L
> The first camp had more proponent, so it won, but its hard to convince the second camp because they disagree on the meaning of the feature itself. Do both variants of contract have the same syntax? If the contract is declared along with the signature, then you just have flags to say callee/caller/both/disabled. If it's callee, the flags are checked at runtime on ingestion. If it's caller, the flags are checked as far as possible by the compiler, bit like safe Rust presumably does, otherwise…
L
I agree this does seem a low-hanging fruit. But it's not that big a deal because it's the dependencies in Cargo, the tool that will fetch external dependencies, not the C equivalent of #include for which there was already a warning. Also, like said in the release notes, it should check all the build target, so the C equivalent of checking all the #ifdef, so you can't just compile one time and check if the symbols are used.
L
Yes, the global village is a bit of a numbers game. But there is an other reason: Meeting people in person, having a short chat with them, already gives a lot of context (ha, that has a bit of a different meaning these days, doesn't it). Even just seeing the person helps to prevent being as direct to say something you might still do behind a keyboard. But it can also swing the other way: trust in traditional media has been lost, especially in the US. And trust in online media you didn't know…
L
> First, the cond may or may not be executed. Seems like a weird choice: always enforcing them seems like a better outcome. This is probably for performance reasons .. The proposal is based on the idea to add redundant code to catch errors in the implementation. As soon as the conditions are guaranteed to execute, they become part of the implementation and can not be used to independently test the implementation anymore. Think of it like unit testing: Nobody stops you to include your test suite…
L
From the outside (I'm not a Rust developer), it looks like a low-hanging fruit was taken with “reporting of unused dependencies in Cargo” (even though things are never as easy as one might expect). Talking about unused deps and dead code removal: I'm still a C developer, and the C language has many tools, but I don't know a tool that works for shared libraries (with unstable API) with code scattered in multiple repositories (for a monorepo that's easier). Do you know the existence of such tool?…
L
I make that title with love - I used MATE as soon as Fedora adopted it (v17? v18?), to my move to Arch and up until the end of 2024. The thing that finally drove me away was getting a *good*, cohesive dark mode working for a desktop (carrying my own theme mods and everything, still just "eh"). After having to use an Ubuntu distro at work (Mint w/Cinnamon) and seeing how good a dark theme could actually look, I set up Xfce and was greeted with a beautiful, default out of the box dark mode. We're…
L
> workers are choosing those jobs because they presumably don't have better options, and adopting policies that make those jobs inaccessible or cease to exist would not benefit those workers. Many mines are under the control of armed state and non-state actors that use involuntary labour with the workers, often children, physically abused and prevented from leaving. Problem is especially rampant in Congo, which supplies 70% of the world's cobalt. Myanmar is also very bad, but sadly forgotten…
L
just wanted to get mad Please do not attribute to malice what can adequately be explained by stupidity. Thank you.
L
I agree, but manually writing async code is a pain, unless you have a language that actually supports it well. Then use that... From an application perspective this is certainly not the way to go to save a few kilobytes of kernel stack, if it means massive amounts of manual async state machines. I agree, but the numbers don't add up: Android devices, for example, may host 2-3,000 system threads, and thousands more for running apps, adding up to 5,000 threads on an idle device 2-3000 system…
L
UVC streams from In Terminals to Out Terminals via a DAG that can include processing nodes; the specification gives an LCD screen or a USB endpoint as examples of Out Terminals, and a USB endpoint or a camera as examples of In Terminals. The only reason it's commonly used for input devices like webcams is that Microsoft made it hard for video input device manufacturers to supply their own drivers while simultaneously providing a high quality UVC capture driver as part of the OS; device…
L
There are many projects with BSD/MIT licences that are successful and alive and some of them (like LLVM) have even significantly eroded their GPL competition. Specifically, in LLVM's case, this is precisely because many companies have their internal forks and closed versions of LLVM and only give back the patches which doesn't affect their position regarding to their closed source versions, and can work with others' patches. In other words, LLVM is one of the *ium projects (e.g. VSCode /…
L
This seems likely to lead to another project regressing to the mean, like Debian did.
L
>In GPL projects, those changes also have to be made available in source code form to both the users of the software and to the original source code project. I don't think the second part is true. Yes GPL requires providing source code to those who receive binaries but it does not, according to my understanding, require providing the code to the original project. Of course it also disallows forbidding recipients of the binary and source from providing the source to others. So if a commercial…
L
...don't you loose the users that are willing to try something because they are somewhat familiar with already? Not the OP, but I'll disagree. I'm a long time Debian user like you, but it doesn't prevent me to install something completely unfamiliar and give it a fair go. I have found many nice tools, things, technologies like that. As an example, we have significant number of servers running RHEL family of distros, amongst many things which are not Debian based, or not used as a part of a…
L
> Neither license prevents rebranding (in my opinion it's a real weakness of free software licenses). Both require minimal attribution in the form of a copyright notice, but not more. Neither prevents aggressive monetization. Modern Free Software licences now try to extend their grasp beyond copyright, and imho that is a BIG mistake. Neither patents, nor branding, are anything to do with copyright. And blocking re-branding would pretty much kill Free Software! If you can't rebrand then either…
L
> Do note that pretty much all desktops also have co-processors (Intel ME and AMD PSP). BTW these are only the tip of the iceberg, there are many more. Search the Internet for "ME analyzer" and then look for "engines". But even better: take a look at /lib/firmware/ - hidden in plain sight! Funny how most people know this yet miss it. All these matter a lot from a security standpoint, not just the ME/PSP. > Other than usually signed firmware, is there any way to even persistently choose which…
L
This is basically M:N threading - and we don't have a good sense of what kernel interfaces are needed to make M:N threading capable of beating 1:1 threading on latency (they do beat it on throughput, which is why I use userspace scheduled options, but the latency is, AFAICT, still an unsolved issue).
L
> Well, Kubuntu 26.10 (which was released after Fedora 44) ships KDE 6.7.5, hard to understand why Fedora 44 was forced to upgrade to KDE 6.8 on this background. ...Because Kubuntu isn't Fedora, and makes different support lifecycle commitments? Does Kubuntu commit to doing security/etc backports during the ~6 months [1] that Kubuntu 26.10 is "supported" but upstream KDE is not? At best, sticking to KDE 6.7 would have bought Fedora 44 another couple of months on 6.7, at which point it would…
L
But, If you are deviating from desktop too much don't you loose the users that are willing to try something because they are somewhat familiar with already? For example I am using Debian in my servers because I was already for years a Debian user in the desktop. I understand that I am minority but then again Redhat has both Fedora and their RHEL. Would it make sense to only have RHEL with no way their potential users to get familiarized with their offerings?
L
> they make some goodies that they may help them on their jobs and they don't want their employers to steal these goodies and make them property of their employers… but then GPL people come and steal them, instead! These two definitions of "steal" are... very flawed. > but still: they have written the terms, these terms allow GPL-relicensing, if someone is doing that — they shouldn't complain. Exactly.
L
As a sysadmin who manages fleets, I don't agree that reproducible system installation and update in package based systems is impossible. I have been doing it via xCAT (install) and SaltStack (update/manage) for a long time. You don't need these exact tools, but they are useful to us due to size of our fleets here. If you want extremely strict, SaltStack even has CF-Engine like drift-protection features yet I don't use them, because a system you don't touch doesn't drift. Given a fixed state,…
L
But they want other super-rich corporations doing the same when they switch jobs! Essentially: they make some goodies that they may help them on their jobs and they don't want their employers to steal these goodies and make them property of their employers… but then GPL people come and steal them, instead! That's not nice, from their POV and I can understand why they feel that way… but still: they have written the terms, these terms allow GPL-relicensing, if someone is doing that — they…
L
> To be blunt -- KDE would have been dropped from Fedora had this policy exception not been granted. Well, Kubuntu 26.10 (which was released after Fedora 44) ships KDE 6.7.5, hard to understand why Fedora 44 was forced to upgrade to KDE 6.8 on this background.
L
> An image-based distribution wouldn't have helped because the images need to be installed (and activated) as well. Concur. Even with base image + app overlay based fleet management, you will never have uniform systems. Fleet diversity will scale at least logarithmically with size of fleet. You need to be able to live with diversity, because you're never going to have just 1 image deployed.
L
> BOTH licenses require you to disclose whose code you are using because you have to attribute copyright to them. So your ability to monetize can be held in check by the "consumers" ability to compare what you are selling with the "free" alternative. No, MIT license do not require to disclose your code, it only mandates to preserve attribution and the legal notice. But if you change just the 10% of the code and leave the remaining 90% unchanged then you can keep all the source private as long…
L
> Does anyone still care? Hibernation lasts longer, and it's more secure.
L
In that case they should double license to avoid other super rich corporations doing the same :)
L
It is unfortunate that these license changes and even discussions devolve into how we feel about things instead of sticking to the facts. Asahi does not like the competion. How does the choice of license factor in? Does MIT allow forking without attribution? No, not it does not. You have to say that you are using code based on the MIT license. The requirement to attribute copyright reveals the original authors and the existance of the original project code. Does it allow you to "monetize…
L
> So only GPL projects end up being "not nice" while super-rich corporations doing the same thing are OK? Just one more contradiction I guess. Not contradiction, different motivation. Many people release under BSD-like so they can use their code at work while employed by a super-rich corporation.
L
The people creating the heavily publicised unattributed forks compete for attention and financing, that’s how that harms the project. Really one should not be naïve, some of the biggest proponents of permissive licensing earn their lives selling “coding” which is just lifting of publicly available permissive code, sparing no energy to convince employer or customer to use their services instead of hanging their wagons on the public FOSS project they took the code from. That’s where “militant…