful and would not be aware that it effected MORE THAN ONE product line. Please don't let @dang bury this IMO. If you have an alternative URL please post it!
Agreed. This story is clearly pushing an agenda to an extreme degree. They spent a lot of time linking to different things and past stories, but the claim of having killed seven people gets almost no coverage in the story. Can we at least get a source to where they’re getting that information?
> Can we at least get a source to where they're getting that information?
Fourth paragraph of the article, first sentence, the hyperlink text says, "the US FDA announcement". The link[1] contains the following under the heading, "Reason For Early Alert":
> Abbott Diabetes Care stated that certain FreeStyle Libre 3 and FreeStyle Libre 3 Plus sensors provide incorrect low glucose readings. If undetected, incorrect low glucose readings over an extended period may lead to wrong treatment decisions for people living with diabetes, such as excessive carbohydrate intake or skipping or delaying insulin doses. These decisions may pose serious health risks, including potential injury or death, or other less serious complications.
> As of November 14, 2025, Abbott has reported 736 serious injuries, and seven deaths associated with this issue.
This is an AI-written summary of someone else's blog post, which it does not cite. This sentence is a giveaway: "The Asahi Linux blog post reveals the intricate complexity of implementing sample shading on AGX hardware." There are no other mentions of Asahi and no links.
Seems like this is someone who just started learning GPU programming and is just throwing stuff onto the internet. See, their [recent blogpost, which is just a list of unwritten vulkan tutorials](https://hyeondg.org/vulkan_tutorial/0).
The mistake you are making is letting the author choose your points of comparison, without having a high-level picture of where water usage goes. Comparing water usage to a city is misleading because cities don't use much water; large-scale water use is entirely dominated by agriculture.
I think of homebrew as a curation service; it lets me name a piece of software and install it without having to any special diligence on it. In that use case, I _want_ them to enforce code-signing requirements; that reduces the risk that some software-supply-chain compromise will spread to my computer.
I do want the ability to install unsigned software, either because I wrote/compiled it myself locally and can't be arsed with signing, or because I'm getting it from a non-public source that doesn't want to share a copy with Apple, or because it's from a developer I trust who can't be arsed. But I never want to get unsigned software _from a curation service_.
The reason this happened will probably never be revealed, but I predict it's probably because the channel was uploading videos through a VPN, and wound up sharing an IP address with someone who was using the same VPN for piracy.
Your example use case is automatically filling out an IRS form, operated by the sort of IRC department that makes a webform that's only up during business hours? Do you realize how legally risky that is to create, and how legally risky that will be to operate?
If you're automating filling out the form, you aren't reading the instructions and you aren't checking what you're putting into it as much as you should be. And if you put in incorrect information, it tends to be considered fraud, even if it's downstream of a sloppy LLM rather than downstream of a particular fraudulent scheme.
You're right, but this is where the LLMs are especially useful. Our customers all prompt it to terminate if it doesn't have the right information / the pre submission confirmation doesn't match
The captcha put you on notice that your scraping wasn't authorized. Depending on the details and circumstances, bypassing it and scraping anyways may have been a crime.
reply