stkdump

I would argue LWW is the opposite of a merge. It is better to immediately know at the time of writing that there is a conflict. CRDTs either solve or (in this case) don't solve a problem that doesn't really exist, especially for package managers.

Sorry, I am turned off by the CRDT in there. It immediately smells of overengineering to me. Not that I believe git is a better database. But why not just SQL?

I think this is speaking in the absense of sufficient protections like laws. Because how does it help that a hypothetical law protects you if that law doesn't exist?

I am surprised that Taiwan and China are not mentioned in the article or in these comments. Given the threat that China will get to Taiwan and capture control over the unrivaled TSMC, investing into silicon while we can is pure and simple de-risking in a de-globalizing world.

The meaning of the term "opt-in" is that it is off by default and has to be manually enabled. "opt-out" means it is on by default and you have to manually turn it off. "opt-in-by-default" or "opted in by default" are needlessly confusing.

Well I would say that the machine should not override the human input. But if the machine makes up the plans in the first place, then why should it not be allowed to change the plans? I think that the hilarious part in modifying tests to make them work without understanding why they fail is that it probably happens due to training from humans.

It's the kind of thing you do with macros in C++.

Far more common than what exactly? Because if you look at the range of the exponent, you should still come to the conclusion that floats can represent more non-integer numbers than integer numbers.

Rebuilding from scratch also takes longer than installing a prebuilt package. So while it might be worth it for a heavily used application, in general I doubt it.

Also I think in earlier days the argument to build was so you can optimize the application for the specific capabilities of your system like the supported SIMD instruction set or similar. I think nowadays that is much less of a factor. Instead it would probably be better to do things like that on a package or distribution level (i.e. have one binary distribution package prebuilt by the distribution for different CPU capabilities).

> authentication, validation and security

Those are generally solved using SSL, no need for centralized storage.

Just that none of the parsing methods I learned are actually used commonly in most real compilers.

I find it exceedingly likely that they lose money on every kindle fire sold (at least during one of their regular sales), unless it causes more content sales. So the solution is of course to make it as useless as possible for anything else, otherwise they would just lose more. Same as inkjet printers.

Same here. My ISP recently did a promo to try out 1G/1G for free for a few months. I decided not to buy it after the free trial and went back to my old 500/200 line instead of paying 40% more. Yeah, it takes a minute longer downloading the latest LLM from huggingface, so what.

> I'm told similar economics are why you can't buy laptops with cellular modems in them

I don't know what you mean. My current laptop (Lenovo L13) has a cellular modem that I don't need. And I am certainly a cost conscious buyer. It's also not the first time that this happened as well.

Interestingly page 2 of the file doesn't seem to indicate a crisis. All the numbers a roughly stable, some are actually improving (for example less people struggling with housing costs in Germany and generally fewer people struggling in France). What sticks out is that more Germans expect a recession around the corner.

Also I thank that moderating a popular forum is dang hard.

This logic makes absolutely zero sense. If a house is uninsurable, people will choose to live there without insurance. But if the house is insurable for a high cost people will not? They can still choose to not buy the expensive insurable and be in the same boat as inunsurable home owners.

> how to do it has been well known for more than 3 quarters of century

Don't leave me hanging! How to do it?

Mixtral and Deepseek use MOE. Most others don't.

While reading the definitions I was imagining this flat male comic with gender jokes "monosaturated, am I right?" When I came to the quotes I was kind of relieved to see women representing this experience.

I think evolutionary speaking, it is neither that humans are by nature strictly monogamous nor polyamorous. I think what people are confused about is that evolution often benefits from conflicts of interest between individuals. It also can both benefit from societal norms and from deviations from them. It can benefit from honesty and from deception. It can benefit from people doing their best to figure out and live by what their personal nature is and by people trying to overcome their nature to become something they think is better than that.

I bought an old used desktop computer, a used 3090, and upgraded the power supply, all for around 900€. Didn't assemble it all yet. But it will be able to comfortably run 30B parameter models with 30-40 T/s. The M4 Max can do ~10 T/s, which is not great once you really want to rely on it for your productivity.

Yes, it is not "local" as I will have to use the internet when not at home. But it will also not drain the battery very quickly when using it, which I suspect would happen to a Macbook Pro running such models. Also 70B models are out of reach of my setup, but I think they are painfully slow on Mac hardware.

Agreed. The workaround (arrays of strings) isn't great as it means an extra transformation has to be done between the reader and the usage. I would go so far as to say this is more important than comments.

It isn't. Targeted ads are much more valuable, to the point that fully random ads have almost no value at all. Btw. you can tell. If you regularly delete cookies you get much fewer ads. They like to say that if you don't allow/keep cookies, you get the same amount of ads, just less relevant. But that just isn't the case.

If browsers had the functionality to only preserve cookies that are needed to stay logged in across sessions, that would be quite powerful. That is all the cookies I need.

The thing is that without migration, demographics are extremely predictable for a very long time. People that don't exist can't have kids. They can't be coerced to have kids. And they can't have grandchildren either.

Even if you start coercing people to have kids: that will be much harder than enforcing the one-child policy. And it will have a slow effect either way, because it is starting from such a low base.

Correction: if you want to know if your compiler is correct, you look at the ISO standard. But even as a compiler writer, the ISO standard is not exhaustive. For example the ISO standard doesn't define stuff like include directories, static or dynamic linking, etc.

I don't think this is true. The target audience of the ISO standard is the implementers of compilers and other tools around the language. Even the people involved in creating it make that clear by publishing other material like the core guidelines, conference talks, books, online articles, etc., which are targeted to the users of the language.

There are also people who write COBOL for a living. What you say is not relevant at all for 99.99% of C++ code written today. Also, all compilers can be configured to be non-standard compliant in many different ways, the classic example being -fno-exceptions. Nobody says all kinds of using a standardized language must be standard conformant.

You are aware that D and rust and all the other languages this is being compared to don't even have an ISO standard, right?

I mean practically speaking in C++ we have (it just hasn't made it to the standard):

1. char 8 bit

2. short 16 bit

3. int 32 bit

4. long long 64 bit

5. arithmetic is 2s complement

6. IEEE floating point (float is 32, double is 64 bit)

Along with other stuff like little endian, etc.

Some people just mistakenly think they can't rely on such stuff, because it isn't in the standard. But they forget that having an ISO standard comes on top of what most other languages have, which rely solely on the documentation.

Can you elaborate on this? It is my understanding that electors are not allowed to change the vote based on anything like being more informed or something similar. Is this wrong? Or did the mechanics of the election change so much over time?