roca

84 ms·

871.
by roca 6y ago
That's a fair point. I do think it's appropriate to point out that build times are a problem for a lot of projects and may affect the kernel so careful attention needs to be paid to that issue.
872.
by roca 6y ago
We have a 170K line project written in Rust ( https://pernos.co ). I've written a lot of C++ in the past, but also Java and other things (I've been programming for nearly 40 years now), so I'm familiar in developing
873.
by roca 6y ago
This is good but it doesn't work well enough in my experience. I still spend a lot of time waiting for dependencies to build before the crates that depend on them.
874.
by roca 6y ago
That is a problem. There is another major problem, which is that rustc/cargo do not do a good job of compiling crates in parallel when one depends on the other. There may well be other problems too.
875.
by roca 6y ago
Rust has great support for modules (crates in Rust lingo). It can compile modules in parallel when they don't depend on each other. The problem is that when you have module A depending on module B, rustc doesn't do a good job of c
876.
by roca 6y ago
If you want to convince people that Rust should be thrown away in favour of mbeddr or something else, you need to make an argument based on specific design flaws of Rust, not just tar it by association with C++. That argument would have to
877.
by roca 6y ago
Nick Nethercote and others have done a lot of work with traditional profiling and optimization like that. They've done a great job, but for my project the fundamental problem seems to be limited parallelism, which I think requires more
878.
by roca 6y ago
Right on both counts.
879.
by roca 6y ago
There are a lot of people out there who can do this kind of work. I know a lot of such people. Many of them don't have traditional CVs. Yes, most of them are employed. Some are under-employed. Some are well employed but might be keen t
880.
by roca 6y ago
It's not my idea. This is well-known. As for who pays for it: that's a tricky issue! But if we get into a situation where (for example) 10,000 developers spend an hour a day waiting for builds, we know 50 developers could fix the
881.
by roca 6y ago
Depends on what you mean by "research". If you mean "problems with a high degree of novelty" then quite a lot of research, I think. Rust is unusual: it requires a lot of tricky type inference and static analysis from th
882.
by roca 6y ago
That might be true for drivers. Outside the kernel, I don't think it's going to be true for Android or Windows.
883.
by roca 6y ago
Requiring certain classes of drivers, for example, to be written in Rust could be great for maintainers. The person who submits the code has to do more work up front to satisfy the Rust compiler that there are no ownership issues, data ra
884.
by roca 6y ago
Rust in Linux will be fantastic except for compile time. Rust (and the world) needs a Manhattan Project to build a fast Rust compiler (where by "fast" I mean both efficient and scalably parallel when compiling a workspace with m
885.
by roca 6y ago
One argument made for unit testing which this post doesn't address is that it's easier to understand and debug unit test failures. To which I say: yes it is, but quality debugging tools (which don't exist in many domains, and
886.
by roca 6y ago
OK, that's a fair point. I have had the impression that the age effect on sickness and transmission is much less than on mortality, but I admit I don't remember any specific data on that.
887.
by roca 6y ago
As to your final question(s): I think the key differences are: developing a national strategy to defeat the virus, not live with it; clearly communicating that strategy; and sticking to it. Countries with SARS experience were quick at devel
888.
by roca 6y ago
That's right. That's where I live. Auckland is not super dense, but then neither is LA and look at that now.
889.
by roca 6y ago
FWIW Vietnam's youthful demographics are pretty much irrelevant; what's impressive about Vietnam is not its mortality rate but that they shut down viral transmission. My point is that many countries with wildly different condition
890.
by roca 6y ago
Sure, there are many other countries that likewise lack a clear strategy. But it's not good to grade on a curve here. Americans should be demanding a clear strategy, both from the states and from the federal government; they mostly are
891.
by roca 6y ago
Now explain New Zealand. Lots of very different countries have beaten the virus in different ways. Apologists for the failed US response have to resort to special pleading: "it's not fair to compare the USA to South Korea ... or T
892.
by roca 6y ago
US states do not have a clearly articulated strategy. A clear strategy would be something like "apply mitigations and lockdowns as lightly as possible while keeping R below 1 so the number of active cases continually declines, until we
893.
by roca 6y ago
Australia and New Zealand aren't good comparisons for testing rates because they have negligible community transmission. When no-one's infected with COVID-19, people aren't going to seek tests as much, and a higher testing ra
894.
by roca 6y ago
NZ ditto, except just about all our COVID deaths were in rest homes. Not all elderly are in rest homes, either. I don't know of anywhere with widescale community spread that has been really successful at protecting the elderly.
895.
by roca 6y ago
We aren't economically isolated from the rest of the world. We're still exporting and importing. Exports in some industries are higher than ever. We're still bringing in workers in certain key roles. NZers can leave (though
896.
by roca 6y ago
You can constrain the features used by an HTML document using CSP headers. For example setting "img-src 'none'; style-src 'none'" will disable images and CSS styling. So this comment is wrong, basically.
897.
by roca 6y ago
Maybe the end of anonymous ungated social media, yes. We can still talk to humans and sources we know and trust. Probably that's what we need to do already.
898.
by roca 6y ago
If this explanation is true, it suggests a very important experiment would be to develop AI models that train on video and (much less than 500 billion words of) text and then tackle these test problems that GPT-3 is evaluated on.
899.
by roca 6y ago
Where are those weights encoded? It doesn't seem like the genome has enough bits to encode a lot of weights. I agree with other people in this thread that structure/architecture is encoded, but not weights.
900.
by roca 6y ago
That makes sense but it kind of falls into my "brain has a better architecture" possible explanation. What's interesting is that the thrust of the GPT approaches is "architecture doesn't matter while we can just k

More ›