ddevault

36 ms·

361.
by ddevault 4y ago
Copilot can certainly produce an output which appears "transformative". It can also produce verbatim copies of licensed works. The output exists along a spectrum from probably transformative to clearly non-transformative, and it o
362.
GitHub Copilot and open source laundering(drewdevault.com)
87 points by ddevault 4y ago|25 comments
363.
by ddevault 4y ago
Remember that Hare programs ordinarily have only a small number of dependencies, countable on one or two hands. If you need a newer version of a specific one, you can send a patch for your distribution updating the package when you need it,
364.
by ddevault 4y ago
This, in my opinion, is a trait of a poor engineer, and falls outside of our target audience. Strong engineers are serious about the entire system, including the OS it runs on, and treats it as an engineering target no different from any ot
365.
by ddevault 4y ago
Use a distribution whose release cadence matches your needs. I use Alpine Linux, which has frequent updates on edge and biannual stable releases, which allows me to enjoy both approaches. Debian is designed for the longer term, Arch is desi
366.
by ddevault 4y ago
We're well aware of the relationship between borrow checking and generics in Rust and, though it may rule it out for Hare, I do not think that we need generics or anything like it to obtain most of the value-add of a borrow checker.
367.
by ddevault 4y ago
No. The article details that Hare's package manager is your system package manager, e.g. aptitude. Such package managers have independent signatures, features against partial updates, reproducible builds, and independent accountability
368.
by ddevault 4y ago
I did not want to fill up an entire article about the garbage collector problem, but rather just wanted to use it as a brief example to illustrate the thinking around trade-offs when it comes to features like it. It might make sense to writ
369.
by ddevault 4y ago
We were discussing encouraging syntax highlighters to add something like this: https://l.sr.ht/Q_D-.png
370.
by ddevault 4y ago
Reference counting is still pretty high level, given that you have to find somewhere to put the count. It's not really suitable for kernels or some embedded applications, for instance. That makes it less desirable as a language feature
371.
by ddevault 4y ago
A lot of languages have been accused of "ignoring decades of PL research", an almost word-for-word criticism I've heard of a dozen languages or more. In fact, Hare includes a lot of innovations from newer PL research, but ind
372.
by ddevault 4y ago
Thanks!
373.
by ddevault 4y ago
I sincerely hope that this does not happen, but it's a matter of shaping the culture appropriately. We explain our motivations for not having one and re-enforce that through the community and hope that this prevents people from buildin
374.
by ddevault 4y ago
>I wonder what the circumvention is for uninitialized memory, which is important for large data structures in high performance code. Actually there's not really much of a means around this. You could use static memory, which is only
375.
Safety features of the Hare programming language(harelang.org)
62 points by ddevault 4y ago|41 comments
376.
by ddevault 4y ago
I have worked with Bazel, in a large, multi-language organization, but most of these points are solving the wrong problem . Often I argue for questioning the problems rather than taking them at face value and building an unnecessarily comp
377.
by ddevault 4y ago
Helios currently uses legacy boot. The first thing is to get something to boot and print out "hello world" to serial or some such like. I also wanted to get into long mode fast, because Hare does not support 32-bit targets. Then
378.
by ddevault 4y ago
The high-level OS will be more like Plan 9 than Unix, but appreciably distinct from either.
379.
by ddevault 4y ago
I have already spent over 3 years on Hare and it has nearly 50 people working on it. It is also much simpler in design and smaller in scope than other new languages. Furthermore, building a kernel with it helps to validate the language desi
380.
by ddevault 4y ago
The name of the complete system is called Ares, and Helios is a small component of that. Essentially, to most people, it's an implementation detail/internal codename that will not really see much use for marketing efforts. It does
381.
by ddevault 4y ago
As I implied in my answer above, it's the use-cases themselves that are wrong. Helios does message passing completely differently.
382.
by ddevault 4y ago
>[Bazel and gRPC] are not "questionable tools" in the sense that they solve the wrong problems I disagree, and there is a huge philosophical gulf of understanding between my view and those who feel differently.
383.
by ddevault 4y ago
The actual API differences between Helios and seL4 are somewhat subtle, rather than a broader difference of design philosophy. For instance, the layout of the bootinfo structure is different - PML4/PDPT/PD/PT objects are stor
384.
by ddevault 4y ago
seL4 has an interesting API design but I'm not fond of the internals or implementation. It's largely an academic exercise, in my opinion. They also have no ambitions in userspace, and limit themselves to working on the kernel alon
385.
by ddevault 4y ago
Aye, this is the answer. https://drewdevault.com/2022/03/14/It-takes-a-village.html
386.
by ddevault 4y ago
Helios author here. I tried to think of a nice spin for this comment, but came up short. If there is a future of computing with Fuschia in it, it will have everything to do with Google's industry weight and nothing to do with its good
387.
by ddevault 4y ago
>Does io::handle maybe exist only as an optimization for sendfile()? io::file exists for where it's needed to interface with the host system. io::stream is needed for userspace streams. io::handle is needed so that code which just w
388.
by ddevault 4y ago
It depends on your specific needs and is difficult to characterize in hypothetical. I imagine that it would probably involve Hare's native tagged unions, though.
389.
by ddevault 4y ago
Take a look at this comment where I explained why a first-class hash map is not a good fit for Hare: https://news.ycombinator.com/item?id=31258759
390.
by ddevault 4y ago
>You were able to "get away" with implementing a pretty bare bone thing, but you called it out as something that should be viable in general. I can see how you had this impression from the language of the blog post, but what I

More ›