HNHacker News
TopNewBestAskShowJobs

geofft

29,438 karma · joined April 10, 2012

https://ldpreload.com/

@geofft on Twitter and GitHub; also that username at that domain name for email.

I live in Brooklyn, NY, and can generally be found thinking too hard about kernels, linkers, or crypto.

'lizdenys and I make a podcast about practical personal digital security: https://looseleafsecurity.com or search for "Loose Leaf Security" on your favorite podcast app.

submissionscomments
geofft··on Stable Diffusion is a big deal
Yes, but the consequence for society is that professionalization is coming to every single field, and the blue-collar jobs are evaporating, and we had better figure out how to get all those people either white-collar jobs or some other form of income if we want to avoid significant social instability.

It used to be that you could find work as a political cartoonist and draw a picture of some satirizable politician each day - maybe Dr. Oz with a funny-looking vegetable platter, or Biden falling off a bike labeled "build back better," or something - and that would be a career. Now it seems you can just type "Dr. Oz holding a vegetable platter, political cartoon" into one of these AIs and the bulk of the work is automated. Sure, you could spend the rest of the day refining it, but nobody's really looking for perfection or transcendent skill in their political cartoons.

You can still find work making political videos of Dr. Oz and vegetables (e.g. https://twitter.com/JohnFetterman/status/1564432981841907713). Today's image generation AIs cannot do videos like that. Tomorrow's will, before we know it, and again nobody's looking for more than a baseline level of quality there either.

And even the local newspaper that might have been employing a political cartoonist is being swallowed by high-capital-ownership companies that can replace a lot of the writing with AIs. The expectations of quality are a bit higher - though the AIs have mostly gotten the hang of sports writing - but again people don't expect The Smalltown Gazette to have the standards of The New Yorker.

Sure, the highest-quality drawings, feature films, and longform journalism will get a lot better. And that's great! But most people don't work on such things. What are they going to do? If nothing else, how will they remain an audience with money to spend on the highest-quality works?

(I am not advocating for stopping this process, to be clear. Smashing the AIs isn't a coherent proposal, and smashing the physical machines of automation didn't actually work when the Luddites tried. I'm advocating for admitting that this process is happening and drying up employment, admitting that keeping as much of humanity as possible under a good standard of living is important for humanity, and figuring out what to do about it.)

geofft··on Why are D-sharp and E-flat considered to be two different notes?
Singers will do this via intuition - you don't think of, say, a perfect fifth as 2^(7/12) = 1.4983x over the root. You think of it as a particular pair of sounds that resonates well, much like when you picture "red" in your mind you're not thinking of exact HSV or Pantone values. At most, you'll think of a perfect fifth as exactly halfway between the octaves (1.5x over the root). As the sibling comment points out, this isn't the singers choosing a particular temperament for the entire song; it's them constantly tuning individual chords and intervals to each other and to their previous notes as the song goes on. The same note on paper can be several slightly-different frequencies in different parts of the song, and most singers won't even be able to tell you that they're doing that.

(This is also the same mechanism at work when an entire choir singing an unaccompanied piece goes flat without realizing it. Someone will not quite make an ascending interval, and everyone else will adjust to cover it.)

geofft··on Let's Try Gnome Boxes
I believe https://access.redhat.com/documentation/en-us/red_hat_enterp... has decent docs on what to do here:

- Run NTP in the guest

- If at all possible, run the qemu guest agent, which can inform the guest to resync when it resumes from suspend

- If available, use the "kvm-clock" clock source (but that's just a source of timing ticks, not of actual time, so it doesn't help a lot with suspend)

geofft··on Removing HTTP/2 Server Push from Chrome
But the only real way to be sure if something is a good idea, especially in complex distributed systems such as the web platform, is to try it out. And when you try it out at scale, standardization is very important. The alternative is to have everyone using new ActiveXObject("Microsoft.XMLHTTP") because you have a non-standardized thing that's rapidly turning out to be a good idea.

The purpose of a standardization process isn't to stamp approval on good ideas; it's to ensure interoperability. Server Push was a plausible idea. It very much could have been a good idea. It turned out not to be, after we tried it in the real world for many years, but that's okay. And we know one thing for sure - the reason it didn't work wasn't that people were afraid to use a nonstandard X-Chrome-Server-Push extension.

geofft··on Guessing C for Every Answer Is Now Enough to Pass the NY State Algebra Exam
I think it's okay for wild guesses to be zero expectation; if you look at it from an information-theory perspective, leaving the question blank and wildly guessing both communicate the same thing, "I don't know." What's the benefit of incentivizing test takers to leave questions blank instead of wildly guessing? I get that it feels better to discourage guessing, but how does it improve the accuracy of the test as a measure of understanding?

And if you can rule out at least one answer, that communicates that you know something about the material under test - provided, of course, the test has no joke or otherwise implausible answers, which appears to be true of the Regents test that I posted above. Take question 6, for example, which asks whether the days of the month are best described as "(1) integers (2) whole numbers (3) rational numbers (4) irrational numbers". Suppose you forgot the distinction between integers and whole numbers, but you knew that the other two answers were incorrect, and so you randomly picked one of the first two options. Doesn't that convey understanding of part of the material under test, such that a score of a positive fractional point in expectation is reasonable?

(A more logistically complicated system would be to permit students to mark any answers they think might be correct and therefore remove the "in expectation" part. If you don't select the correct answer, you get no points, but if you select n answers including the correct answer, you get 1/n points, and leaving a question blank is considered equal to selecting all the answers. But the cognitive overhead of this system probably makes it not worthwhile.)

geofft··on Guessing C for Every Answer Is Now Enough to Pass the NY State Algebra Exam
I'm genuinely confused at the criticism of the content of the questions and especially the criticism that they're "not hard". I mean, I guess they aren't algebra in the sense of "use the quadratic formula," but I'm not sure that's a useful sense. The question about percentages and the one about reading the graph are both eminently practical questions, ones that very directly respond to the algebra student's "When are we going to use this in real life" plea. I can see people needing to know this in any practical field of work - and I can realistically see people not knowing it, too. And plugging numbers into the quadratic formula isn't hard, either. It's a matter of knowing the definition, which is precisely a thing the author criticizes, and it's much more direct and mindless application of the definition than that question about whether a certain number is rational.

(And is it supposed to be hard, or a good thing if it's hard? The goal should be that everyone passes; nobody needs a weed-out ninth grade math class. The questions should be meaningful, yes, and they should show that you actually understand the subject. But this is a subject we would, ideally, want everyone to understand well.)

The actual June 2022 Algebra I Regents exam is available online and you can see what the questions look like: https://www.nysedregents.org/algebraone/622/algone62022-exam... Most of them look like conventional algebra to me. And there is indeed one use-the-quadratic-formula question!

(The criticism that a zero-bits-of-information response passes is certainly valid, and could be easily solved by subtracting fractional points for incorrect answers, a common approach on other standardized tests.)

geofft··on Fully Dockerized Linux kernel debugging environment
Docker containers don't have their own kernels. They share the kernel with the host operating system, and the host kernel handles interrupts. The Docker container is just a group of processes started in a particular way so they have a different view of the filesystem, the network, and so forth. Like any other userspace process, they get UNIX signals, but they don't get hardware interrupts, which are handled by the kernel.

For this project, in order to have a kernel to debug, they run a separate kernel inside the qemu virtual machine emulator. qemu is a normal userspace process, and therefore that process can run inside a Docker container. So the kernel being debugged has interrupts, but the "hardware" generating those interrupts (and all the other "hardware," including the CPU itself) is all just software emulating a PC.

geofft··on Incident report: Employee and customer account compromise
I don't think "more security training" will get us anywhere - we've been trying it for a long while. This is a technical problem and admits technical solutions. The most obvious, simple, and straightforward solution here is logins with security keys (WebAuthn/FIDO), which cannot be phished. Perhaps Twilio fell for this precisely because they don't want to admit to themselves that the SMS verification product they sell (https://www.twilio.com/verify) is an obsolete joke compared to security keys.

If you don't want to use that, for some reason, or if you want additional protection, other solutions include using dedicated machines (not personal cell phones and certainly not whatever devices the attacker was using) ideally with hardware-locked keys (TPMs etc.) to access the corporate network or at least to access sensitive systems like customer data, or giving people separate privileged accounts that they don't use for day-to-day access, or establishing a two-person rule for logging into sensitive systems (so two people with the same access need to get successfully phished at the same time), or setting up some real-time auditing of changes (e.g., a Slack channel gets notified when people manually log into prod).

geofft··on KDE Connect Now Available on iOS
Looks like it is based on the Mosh waiver https://github.com/mobile-shell/mosh/blob/master/COPYING.iOS, which went in about ten years ago. Blink Shell (which is of course also GPL'd) uses the Mosh code and is available on the App Store.
geofft··on We should have a $25 minimum wage
But automation gets more efficient over time. Twenty years ago, an automaton that could drive you around city streets required a whole university research lab and barely worked. Today, it's very realistically possible. Twenty years from now, it's hard to imagine it being more expensive than a taxi driver. Twenty years ago, we had drip coffee makers and early Keurig machines. Today, we have latte art printers. Twenty years from now, vending machines that produce Instagram-worthy iced lattes with whipped cream and caramel drizzles will likely be commonplace.

If your minimum wage plan is taking automation into account, your plan would have to be to slowly lower minimum wage over time so that, as robots get more efficient, humans are always paid a little less than them. That would be a coherent policy - for a few years. But eventually that leaves you with a minimum wage under a penny. What's your plan then?

geofft··on We should have a $25 minimum wage
> The truth is that when you cut people out of the market you will replace people with machines and concentrate wealth in the hands of the machine owners.

Is it not economically rational to replace people with machines and concentrate wealth in the hands of the machine owners regardless of minimum wage policy?

Like, the Luddites didn't get mad during an increase in minimum wage - there was no minimum wage at the time, at all! It would be about a hundred years until England established one. They got mad during an increase in industrialization and mechanization. Marx didn't talk about ownership of the means of production, i.e., machines and who owns them, during an increase in minimum wage. This has been happening for centuries, and from a quick look around this very forum, there seems like no sign of automation slowing down or even settling at some baseline level of expense where $X/hour workers are worth paying but $Y/hour are not for any $X > 0.

There are only two long-term sustainable ways to solve this problem (and it is a problem!). One is to hope for, and work towards, technological improvements where each stage just brings more work to be done. There were more serious fears of computer programmers automating themselves out of a job a few decades ago than there were last year when GitHub Copilot was released, and that's certainly not because Copilot is unimpressive. We're just confident there's plenty more code to write and plenty more problems to solve. The other is to hope for and work towards a world where technological progress means we are post-scarcity and we can afford to pay people the same inflation-adjusted daily wages for increasingly small amounts of labor and eventually none at all.

Any plan that says that we need to maintain a low minimum wage for the sake of keeping workers employed is one that very soon will end up compromising that low value all the way to $0/hour.

geofft··on We should have a $25 minimum wage
There's two reasons this argument doesn't work.

One - some 15-20% of Americans have a zero or negative net worth, which straightforwardly inverts the argument. Let's say minimum wage is $1 and the family owes $100 in credit card debt that they are not paying off. They must perform 100 hours of labor beyond their current production to pay off their debts. Now some brilliant economists move minimum wage to $100/hr. Their debts are wiped, as the banks didn't automatically increase their balance to $10,000.

Two - it's not regressive. It's a tax on everyone, proportional to how little labor you do. A defining characteristic of the poor, compared to the rich, is that the poor make their money from their labor and the rich make their money from their existing money ("savings" / "passive income" / "investment" / "401(k)" / etc.). Suppose, in your example, your brilliant economists move minimum wage to $100/hr. Immediately, one hour later, their bank account has twice as much money! Sure, they have suffered a temporary loss of purchasing power, but after 100 more hours of labor they've regained it. Meanwhile, some other family in the same economy has saved $1,000,000 in a bank account, representing some five hundred person-years of full-time work at $1/hr, perhaps inherited over the centuries, perhaps gained from business. Their savings are also rather wiped, leaving them with roughly 1% of what they previously had - but regaining it will take them another five hundred person-years of full-time work.

So it is certainly a tax, but it is very progressive. Really these are the same mathematical effect. The more money you have, the more it affects you; the less money you have, the less it affects you; and if you have negative money it even affects you positively.

geofft··on Ask HN: Why is there so much snark on HN?
Snark, I think, correlates positively with terseness, and terseness correlates positively with upvotes - it's much easier for someone to read a short comment and say "Yes, that's it" than a very long one. So there's both a prevalence of those comments higher on articles and an effective pressure to make those comments.

(There are other things that correlate positively with terseness, like outright insults, but also strongly correlate negatively with upvotes. Or there are things like pun threads, which are popular on Reddit etc. but are just culturally unpopular here, for some reason.)

geofft··on Richard Stallman – The state of the Free Software movement
> But you have to keep in mind that RMS advocates the free software philosophy - the complete left end of the software spectrum

I don't think this is true! There are notable portions of the free software community that are farther left (or whatever direction you may consider it to be) than Stallman and GNU.

Debian, for instance, rejects the GFDL as a non-free license when the "invariant sections" or "cover texts" mechanisms are used. These are clauses that allow GFDL documents to include words or even entire chapters that cannot be modified or removed from copies of the document. Stallman uses this clause to include a section about the ethics of the free software movement in documents like the GCC manual. Now, explaining the ethics of the free software movement is of course an important cause. But the manner in which this is done - preventing the GCC manual, or portions of it, from being freely modified in its entirety, from being recombined freely with other documents or code with otherwise-compatible licenses - is, in the view of Debian and others, incompatible with the free software philosophy.

This is a farther / more radical view. Stallman's view is that making some unmodifiable and unremovable sections in an otherwise-free work is an acceptable compromise. Debian's view is that there is no room for compromise. And Debian is an entire (and quite influential) Linux distribution.

Or take the AGPL, for instance, sponsored by now-defunct startup Affero. There are many people (myself included) who believe the AGPL is a restriction on use, freedom 0, because it prevents you from using AGPL software in certain contexts where the requirements cannot be fulfilled. Suppose I find some AGPL code with a neat algorithm for a session cache, or something, and I want to incorporate that into a TLS library. There is absolutely no way to fulfill the AGPL's requirement to make source code available to users who interact with my code over the network. The clause triggers, because users are indeed interacting over the network, but since there's no TLS extension to offer source, I can't actually do so. This is, it seems to me, as much of a restriction on use as "You can't use this in commercial settings" or "You can't use this for military applications." There's a reason that even GNU still uses the GPL for much of its software - they know the AGPL would make lots of their code effectively unusable in contexts where it ought to be usable.

Or, in the left-libertarian direction, consider the folks who want to abolish copyright entirely. The GPL fundamentally relies on copyright law, else it cannot have its particular effect over the MIT/BSD/etc. licenses where it requires redistributing source along with binaries. The GPLv2 itself starts by saying, "To protect your rights, we need to make restrictions" - there is a coherent political view that this is contradictory. There is certainly a popular view, popular among the BSDs and others, that those licenses are freer than the GNU ones. But there is also the farther and more radical view that even the MIT and BSD licenses are a compromise, and the goal should be questioning the validity of software copyrights the same way much of the free software community questions the validity of software patents.

(In fact, among non-experts, you can find a very common view that because something is released to the public in certain ways it ought to be "public domain." The Berne rule of every creative work being automatically copyrighted isn't clearly obvious to the general public, and it's a relatively recent rule, too!)

Stallman is influential, yes, and very passionate. But that's entirely different from him being all the way at the end of the range of discourse. It is worth listening to folks whose positions are less well-known but more radical.

geofft··on RDP Clients for Debian: An incomplete review
There's also FreeRDP (xfreerdp/wlfreerdp) and rdesktop. Looks like all of the ones in the article (Vinagre, Remmina, Boxes) use libfreerdp as a backend.

I was very happy with xfreerdp on Debian 10, but something on Debian 11 is making it crash/disconnect from my work computer every so often. Haven't yet figured out what it is.

geofft··on Ask HN: Could Ukrainian embassies facilitate US hackers to join the cyber front?
https://twitter.com/lizthegrey/status/1497843649668739072 is a short thread on basically this subject, but no clear answer. 18 USC 958 makes illegal "within the jurisdiction" of the US to accept "a commission" to fight against any state "with whom the United States is at peace."

(As the thread points out, public perception and the willingness of prosecutors to go after you are probably more relevant from a mechanics/plausibility standpoint.)

geofft··on Ask HN: What do I need to learn to be useful as a hacker to defend my country?
If you're asking in the general case, as opposed to "How would I respond to this call" - my opinion is that the most practical and effective thing you can do is to help secure code used by individuals - web browsers, the Linux kernel, image decoders, etc. I think there are a number of reasons for this:

1) There is a real asymmetry between offense and defense in (nominal) peacetime. Helping the defense generally helps the "good guys" much more than the "bad guys" (at least in my worldview). If you look through attempted attribution of 0day attacks in the wild, what you generally see is repressive governments attacking individuals - watering-hole attacks on news sites, targeted malware against lawyers and journalists, etc. Cases of attacks on governments (like the 2012 MD5 collision that took down an Iranian nuclear reactor) seem to be rarer, and in particular, cases that rely on bugs in mass-market software (as opposed to supply-chain attacks or DoSes or very targeted attacks) are rarer, and attacks from less powerful / "hacker underground" groups towards governments are rarer still.

2) Structurally, it makes more sense. Cyberattacks aren't like physical attacks. When ammunition hits a target, there's a very classical-mechanics effect of the energy of the weapon versus the strength of the structure or shielding. Offense, at a very high level, is about more and stronger weapons, and defense is about withstanding or escaping attacks. Software, naturally, doesn't work that way. It's more mathematical; either the attack works, and is potentially completely compromising, or it doesn't. If you can make a system that robustly parses input without bugs (or sandboxes the parsing, or whatever), there is no cyber-weapon that can get past it.

3) You can make a real impact. A huge number of the systems that ordinarily people use are open-source software projects that accept contributions. (And note that this includes a whole lot of security-sensitive code in system that are not open-source as a whole product - for instance, most of the attack surface on iOS is in WebKit, image parsers, or the xnu kernel.) A lot more is available free-of-charge and accepts security reports. And there is, unfortunately, a lot of relatively low-hanging fruit.

Pick something you're interested in, go look at recent exploitable CVEs, and do some reading on how the exploits work and how they might be systematically prevented. A little bit of your time spent making it easier to systematically prevent exploits has a real long-term benefit on the world.

As a good historical example - most database libraries around 10-20 years ago made it most natural to construct database queries by appending strings together, which made SQL injections entirely too common. Since then, there's been a combination of a push for libraries to make it easier to do parametrized queries, a cultural / documentation push to get programmers to be aware of this, and a move towards database abstractions like ORMs that avoided the problem entirely.

Someone who wrote some docs 10 years ago about these libraries probably helped hundreds of annoyed enterprise programmers get their system built in the right way when their boss was yelling at them about deadlines, and may well have prevented millions of people losing their data in a breach.

When you think about attacks on secure messengers, etc., it's not hard to imagine that the same amount of effort could save countless lives just a few years down the line.

I think memory safety is one of the highest-impact changes we could make in development that would help the cause of defense, and there's a lot of work to be done. Most of it, mind you, is not merely showing up places and saying "I'll rewrite this in Rust" - it's helping people be able to integrate incremental rewrites and ship things in new programming languages, or perhaps helping them avoid memory-safety problems in their existing programming languages. Chrome has a good article on this https://www.chromium.org/Home/chromium-security/memory-safet... , and "fuzzing," the technique of throwing generated inputs at libraries to see where they crash and then fixing those crashes, is also highly valuable.

But there are a whole lot of other similar classes of problems to help with, too. The Linux Kernel Self-Protection Project https://kernsec.org/wiki/index.php/Kernel_Self_Protection_Pr... is working on eliminating classes of bugs in the upstream kernel. Some of these have already been addressed, to some extent, in security-focused forks of the kernel, but getting the fixes into the standard kernel is important for getting them in everyone's hands and also a good way to learn about things.

geofft··on No minimum parking requirements? No problem for Fayetteville, Arkansas
That part is getting a lot better with e-bikes, which is part of why I mentioned them. :) I'm in NYC, where the temperature does vary at least that much between summer and winter, and I recently got an e-bike in preparation for our office reopening, because I wanted to be able to avoid the subway in more weather than I'd be happy using an ordinary bike for (especially while the office's showers are closed!). On warm days, I can easily get up the bridge without sweating if I let the pedal assist do its thing and don't try to go super fast. And on cold days, if I do the same thing, I don't have the problem where I start to sweat under warm clothes - I can wear exactly what I'd wear without a bike (a warm jacket, decent inner layers, and gloves, nothing more complicated than that) and it works fine, too. It's a lot harder if it's more than a few degrees below freezing, extremely hot, or heavily raining/snowing, but I think it's pretty manageable most days.

But also, that's one of many reasons that bike infrastructure isn't enough to displace car infrastructure on its own. Good public transit helps, apart from being a viable year-round transit option in its own right, helps here because it allows people to decide that they can bike on days with good weather, knowing they have fallback options the rest of the time, whatever their own comfort level with with weather may be. And frankly taxis/ride-hailing services can also solve the problem of getting people from place to place without requiring a huge amount of parking spaces.

geofft··on No minimum parking requirements? No problem for Fayetteville, Arkansas
Not disagreeing, really, but - habit, culture, perception, and infrastructure. Minimum parking requirements, and more generally streets designed around cars, are a form of infrastructure. I'm not saying it's a good form of infrastructure, but it is one. Local government needs to invest in alternatives: pervasive sidewalks, well-designed protected bike lanes so that people who are interested in biking but worried about being hit - which is over 50% of adults in US metro regions, see https://jenniferdill.net/types-of-cyclists/ - actually decide to bike, legal use of e-bikes, intersection designs that don't prioritize high-speed cars, reliable and frequent and ideally 24-hour public transit with dense stops (so probably buses and bus lanes), and so forth.

And yes, to be fair, habit and culture and perception play into this too. Cities build this infrastructure when people demand it and can convincingly argue that it will get used. Potential bus stops are more useful when there isn't a parking lot separating the bus route and the store people want to get to. Useful sidewalks are a natural consequence of commercial districts designed to be accessed by foot/bike/transit instead of by car, and so forth. But it does take both of these.

geofft··on Jujutsu – A Git-compatible DVCS that is both simple and powerful
It's doing something weirder than that, but "it applies a diff between a specified commit and its parent on top of your current work" is more accurate/intuitive than "it does a three-way merge with a common ancestor". No common ancestor is involved.

The first hint that it's doing something interesting is that the implementation of cherry-pick is in revert.c, because it's implemented as a variant of revert: https://github.com/git/git/blob/v2.35.1/builtin/revert.c#L23...

Both the "revert" and "cherry-pick" operations initialize a sequencer object with a single operation in the sequence. (This is the same mechanism underlying "git rebase -i").

From there we can find the call to sequencer_pick_revisions, which leads us to the sequencer implementation, where there's a fast-path for ordinary cherry-picks leading us to a short function single_pick, which in turn calls do_pick_commit: https://github.com/git/git/blob/v2.35.1/sequencer.c#L2066

This operation then does the following:

- Determine what we're applying a diff on top of: https://github.com/git/git/blob/v2.35.1/sequencer.c#L2083-L2...

- Find a parent commit. If the commit has more than one parent, it requires you to specify a "-m" option indicating which parent to diff against. https://github.com/git/git/blob/v2.35.1/sequencer.c#L2109-L2...

- Set up the commit message and some other stuff.

- Set the variable "base" to the parent it found and "next" to the commit being cherry-picked. https://github.com/git/git/blob/v2.35.1/sequencer.c#L2187-L2...

- Assuming no merge strategy (or either of the standard strategies "ort" or "recursive") was specified on the command line, call do_recursive_merge. https://github.com/git/git/blob/v2.35.1/sequencer.c#L2237-L2...

In turn, do_recursive_merge calculates a head_tree from the current HEAD (https://github.com/git/git/blob/v2.35.1/sequencer.c#L645), and then calls either merge_incore_nonrecursive (the current default, from the new "ort" merge strategy) or merge_trees (from the older "recursive" merge strategy) with three trees (i.e., with no further ancestry information): the base is the parent it found, and the two sides being merged are your current HEAD and the (tree of) the commit being cherry-picked.

That is to say, at no point does it care whether the two commits even have a common ancestor! It's just doing an operation on trees. It is doing a three-way merge, yes, but the graph of the merge it's doing is one that potentially doesn't actually exist in reality.

Or, in other words, it's trying to compute the tree that could be equally well described as the result of

- applying the diff of the commit you're cherry-picking to your current HEAD

- applying the diff between your current HEAD and the parent of the commit you're cherry-picking to the commit you're cherry-picking

So this is more powerful than purely applying a diff with no information about the base of the diff, but it is very much like applying a diff.

One way you can test this without drilling into source code is to make two independent commit histories (using either two git repos, or git checkout --orphan, or whatever) where a commit in one history has a diff that would apply to a file in the other history if applied with the "patch" command. Then try cherry-picking it into the second history. It should work.

(In the revert case, it more or less does the same thing, just with "base" and "next" flipped. That is, if you have commits A, B, and C, and you want to revert B, it does a three-way merge where the base is the tree after B, one side is the tree after A, and the other side is the tree after C!)

geofft··on Ask HN: Free Countries
Oh I totally missed that this was from Cato, got it :)
geofft··on All baptisms performed by Phoenix priest invalid because he changed one word
I think that's precisely what the very idea of the sacraments is! Full disclosure, I'm a Lutheran, and so I obviously don't fully agree with Rome's take on things, but in this very specific bit, I do agree this is how the means of grace work.

Even in the Lutheran service, where we have a norm of corporate confession in the middle of the service instead of individual confessions (and prescribed acts of contrition) beforehand, the preacher says to us, "As a called and ordained minister of the church of Jesus Christ, and by his authority, I therefore declare to you the entire forgiveness of all your sins." This phrasing is clearly not meant to mean just "Because I am a called and ordained minister, I can tell you for sure that Jesus already died for your sins." It is meant to be an actual thing that happens, through the authority given by God to the clergy - i.e., God is obliged, because God has chosen to grant that authority.

Here's my denomination's take on sacraments in particular: https://download.elca.org/ELCA%20Resource%20Repository/What_...

> Historically a sacrament was viewed as "an outward and visible sign of an inward and spiritual grace given to us, ordained by Christ himself, as a means whereby we receive the same, and a pledge to assure us thereof" (Anglican Catechism). Lutherans have traditionally tried for greater clarity by stating that a sacrament is an act that is commanded by Christ, uses a material or earthly element, and through connection with the Word is the bearer of God’s promise, as the definition above indicates.

I think I see why it feels sort of "unfair" that God is obligated to do something when we invoke a sacrament - but we're not the people imposing that obligation. God did! It's a "pledge to assure us [of grace]" / "promise" in the wording above, and it's not unfair to make use of that pledge / promise. Going back to the analogy - if I'm root and I give you a sudoers file, you're not imposing on me by running sudo. Yes, you can always ask me to do something and have me form an opinion in the moment about whether I should, but for the things in the sudoers file, you can also just run sudo, I'm not going to be unhappy about it. That's why I put it there!

It seems to me that the alternative position is that every baptism and communion is nothing more than a prayer that God might choose to do something - and might just as easily choose not to do so. To me, that seems harder to believe: you'd have to say that God gave us the sacraments even though there's no particular point in them beyond just regular prayer. Why should they exist, then?

(Also, there's a connection here to the ancient argument of the Donatists, that priests and bishops who apostatized in the face of Diocletan's persecution lost their "powers," so to speak. That view was declared heretical, and the view that won out, advocated by Augustine and others, is that sacraments derive their power from God's grace and the simple fact that they are done, not by the merit of the person trying to do them.)

geofft··on All baptisms performed by Phoenix priest invalid because he changed one word
CCC 1257 says, "God has bound salvation to the sacrament of Baptism, but he himself is not bound by his sacraments." I think I've seen the hierarchy quote this a few times re this story.

I believe that Rome's position here is that they have been entrusted with specific sacraments as tools / "means of grace," and it's not their place to tinker with them, but that doesn't imply that anyone is automatically consigned anywhere as a result of such tinkering. It's entirely reasonable to believe that God is extending grace in a non-standard way because God agrees that this does not make sense. But the fact remains that it is still a non-standard way and not actually the use of the sacraments as given to the Church, and the Church doesn't have the power to decide how much tinkering is too much. It just gets the sacraments the way the sacraments are.

A bad attempt at a nerdy analogy here is to a sudo rule. An unprivileged user who doesn't have "ALL" sudo access must run specific commands in the way they're phrased in the sudoers file. If the sudoers file says you can run "sudo shutdown -h now," and you type "sudo shutdown --halt now", it won't work. But that doesn't mean that the root account is unable to run "shutdown --halt now" or that root doesn't think the "--halt" version is a good idea. It certainly doesn't mean you're unable to contact whoever is root and say "Hey, please run shutdown --halt now," or that they'll ignore you if you do. If root runs it, it will work, and have the same effect. It's just not what's in the sudoers file.

geofft··on Google Release Chrome OS Flex
Google is a business - they are neither a kindly old wizard doing things for the good of the people, nor are they a game show wheel, randomly picking a product to kill today.

You can pretty accurately model what they will and won't do with this one unusual insight that they're a business.

For instance: are they going to kill Chrome? No, Chrome has given them market dominance as well as a seat at the table to define web standards, improving and shaping the web platform both to be a more appealing product to developers and users, so that they can make more money when they stick ads on it, and to more easily support sticking ads on it. Or, briefly, Chrome - though it's not a paid product per se - makes them huge amounts of money.

What about (from that website) Google Currents? It's the rebranded, corporate-only version of Google+. Its biggest user was almost certainly Google itself, and it was a venue for employee activism and therefore was a threat to corporate profit. Briefly, though it was a paid product, they had few paying users and it cost them money.

A sibling comment mentions the end of legacy free G Suite (formerly GAFYD) accounts for very small users. This was, by definition, a thing that didn't make them money and a thing that was hardly an on-ramp for subscriptions that did make them money. Again, you could predict this.

So what abut Chrome OS Flex? It remains to be seen - it's not as clearly worthwhile as Chrome or even Chrome OS itself. But Google does absolutely make money on Chrome OS Enterprise, which this product is clearly linked with. It's nowhere in the same boat as things they provided out of the kindness of their corporate-personhood heart like legacy free G Suite or Google Reader.

It might end up being a failed experiment, certainly. But "It's a Google product" is a weak signal about whether it's going to get killed, and there are many stronger signals. Let me know when they kill Ads and Cloud.

geofft··on Ask HN: Free Countries
I'm genuinely curious - this is the first I've heard of those countries being referred to as "libertarian." What makes them libertarian?

Among other things, they all have substantial taxes (in the 25-40% of take-home income range) and use it to fund substantial civic infrastructure / state-run welfare. My understanding is that libertarianism is generally sympathetic to the position that taxation is theft, and libertarianism is incompatible with large governments and large social safety nets. But it seems like I'm missing something.

(I usually see these countries being described as social-democratic - is that incorrect?)

geofft··on Ask HN: Free Countries
The word "freedom" gets used in so many ways by so many partisan folks. Can you give concrete examples of what you mean?

For instance, even within the USA, there are certain people who find more "personal freedom" in rural, sparsely-populated areas, and certain people who find more "personal freedom" in a high-rise apartment building in a big city where nobody makes eye contact with their neighbors. Which one is more up your alley?

geofft··on An update on AirTag and unwanted tracking
Right, or even consider the original case where you have an AirTag on your keys for if you lose them - given a name and a phone number, someone who finds your keys can quite possibly figure out where you live and use the keys.

Most things you lose are both less dangerous lost and worth less to a criminal (and hence less likely to induce someone to become a criminal) if they can't be tied back to you. Don't put "If found, return to 123 Main Street" on your keys, don't put "Apple Confidential" on your USB keys, don't engrave "Property of North West" on her Switch, etc.

geofft··on Please, stop pushing for static typing in Python
Asyncio is weird. But check out Trio, an alternative and easier-to-understand async framework: https://trio.readthedocs.io/en/stable/tutorial.html

Trio solves a problem that threading does not: switching between tasks only happens at deterministic points (they don't run in parallel), and tasks are in a hierarchy so that exceptions are handled in reasonable ways (and you can do a lot of other associated cool things like cancel an entire tree of tasks).

Python has a GIL, so you're not getting parallel Python code anyway without multiprocessing.

This means that if I'm writing code that implements, like, a basic network server, by far the easiest way to do it is Trio. Unlike with threading (in Python or any other language) I don't have to worry about locks from concurrent access / tasks switching at unexpected points, and unlike with e.g. the select module, handling both directions of communication and multiple connections is doable with normal straight-line code instead of weird state machines.

It's not needed, yes, but neither is Python itself. You could just write C.

geofft··on Tell HN: Google doesn't work anymore for exact matches
There's a "verbatim" option under search tools for that.

I agree with you that that's what I expect from quotes, but... I really think there are way more Google users who do not expect that. A lot of people think quotation marks are the way you do emphasis in English, for instance!

geofft··on The Pandemic Boom Goes Bust
But most of that stimulus didn't end up in the hands of consumers.

People weren't purchasing Pelotons with $600 stimulus checks. The sort of person who has nothing better to spend an extra $600 on than a Peloton is the sort of person who wasn't eligible for the stimulus in the first place. (I do fully agree that stay-at-home orders motivated some purchases, though - but again, the sort of people who could actually stay at home and take meetings from their home office instead of driving around town delivering Pelotons were the sort of people who already had the cash to spare for these things, anyway.)

That stimulus ended up in the hands of business owners, which effectively caused a partition: because business owners, as a group, had more money, they could all raise prices and afford it. But the working class, as a group, did not have noticeably more money.

In other words, there was a sharp rise in inflation - for the owning class. The effect was the same as if there were two neighboring countries, and one experiences inflation. That country's cost of living goes up, but so do their wages and the valuations of the things they own. So they're fine, and effectively they can buy things from the second country for even cheaper in terms of purchasing power. But people in the other country find it harder to afford to buy things from the first country. They're the ones most hurt by inflation.

← PreviousPage 2 of 34Next →