Google Merges Android & Chrome OS, AI Slop Invades FOSS  ## Metadata - Author: [[Linux & Open Source News]] - Full Title: Google Merges Android & Chrome OS, AI Slop Invades FOSS - Category: #podcasts - URL: https://share.snipd.com/episode/341d0a48-1e24-43db-b8fc-a14df8bd2d64 ## Document Notes ## Highlights - **Google to Merge Android and Chrome OS** - Google plans to merge Android and Chrome OS into a single unified platform. - This could mean laptops running Android with a better multitasking desktop interface in the future. Transcript: Nick So Google's president of Android at Google gave a little interview to TechRadar and within it he slipped a little nugget of information saying that Chrome OS and Android were going To be combined into a single platform. They're also saying that they are very interested in seeing how people use their laptops because, well, they don't really have real laptops these days. They have Chromebooks, which don't really work like full-blown laptops. We already knew that Chrome OS was going to move away from a real full-blown Linux system, getting its underpinnings and libraries from Android instead, but this basically confirms That the two systems are going to merge. It's not just Chrome OS grabbing a few elements from Android but still existing as its own platform. Now they're going to be the exact same thing, so probably you're're gonna buy laptops running Android. Google is also trying to implement a better multi-window desktop multitasking interface into Android and the first Android 16 beta shows some of that. There's likely another component around that. Google might be forced to sell the Chrome browser at some point if the current US-based antitrust issues go further. And since Chrome OS is basically just Google Chrome running on top of a Linux base, this could have all kinds of implications for Chrome OS's existence. By replacing all of that base with Android, they only have one system to manage, and if Chrome has to go away and be sold, chances are Google can just strike a deal to have it pre-installed By default, or they could even just build their own web browser anew. Contribute to Linux for Android, for servers, but also on the drivers front for their Chromebooks. They would likely still need to do that, fortunately even by replacing Chrome OS with a desktop version of Android, because Android would still need drivers for these laptops or these Desktops. So chances are this won't really change ([Time 0:02:00](https://share.snipd.com/snip/206cfc96-597c-4324-9a83-fd7fa7efb14e)) - **AI Slop Floods Curl Bug Reports** - AI-generated bug reports, especially for Curl, are overwhelming maintainers with useless submissions. - This flood threatens bug triage quality and may end bug bounty programs. Transcript: Nick Interface. Now, if we needed another proof that unchecked AI is just a net negative for the world, projects are now being overwhelmed by AI-generated slops. Specifically, Curl, for example, is now getting tons and tons of crappy bug reports that are just AI-generated. And what I'm saying tons and tons, it's tons and tons relative to Curl's usual bug reports. It's not like they're getting hundreds every day. The project's lead developer simply cannot deal with that anymore. He says the general trend has been more and more AI slop. About 20% of bug submissions are now clearly AI generated and properly useless. This obviously drowns out all the real reports and it makes triage and identifying what's a real issue and what should be worked on a lot harder. This might even lead to the bug bounty program for curl ending. This is something that rewards submissions that identify security vulnerabilities. It has already paid $90,000 since 2019 and it's helped identifying some flaws that Curl then fixed. Technically, anyone submitting a bug report to that bounty program should disclose the use of AI, but of course they don't do that, either because they don't read the disclosure agreement Or they would just spam the bug bounty program using AI and so they don't want to tell that they're using AI because they are just terrible people. The project lead said that you're generally just better off avoiding AI as a whole, which I entirely agree with. Each submission apparently takes from 30 minutes to three hours to be looked at and each submission is looked at by three or four reviewers out of the seven who work on the project so that Is a lot of wasted time so even if these aren't hundreds of ai slop reports per month it is still taxing on people and the worst thing is that the project lead says that a lot of these bug reporters Think that they really are helpful by submitting their AI-generated bug report, meaning that these people do not know anything about what they're trying to fix or the project. They're not even evil. They just really believe AI is capable of identifying giant flaws, and so they're just spamming projects with these bug reports trying to help. Let's face facts. Generative AI has just made the internet a way worse place. It's just pure garbage. It's misinformation, it's scams, it's spams, and a few legitimate use cases are drowned out by all the problematic ones. Even if it's just for your own enjoyment, the energy use itself just makes it completely unethical to use it in today's current world. So hopefully this freaking trend goes away. These tools just stop being used because people finally realize it's just a ([Time 0:05:47](https://share.snipd.com/snip/2bc22197-5520-4dca-be6f-896e129c014a)) - **Avoid AI-Generated Slop in OSS** - Avoid relying heavily on AI-generated content in open source projects due to inefficiency and misinformation. - Use filters or tools to combat the flood of AI-generated useless submissions. Transcript: Nick Of trash. But I'm not too confident in that. In the meantime, we're gonna have to find solutions or filters or tools to get rid of all that crap. And I don't know if that's even feasible anymore. Now the GNOME Foundation is firing on all cylinders since they elected their new executive director. They've now been joined by PostmarketOS, maybe the best Linux on phones effort that we currently have. They've joined the GNOME Advisory Board, meaning maybe we'll see more effort on that front. Funding is also a big area the foundation is looking into, obviously, with a new donation page in the works to fund the work of specific contributors, letting people get a bird's eye View of what needs funding, who funds what, but also letting contributors apply for grants, for fellowships, and for jobs. They're also following the KDE route, trying to build partnerships with hardware makers like Framework or Slimbook to make sure that GNOME works really well on these devices out of The box and maybe at some point showcasing the future GNOME OS as a pre-installed option, once at least it is a real option for a daily driver. Now, all this transparent communication and ([Time 0:08:42](https://share.snipd.com/snip/e38faef3-bed7-49de-929c-d499e1f71eaf)) - **GNOME Foundation Firing on All Cylinders** - GNOME Foundation is becoming more transparent and proactive under new leadership. - They focus on funding, partnerships, and potential GNOME OS pre-installations. Transcript: Nick New executive director. They've now been joined by PostmarketOS, maybe the best Linux on phones effort that we currently have. They've joined the GNOME Advisory Board, meaning maybe we'll see more effort on that front. Funding is also a big area the foundation is looking into, obviously, with a new donation page in the works to fund the work of specific contributors, letting people get a bird's eye View of what needs funding, who funds what, but also letting contributors apply for grants, for fellowships, and for jobs. They're also following the KDE route, trying to build partnerships with hardware makers like Framework or Slimbook to make sure that GNOME works really well on these devices out of The box and maybe at some point showcasing the future GNOME OS as a pre-installed option, once at least it is a real option for a daily driver. Now, all this transparent communication and the efforts that they're making to try and make GNOME viable are, I think, a really good indication that the foundation is now in competent And good hands, so props to the new executive director, looks like he's doing a good job, although the GNOME Foundation does have a long road ahead to actually become sustainable itself And then be able to invest in GNOME's developers. ([Time 0:08:58](https://share.snipd.com/snip/ff001e64-44df-46a8-bb1c-19c60dc8da07)) - **GNOME Enhances Parental Controls** - Expect modernized and integrated parental controls in GNOME with features like screen time and web filtering. - This will aid Linux users with children in managing device use more effectively. Transcript: Nick Parental controls section of GNOME, something I personally don't have an interest in because my only son is really hairy and he's actually a cat, but I'm sure that this will help a lot Of people who have kids. The goal is to redesign the parental controls UI to make it work like other parts of Gnome, so basically modernizing, libidwaita-ing this part of Gnome. But they also want to have these controls integrated in the lock screen as well, and they want to add new features like screen time monitoring, screen time limits, bedtime schedules, And web filtering. All features that should probably be in any basic parental control software, so having that on GNOME directly is probably really good because it really did not do anything if it didn't Have that, did it? Work has been contracted, and the UI refresh is already completed, and it might even become part of GNOME 49 in September, although it's not certain yet. ([Time 0:10:28](https://share.snipd.com/snip/26243b8f-0f3c-4ccf-9d55-51550e4797c7)) - **PHP Proposes License Change** - PHP plans to switch to a BSD license to unify licensing and improve GPL compatibility. - This change simplifies integration without changing user or contributor rights. Transcript: Nick Ones the php license for all the base language and the zend engine license for all the Zend engine code, as they wanted to make sure you could use any of those two parts separately. But as both projects got more and more interconnected, this dual licensing scheme proved problematic, notably because it could make PHP incompatible with the likes of the GPL. So there is a new proposal to update that PHP license and that Zend license to instead adopt the three-clause BSD license for both parts of PHP, starting with PHP 9. This would be an OSI-endorsed license, so it should be compatible with other free software and open source licenses, and it has minimal restrictions to use, package, modify, and redistribute The code. The change is being discussed currently. It will not change any of the rights granted by contributors or granted to users, so it doesn't seem like a bad idea on paper, although as always, license changes are controversial. The proposal has a nice detailed historical breakdown for why that dual licensing scheme was adopted and why the person behind the proposal thinks that this needs to change. PHP doesn't need permission from all of its contributors to change that license, but they want to do things properly so they are holding a vote and they are giving people six months of Discussion before holding that vote. Which is probably the right way of handling this. Just because you don't have to ask for permission to change something doesn't mean you shouldn't consult the contributors and the community. So this is being done in the right way. And in the end, it probably won't change anything for PHP developers and users. It's just going to simplify how you can integrate PHP in other projects, so probably a good thing. ([Time 0:12:06](https://share.snipd.com/snip/4dc9a883-b6cb-47cf-b0ff-387cbcfa2525)) - **Nick's Haiku OS Experience** - Nick tried Haiku OS a few times in a VM and liked its nostalgic feel. - He views it as a modern take on the classic BeOS with distinct features. Transcript: Nick BOS was back in the day, but also to bring it into the modern computing world. Haiku got a nice update this week, getting better Wi-Fi drivers off of the OpenBSD code base, but also improvements to high DPI scaling for the windows, for mouse cursors, for icons, Also better support for FAT file systems. The UI rendering code has also been simplified, which should lead to better performance. Dark mode support should be nicer with an easier way to pick the colors that you want to use for the various controls, which also improves custom color schemes. The package manager can also now install software without refreshing the repos cache if you want, which should speed up the installs on your system. Workspace management was also improved, and there are plenty of fixes to the kernel and the underlying libraries, which I must remind you is not the Linux kernel. It's something that they built themselves off of various other parts. Now, I never tried Haiku on the channel specifically. I ran it a few times in a VM just for fun, but I never really stuck with it after a few hours of trying things out. I do like how it looks. I ([Time 0:14:06](https://share.snipd.com/snip/7535aa89-0eaa-4dce-a8f2-e384bc95b860)) - **NVK Nvidia Driver Progress Update** - NVK open-source Nvidia drivers currently achieve less than half the performance of official drivers. - Despite low performance, NVK can run benchmarks and games, showing promising progress. Transcript: Nick We can do about this. Now, as the NVK open source drivers for Nvidia are now part of Mesa and they allow Nvidia GPUs to run demanding stuff without using the official drivers, 4onyx did a little benchmark Of the official drivers versus the NVK plus Nuvo stack and it looks like things aren't quite up to par just yet. The TLDR is that NVK plus Nuvo is currently giving you on average 45 to 47% of the performance of the official drivers on modern NVIDIA GPUs, which is, in my opinion, just not good enough To ever envision ditching the official modules. NVK does outperform the NVIDIA drivers in a few cases, in Vulkan benchmarks, but in most cases, whether you're using OpenGL Vulkan, NVK is not even close. In games, the performance is way lower as well, with stuff like Arkham Knight or Counter-Strike 2 getting about a third of the frames with NVK, and other games barely reaching half. Now, obviously, this is not criticism for NVK. They started work on it, I think, two years ago, and they're also still reliant on Nuvo, which is a very old kernel driver that just doesn't work well with recent NVIDIA GPUs. They do need to do a lot of optimization, but it's actually quite impressive that in two years they managed to get a driver that can actually run benchmarks and games and run them properly. Sure, the performance is not up to snuff, but it ([Time 0:15:38](https://share.snipd.com/snip/572a8737-833f-4a33-9e50-7f317a5454f5)) - **Linux Mint 22.2 Beta Highlights** - Linux Mint 22.2 beta will include fingerprint authentication and better integration of Libadwaita apps. - Wayland support improves but won't be default yet; moderation tightened due to community issues. Transcript: Nick Now Linux Mint 22.2 should get its beta in August, which means the stable release should come around September, since Mint betas generally last for about a month, if I remember correctly. This one should come with the hardware enablement stack for the LTS kernel, with also support for fingerprint authentication throughout the system, including the terminal. There's also updates to the default themes to closely match the colors that Odwaita can display. And they also want to better integrate third-party apps in Mint, notably Libidwaita apps, with support for accent colors and their brand new Libidwaita fork to make sure that Libidwaita Apps use the right window controls and the right colors. Wayland support is also improved in there, although it is not fully ready yet and it probably will not be the default either. I think they plan to make it the default for the Linux Mint version that will be based on ubuntu 26.04 lts so maybe next year at best lmde7 will follow mint 22.2 using debian 13 as its base Instead of ubuntu and it will support oem installs something that lmde6 doesn't do the main team also said that they unfortunately had to tighten moderation on their blog as apparently Things were getting out of hand. As they say, they are not here to fight for any ideology, everyone is welcome, and I quote the Mint team here, there is no such thing as freedom of speech. And well, yeah, the internet is a pretty crappy place right now if you have a comment section under anything you create. I can absolutely 100% testify that. And I ([Time 0:17:37](https://share.snipd.com/snip/d372aee1-c8b1-4ec2-a926-cf3227e0e3ff)) - **Google's Haptic Touchpad Proposal** - Google proposes a standard for haptic touchpads on Linux focusing on vibration feedback and force thresholds. - This aims to reduce vendor code duplication and improve app interaction possibilities. Transcript: Nick On Linux. They offered a design proposal to change how the HID drivers that handle these input devices work, and also a change to how the userspace drivers work as well. Now this only affects touchpads that do not click physically, like the ones you would find on modern Macs, not your regular diveboard touchpads that most laptops still use. The goal of their proposal is to have a standard protocol to allow userspace apps to talk to these haptic touchpads and also to reduce duplication of code between vendors. This would allow apps to handle haptic feedback, like vibrating the touchpad when needed, but also potentially letting these apps act on certain presses with maybe certain force Thresholds, like for example, force-clicking a word to get a definition or force clicking a link to display a preview of the web page behind that link. Why this and why now? Well, there are apparently multiple projects to develop these touchpads outside of what Apple is currently doing. I would assume Google is working on one as well. And so to avoid multiple vendors developing multiple implementations, like what we've seen in the early days of touchpads on Linux, Google is trying to take the lead on it at the kernel Level. Now, when I first used haptic touchpads on the first MacBook Pros that used them, I found them trash. They just did not replicate a real click. You could clearly feel that it was vibrating. The extra features stacked on them never really were all that interesting, but on the latest Macs that I tried, I think it's now suitable. I still kind of prefer the regular old diveboard touchpad. I don't quite see the need for the haptic ones, unless maybe they're like more resilient over time. Not sure about that. But hey, if they are going to happen on multiple other computers, then we might as ([Time 0:20:04](https://share.snipd.com/snip/2570f187-52e9-4f71-a792-ac393242a685)) - **Firefox Adds WebGPU Support** - Firefox 141 adds WebGPU support on Windows, allowing web apps to access GPU directly. - This enhances web graphics performance but raises security concerns due to expanded browser attack surface. Transcript: Nick Now, Firefox 141 is getting web GPU support. About two years after Chromium-based browsers implemented that same feature. Web GPU is a way to let web apps or just websites access the GPU directly, meaning it provides much better performance than stuff like WebGL. It is a W3C standard and currently only Chrome and Chromium-based browsers support it, Firefox just added it, and Safari has yet to implement it, probably because they have to build The entire layer that talks to their own graphics API instead of using something that is a multi-API. Unfortunately, WebGPU is Windows only for now on Firefox. It should come to Linux and macOS in a few months because it is built using WGPU which provides a way to just link that web GPU to any graphical API whether it's Vulkan, DirectX or Metal And WGPU is an open source library that Mozilla is a major contributor to. There's also a lot of work left to be done on this thing for Firefox even on Windows to improve the performance, to reduce the latency and to add a few methods, notably to use it for video Decoding. Now I know that these features tend to annoy some people because you're basically turning your web browser into an entire operating system with direct hardware access which does increase The attack surface for websites basically and for your web browser and on your system. So not necessarily a great thing always, but as soon as Chromium-based browsers have it, this means that 90% of the web browsers used in the world have it, and so you can't really not have It if you're an alternative web browser, because if you don't, you're going to get incompatible with a lot of websites really quickly. ([Time 0:22:03](https://share.snipd.com/snip/68fbc40a-7f55-4773-8fdd-a4ef9a25c436)) - **Court Greenlights Anthropic AI Lawsuit** - A US judge allowed authors to pursue a class action lawsuit against Anthropic for copyright infringement. - The lawsuit alleges pirated books were used to train AI, risking billions in damages. Transcript: Nick Won't be solved for years. This time it's against Anthropic as three authors sued them for copyright infringement, alleging that the AI company used pirated book libraries to train their model. A US judge said that this can be turned into a class action lawsuit on behalf of all authors whose work would have been downloaded from those same pirate libraries back in 2021 and 2022. The judge says that Anthropic may have illegally downloaded up to 7 million books, which would potentially expose them to billions of dollars in damages, which I think could probably Bankrupt that company, because even though they probably have billions of dollars in investor money, they have to burn that money to actually work because they don't generate any Cash right now. Of course, Anthropic is looking to challenge that ruling, but the arguments, as always with AI companies, are extremely flimsy. They're not saying, oh, we didn't do it. They're saying it's hard to prove copyright infringement on such a massive scale. But as always, AI companies' arguments for why they don't pay creators and authors are always along the lines of, if we had to pay, it would cost us millions, which, yeah, it's not like They have any ground to stand on anyway. Some use cases for AI are fair use, and the judge apparently in that case ruled that Anthropic did do something along the lines of fair use with those books, but using pirated copies is Still illegal, even if you use them for something that is fair use. So this is why this can be turned into a class action. Hopefully this puts an end to this madness that ([Time 0:23:55](https://share.snipd.com/snip/07a6e425-fc09-4a44-b07c-92e79cf54be5)) - **x86 Emulator Advances for RISC-V** - FeelX86 enables running x86 apps and games on RISC-V Linux, including Steam games. - This emulator bridges software availability gaps for emerging RISC-V architecture. Transcript: Nick Project. This lets you run x86 apps on RISC-V on Linux, and this now includes Steam and some big games like Crysis or The Witcher 2 and 3. Games that use Steam DRM should also work now. It is early days, performance might be terrible, but it is a nice proof of concept. Felix86 is also working on better system integration with file system emulation using a separate executable, and they're also working on letting users run the entire emulator a bit More easily through an emulated bash. Much like ARM emulation, if it progresses well enough and it becomes a stable way to access x86 software on RISC-V devices, it will likely end up being packaged and pre-configured by Distributions like what Fedora did with FEX for example. And sure, RISC-V is not extremely popular right now. They don't really have a big flagship product like you would have for ARM with their recent Macs, but it's still getting a lot of attention. We're seeing a lot of projects adopt it and start trying to provide boards with it and try to port Linux to it. But it's the usual chicken and egg problem. If you can't run your usual software, you're never going to buy it. And if no one uses it, no one is compiling or building the software for that architecture. So having the emulator bridges that gap for the initial years or decades where the architecture is starting to take off. And then you can actually start porting the software to it. And you also ([Time 0:25:57](https://share.snipd.com/snip/5c0f4503-ffbb-4dbb-9379-36bd70f9361a))