GNOME's 5 year plan, Windows Recall is a nightmare, Qualcomm beats Apple?
Linux & Open Source News
Microsoft's AI PCs and EU's AI Regulations
Exploring Microsoft's AI PCs with features like Windows Recall for finding past information through screenshots raises privacy concerns, while the EU enacts strict regulations imposing transparency obligations on companies developing AI systems, restricting biometric surveillance, and banning AI for social scoring.
👏 Support the show here: 👏
Patreon supporters get the daily Linux & Open Source audio show!
https://www.patreon.com/thelinuxexperiment
https://www.paypal.com/paypalme/thelinuxexp
https://liberapay.com/TheLinuxExperiment/
👕 Buy TLE Merch: 👕
https://the-linux-experiment.creator-spring.com/
📹 Watch Linux videos: 📹
https://www.youtube.com/thelinuxexperiment
🎙️ Leave your feedback here: 🎙️
https://podcast.thelinuxexp.com
00:00 Intro
01:23 GNOME’s 5 year plan
https://foundation.gnome.org/strategicplan/
07:08 Qualcomm’s ARM CPUs beat Apple’s M3
10:12 Windows Recall looks like an absolute nightmare
15:00 EU passes AI legislation
17:28 Ubuntu 24.10 will focus on polish
https://discourse.ubuntu.com/t/ubuntu-desktop-s-24-10-dev-cycle-the-roadmap/45120
19:55 Ubuntu doesn’t like GNOME’s point releases
https://www.phoronix.com/news/GNOME-MRE-Over-Mutter-Shell
24:32 Plasma makes big progress
https://pointieststick.com/2024/05/17/this-week-in-kde-all-about-those-apps/ https://notmart.org/blog/2024/05/new-plasma-edit-mode-in-6-1/ https://kde.org/announcements/gear/24.05.0/
30:56 Is the “frozen kernel” model broken?
https://ciq.com/whitepaper/vendor-kernels-bugs-stability/
35:12 Firefox has a roadmap
https://connect.mozilla.org/t5/discussions/here-s-what-we-re-working-on-in-firefox/td-p/57694
39:19 Gaming: Open source, self-hostable Stadia, Nvidia driver update
https://github.com/netrisdotme/netris
https://www.nvidia.com/download/driverResults.aspx/224751/en-us/
43:11 Outro
Transcript
Note: this is auto-generated by Davinci Resolve, it might have errors, badly worded things, I’m not responsible if anything is not representative of what I actually said :D
Hey everyone and welcome back to this edition of the Linux and open source news podcast. I’m your host Nick and this is the show where we discuss everything related to Linux, open source, the open web, privacy and everything around those spaces. So in today’s episode we have GNOME or at least the GNOME Foundation announcing a five year plan to get more funding, to get more people on board, to get more users for GNOME. We have Windows and Microsoft announcing a new feature called Recall which looks like a privacy and security nightmare. We have Qualcomm potentially beating Apple’s latest chip in their own ARM CPUs which have full mainstream Linux support or at least they are working on that full mainstream support. We have the first inklings of AI legislation passed in the EU. We have some news about Ubuntu and some problems they might start having with GNOME. We have some big progress on Plasma and a bunch of other things. So as always if you want to learn more about any of these topics all the links that I used are in the show notes and if you don’t really like waiting for the end of the week to get these news well you can also get them in daily format from Monday to Friday if you become a Patreon member the link is in the description of the show as well. So now let’s get started. So GNOME has published a first draft of their five year plan or at least the GNOME Foundation did that. They published it to get a bit of feedback from the community, from developers, from users and other interested parties who might ship or might want to ship GNOME or to even just fund the foundation. So they divided this plan into three big goals each having three separate objectives. So the first big goal is to have explosive growth of the community of what they call creators and users. I guess creators mean developers here, developers for GNOME but also for GNOME applications, maybe extensions and everything around GNOME. The three sub-objectives are first to unify the community around the shared vision. I think that’s something GNOME has kind of already did. They do have their own vision of what a desktop environment is. They have been pushing hard for this and yes some people did not like this vision but at least the community around GNOME right now is focused on the same goals and it seems to be working for app developers at least because they have the biggest and best ecosystem for third-party apps on any other desktop environment. The second sub-goal is to make GNOME relevant and attractive to more diverse people and probably this means accessibility, making GNOME something that everyone can use no matter if they are fully able, if they have disabilities, no matter what everyone should be able to use GNOME and they are already working on that with their new accessibility framework. And finally they also want to increase the commercial and economic value of GNOME. Now this one I don’t really know what that means. My mind instantly went to “oh no they want to make distro space to implement GNOME” but that’s really probably not what that is. It’s probably just to make the project a more desirable target for fundraising and for potential investors who would just drop money into the GNOME Foundation to support the GNOME project. So it’s probably more communication around what GNOME is doing and less trying to sell GNOME as a product. Now the second major goal is to create a unified and integrated suite of programs, services and processes. And this seems to encompass building more integrated technologies, so maybe having more GNOME APIs that can be taken advantage of by GNOME itself but also by GNOME applications. But it also seems to target how GNOME is organized. For example they want to reorganize their myriad of little events into a single annual event that can include more people at the same time. So presumably maybe livestreams and people being able to ask questions online and them being addressed, maybe taking more advantage of the current technologies to make a one big GNOME event instead of multiple little guadeks around the year and smaller other little development sprints. Not sure what exactly they mean around that but they do want to conflate all events into one. Finally the third big goal is to strengthen the GNOME Foundation as a non-profit. So presumably to make sure it is well funded and that it works reliably. There are objectives such as documenting their impact and their value, which is going to be absolutely mandatory if they want to showcase what they’ve done with the money they have received. They also want to double the annual expense but also the annual revenue budget of the foundation. So that just means get more money so we can spend it. And the final goal is prioritizing the health and well-being of the foundation. So probably not letting it run as it did in the past by just tapping in their treasury and not really trying to find new avenues for funding. All of this seems to just mean we need more budgets to do good things with GNOME so we need to find more money first and then we can spend it. All in all it is obviously not a plan for GNOME, the desktop environment. It’s not a development roadmap, a feature roadmap, it is a plan for its supporting fundraising and governance organism. And I think it makes sense as a bird’s eye view thing. To make that project work you need to find reliable sources of funding and then you can spend that money towards growing GNOME as a project, whether it’s in terms of having a bigger community, having more developers, having more features, having more resources, having more communication.
This also requires making sure that first you have a big community, people that can be hired with that money to work on GNOME features. And to do that you need a strong vision, you need some big events to draw people in, you need some interesting technology people want to use and you need to document all of this for potential funding partners. It makes a lot of sense, the language here feels extremely corporate to me when reading this, but honestly that’s probably what people would expect from a non-profit like trying to find money, you need to appear like a company but not be a company and I think that’s what they did here. So obviously you can submit your feedback on that draft if you have remarks, just remember it is not feedback on GNOME the project, no one is asking you if you like GNOME, if you think GNOME likes certain features, that’s not what this draft is about, this is a plan for the foundation itself, it’s not a development roadmap, so don’t go around submitting feedback saying hey, we need this feature and minimize button needs to come back because that’s not what they’re looking for here. So there’s a button to submit feedback if you click on the link that I left in the show notes.
Now we also got a few interesting details about the Qualcomm Snapdragon X Elite and you might wonder what that thing is and why I’m even talking about this here, this is an ARM CPU made obviously by Qualcomm for laptops and initially it looked like it would be a partnership with Microsoft and it would only land in Surface laptops but it also looks like Qualcomm is upstreaming a lot of that support to the mainline Linux kernel, they already have the CPU support in place and they’re already upstreaming more and more parts of the entire system on a chip including how to handle the battery, the USB host, potential webcams plugged into it and everything in between. So this is a really cool CPU or system on a chip to follow if we’re a Linux user and we’re also looking to move to ARM or at least interested in ARM computing. And the benchmarks are really not bad for this chip as well because they seem to indicate it will more than compete with Apple’s latest computers. In Geekbench 6 they apparently beat the Apple M3 by a pretty solid margin, if I remember correctly, the M3 is like 88% of the performance of the Qualcomm chip, so not bad. It also looks like the Qualcomm chip beats an Intel Core Ultra 7 155H as well. Now the chip seems to run a little hotter than Apple’s M3 but it still manages to beat it in battery life as well, apparently the M3 gets 16% less battery life than Qualcomm’s chip while playing local video files. In terms of graphics though, the chip from Qualcomm seems a little bit less interesting, it’s being beaten in blender render tests by the M3 and by the Intel chip as well. But it’s still pretty impressive and of course it’s just benchmarks and Qualcomm and other manufacturers have a bad habit of optimizing for specific benchmarks to appear more powerful than they are in actual real use. But in the end it still looks like a pretty decent chip and while for now I think it’s only been announced to be used in a Surface laptop from Microsoft, with Qualcomm submitting all the code and contributing everything upstream and in the right places not with proprietary modules and weird things like that although there will be proprietary firmware in the default Linux firmware packages, it’s a good option that we might have to run ARM devices with Linux with native support and it might also be way better supported than anything from Apple because as much as the Azahi developers have been doing an insane amount of work, it’s still reverse engineering and it’s taking a lot longer for them to actually manage to find how things work and to develop drivers and to get these drivers to acceptable performance levels compared to just having the company contribute all the drivers to the Linux kernel itself. So it looks like an interesting thing and I will definitely keep an eye on it to see how good of a support it gives to Linux.
Now we need to talk about Windows and Microsoft because Microsoft held a big conference about AI in Windows and what they call AI PCs which they will brand Copilot Plus PCs and among all the usual AI features that was one that stood out as being particularly bad it’s Windows Recall and if you only look at it from a bird’s eye view, this thing seems insanely useful. You just ask your local AI on your PC something and it’s able to find anything you did on that PC that’s remotely linked to that. So for example if you were browsing for I don’t know Warhammer miniatures and you completely forgot what website you saw this super deal on and you just remember the image on the box. You ask your AI to find this exact image, you type I don’t know golden warriors with lenses and it’s gonna find an entire screenshot of the website you were browsing fully analyzed by the AI and perfectly representing what you were looking for. You could also just be remembering hey I saw this super cool bag that I wanted to buy the other day but I don’t quite remember I just remember it was a brown bag. You type find me a website where I browse for a brown bag and it’s gonna display that for you. Seems insanely useful. In fact it’s screenshots that is displaying to you because the way this thing works is the dumbest I have ever seen. It’s just taking screenshots every few seconds and it’s storing those screenshots on disk and it’s learning from those to identify objects and themes. And obviously this should raise a huge red flag because this means you will have screenshots that might display very private information or very private activities stored without your knowledge. It can use up to 25 gigs of storage on your PC to do so, that’s more than an entire Linux distro’s worth of disk space, and anyone who logs into your PC with your account or anyone who can just access the disk’s content will have access to these screenshots which feels pretty bad. Now Microsoft said that they will not take screenshots when you’re using private browsing in Edge but they didn’t mention other browsers or if they could integrate with that. And they also said it won’t take screenshots of copyrighted materials although how it will identify these is unknown as well. The feature will of course be on by default and fortunately it seems that it might be encrypted which lessens the risk a little bit. All the screenshots could be encrypted on disk but since disk encryption only exists for Windows Pro and Enterprise users because they have access to BitLocker it is unclear if Windows Home users will have those screenshots encrypted on disk or not. And if they don’t it means that anyone with a live CD or a live USB drive could just go and grab those screenshots and pass through them and see any credentials, login information, passwords or whatever else the user might have been doing or any weird activity that they might have been doing on their PC that has been screenshotted by this really fantastic AI. So obviously the problematic use cases are plenty on public PCs if the person managing this PC is ill intentioned or has forgotten to disable this feature anyone who logs into this PC could see what you did, including your credentials potentially or your personal emails. A work laptop will have all of that accessible by the IT team or even by your boss. A family PC that shares a single account is a major issue as well or if you just get your laptop stolen and they manage to log into your account. That last one is already a big big problem for you even if you weren’t using the recall feature but it only adds insult to injury because for example you might have a password manager that has a master password, the attacker that stole your laptop cannot enter that but they could just go and look at your screenshots to see that exact password that you unlocked and maybe you displayed in a full text which yeah that’s bad. So in the end it’s another let’s do AI stuff way too quickly without looking at the consequences type of features and it’s really creepy in my opinion and it also looks like there’s already some investigations at least in the US in terms of privacy to see if this feature might even be legal at all. So yeah, well done Microsoft, that’s another great use of AI that we’ve definitely built trust and confidence from the public in these technologies.
And since we’re talking about AI let’s get this other topic out of the way. Unfortunately, or unfortunately depending on where you stand on AI regulation, it looks like AI regulation is starting to appear with EU member countries agreeing on an AI act. Under this new legislation any company developing AI systems and wanting to put them into the hands of EU citizens will have strict transparency obligations to let people and regulators the type of data they use to train with detailed summaries of where they grab this data, they also will have to explain the specific biases, they ironed out of the model or they voluntarily added to it, they will also have to conduct more testing, do audits and do some reporting on their energy use and cybersecurity measures. This new act also restricts the government’s use of biometric surveillance in public spaces using AI technology. And companies providing AI tools in the EU will have to comply in the next 12 months if they are doing general purpose AI models and they will have more time, 36 months, if they are building, let’s call them more sensitive AI tools, stuff related to weapons, warfare, law enforcement and the like. This means that US companies will also not be exempt from these regulations if they aim to provide their AI tools in the EU market. The fines can go up to 7% of the company’s global turnover. And this law also seems to ban the use of AI on social scoring, on predictive policing and on scraping of facial images from CCTV footage or the internet, which is also a very good thing because it already gives countries a clear no-no on using AI to basically do mass surveillance of their citizens. There are other ways to do that, but at least AI won’t add to the pile. All in all, it is a first barrier to make sure that these tools aren’t trained on anything and everything or at least if they are, we’ll know what they use. And it also will help various countries to ensure that these AI companies are not just going fast and breaking things and just doing whatever they please. It doesn’t address the content property side of things or the licensing side of the material used to train the AI, but I’m sure that this will slowly be fixed through court cases across the world.
Now let’s talk Ubuntu. We have some more information about Ubuntu 24.10, its codenamed “Oracular” or “Riole”, which apparently is a form of small bird. Apparently this new version will be focused not on adding a ton of features, but on polishing what is already there. The installer, for example, will finally gain its TPM backed disk encryption feature, it won’t be experimental anymore. The installer will also give the user more feedback when issues arise, when, for example, the system is trying to configure a specific device but fails, you will get clear error messages and you will know what hasn’t worked when you’re trying to install the NVIDIA drivers. If there’s an issue, it will tell you so. And Ubuntu will also get a new welcome wizard that will presumably replace the one from GNOME.
And the App Center from Ubuntu should also get better at recommending apps and displaying user ratings. As per using Flutter, there were some doubts and issues with Google laying off a sizeable part of the Flutter team, but Ubuntu will keep using it, they are not abandoning this technology and they will keep investing on it, for example, they will migrate Flutter’s GTK backend from GTK3 to GTK4 for better performance. And in 24.10, there’s also one major change which is that Wayland will be the default for everyone, including NVIDIA users, they are probably banking on explicit sync support being released for everyone by the time October rolls in, GNOME will have support for it, so Ubuntu also will. And if the NVIDIA drivers have been updated then, and they’re already in beta with updates for that, I’ll talk about this later in this episode, then probably things should be a lot better for Wayland, so they are going to move that to the default. Now there were also some interesting comments from Ubuntu’s desktop lead, they said they want to focus on the desktop, and they plan to expand Ubuntu’s desktop team by at least 50% over the next year, and this is really good news, because it means that they would like to push the desktop a bit more when it’s been fixed in place for a while now, and it feels like they tried to add a few things, but they also aren’t really focusing a lot of that. So yeah, it’s interesting to see that they do have plans to make the Ubuntu desktop a priority, or at least more of a priority at Canonical.
And still on Ubuntu, they are having a few problems with GNOME. GNOME shipped a major feature in a point release, in GNOME 46.1 they delivered explicit sync in Mutter and the GNOME shell. This is something they don’t usually do in GNOME, usually, or at least for the longest time point updates have just contained bug fixes, security fixes, and some minor changes to the core applications of the desktop, but that’s rare. Features are generally not a big part of those point updates, especially major features like explicit sync. And the fact that they changed this for GNOME 46 makes their relationship with Ubuntu a bit uneasy, because GNOME shell and Mutter will no longer be covered by Ubuntu’s micro release exception, or MRE. MRE is a policy that Ubuntu has to allow releasing point updates to certain components of the distro without having to do any QA or, well, no extra testing, no backboarding of fixes or patches that they apply. So basically they say, okay, we have our patch set, we can just apply that to this minor version, it will work exactly in the same way as previously. And we know that because we agreed on that with the upstream project. This is something that they do with GNOME shell and Mutter, for example, because Ubuntu has a bunch of patches that they apply on top of that to enable certain features or to add stuff that hasn’t been upstreamed yet, like the triple buffering patch for Mutter.
Ubuntu doesn’t tend to ship new features to its applications in its repos, and tend to not ship new features to the desktop either along the life of a distro. So GNOME shipping this major feature here breaks this confidence that they have. And this means that Ubuntu will no longer just ship any point update for GNOME, they will have to look through it to backboard their own fixes and improvements, and to do a lot more testing, they will have to check every single point update on a case by case basis to see if it can be added as is or if they have to do a lot of extra work. And honestly, in this case, I’m not sure if GNOME or Ubuntu is right or wrong, I think both viewpoints are perfectly fine. GNOME needs to add explicit sync as fast as possible, because without it, the experience with Nvidia and Wayland is really bad for a lot of people. So if they had the opportunity to ship that in a point update instead of waiting for September for GNOME 47, I think they really should and they really did well shipping this right now. But on the Ubuntu side, they had basically an unspoken agreement as far as I can understand that GNOME would not break or would not bring major features to certain parts of the desktop in between big releases. And this unspoken agreement is apparently now broken, meaning that Ubuntu cannot just grab the latest version of GNOME, apply their patches, do some automatic QA and publish that because they know nothing has been significantly changed or broken.
Now in the end, I think GNOME doesn’t have just Ubuntu to work with. They ship GNOME to a lot of other distributions and those distributions tend to ship a much closer to vanilla experience of GNOME. So they absolutely should bring more features whenever they feel they can, and they should not really limit themselves to having features only every 6 months.
I’m sure a lot of people will think that it’s GNOME being GNOME again, GNOME being little dictators, but if you look at Plasma or KDE, they always shipped major big features in point updates. Plasma 6.0 got some big big updates as well along its life, and Plasma 6.1 will probably have the same. And no one really said that Plasma has a problem doing that or that Plasma is dictating how distro should work. So I think it’s the distribution’s job to adapt to the desktop that they decide to ship, it is their job to backport their patches or to apply their patchset or to tweak their patchset for every version of the desktop or to decide to just not ship a point update if they feel it’s gonna break the experience. But I don’t think desktop environments or applications should just limit themselves because distributions might have more work to do. I think it’s fair to let desktops and apps do their work and it’s fair to let distros apply an update policy that they want and that they decide it upon. But let me know what you think.
Now we gave GNOME some time in the spotlight, so it’s only fair we do the same with KDE and Plasma. And KDE developers are still working on Plasma 6.1, it should be released in a bit less than a month. One of the things that they fixed in 6.1 is theming of their applications on other desktops, notably on GNOME. I reported on this, I think it was two weeks ago, but some applications when run under GNOME with the Advita theme as the default just had no icons in a lot of places because they assumed Advita was a complete icon theme when in fact it was just an icon theme designed for GNOME and it never wanted to have all the icons KDE apps could want. So this is going to be fixed in Plasma 6.1 and with applications released after that point. The fix will be that KDE applications will first try to use the user defined icon theme, whether it’s Advita or not, and if certain icons aren’t in that theme, they’ll look at the theme that it inherits from. So for example if you used, I don’t know, let’s call it Tango icons and it inherits from Advita. They’re going to look in Tango, if they cannot find what they want they’re going to look in Advita, and if they cannot find what they want in Advita then they’re just going to use Breeze icons instead, and Breeze icons will always be provided alongside KDE apps that implement this. For now only Kate, Console and Dolphin support this, but any application could opt in to do this as well. On top of that, in 6.1 Dolphin will be able to generate, well, I’m saying 6.1, it’s also in the KDE Gear Compilation that should land a bit later. So Dolphin will be able to generate previews for remote locations, although it will warn you that this is very very consuming in terms of network resources, so it might be very slow. Discover will let you know when a Flatpak app is end of life and has been replaced by a new one, and it also will let you switch to that new app in one click. And also you will get a much easier way to get admin privileges in Dolphin, you just have a menu entry to act as administrator, you get a nice big warning pop up telling you if you delete anything, well, we won’t stop you, so be careful. And once you’ve done that, you’re in admin mode and you can browse Dolphin and move files around just like you were using Windows.
Now there are also some details on how you will be able to edit the desktop in Plasma 6.1, and how you will change the widgets, how you will add more widgets or tweak the current panel or add a new panel. Currently to do that you right click on the panel or on the desktop, you choose edit mode and then everything just layers on top of your existing desktop, meaning that some elements are hidden behind others. If you display for example the widget sidebar and you wanted to drag a widget where that side by sidebar appears, you have to drag the widget out of the sidebar, then close the sidebar that move the widget right where you want it. If you have a top panel, it’s gonna be hidden by the configuration bar that appears in Plasma to let you tweak the entire desktop. It’s a bit messy and it’s hard to notice when you’re actually in or out of that entire mode, so they’re gonna tweak that entirely. In 6.1, when you enter the edit mode, your actual desktop and wallpaper will zoom out, your panels will still occupy their current places on screen, but they won’t be on top of your desktop and wallpaper. And everything else that appears, the widget sidebar, the little pop-ups that appear when you try to edit your panels, they will appear outside of your zoomed out desktop, meaning that they will never hide the desktop itself and you are able to interact with elements without hiding whatever you’re doing and it’s also a lot clearer that you’re actually into edit mode. When you’re done with edit mode, the desktop zooms back in, occupies the entire screen and you know that you are done. I think it looks pretty good, I think it’s a lot clearer that you’re entering a special mode when you’re trying to customize your desktop. I think it’s an improvement over the current way of doing things in Plasma. There’s a little video you can watch in the article that I linked in the show notes, so you can see how it looks like. It’s pretty hard to describe, but basically if you know what the overview looks like in KD Plasma, it’s gonna look the same. Your desktop will zoom out and you will have elements to customize it on the right and on the left side of that desktop.
And finally on KDE, they just dropped the KDE Gear Compilation, it’s version 24.05, because we’re in May and they use kind of the same numbering scheme that Ubuntu does. So updates to the core KDE apps include first, improved animations for Dolphin, when you’re moving through and navigating through files and folders, when you’re dragging and dropping items, animations will convey what you’re doing in a more user-friendly fashion. You will also get more detailed information about files, for example in the recent files view, you’ll see the modification time and date, and if you’re in the trash, you will see the origin of the file. Now Kdon Live, the video editor, now can let you apply effects to multiple clips simultaneously, which is really important for productivity, and they also have an AI-powered tool to translate subtitles automatically, and they have some performance improvements. They also update to NeoChat, the Matrix client, it has a dedicated search popup, and it’s can now scan PDFs and documents for travel-related data, and they will display this information in the conversation. Tokodon, the Mastodon client, will let you write new posts in a separate window, Elisa, the music player, now gives you either a list or a grid view to navigate your music collection, and there are also updates to the Archive Manager arc, to the aggregator RSS feed reader, and there are some new apps joining the compilation, there’s Francis, a Pomodoro timer to know when to take breaks when you’re working, there’s Calm, that’s an app to give you guided breathing exercises to reduce stress, and there’s Oidex, which is an app to let you rip music CDs. So depending on your distro, you will get these updates more or less quickly, or not at all until your next major update. There’s nothing revolutionary that everyone absolutely needs right now, apart from maybe Kdon Live users, and if you absolutely need that, you can probably install any and all of these apps through Flatback or Snap to always use the latest version available.
Now this week there was also an interesting blog post about fixed kernel versions, as in a distribution ships with a specific kernel version, and they will not move to the latest stable version, they will instead backboard fixes to the specific version that they decided to stop on. And according to this post, this is not a good way to build a kernel at all. Now the research was conducted by CIQ, that’s a company that offers commercial support for Rocky Linux, and their analysis seems to point to the fact that a frozen kernel is less secure, and a lot less secure than the stable kernel the Linux team makes, and it’s also buggier, and it does tend to get buggier and buggier over time. So they have a white paper on the topic, and as with all white papers, take it with a grain of salt, there’s generally an ulterior motive, it’s made by a company, it’s why it’s not called a research paper, it’s a white paper, it’s made by a commercial for-profit company. But the data behind it seems sound, they took the example of Red Hat Enterprise Linux 8, and they tracked the number of open bugs against their fixed kernel version, and that number has been increasing to insane numbers. And we’re not talking about general bugs that everyone has, we’re talking about bugs that do have fixes in the more recent stable kernels as well, meaning that the fixed version of the kernel shipped by Red Hat Enterprise Linux 8 just did not get bug fixes that could have been there for more than 4000 bugs, and if they had just used the latest kernel available, then they would have had those bug fixes. Now, first, CIQ is a company built by the person who created Rocky Linux, they provide commercial support for Rocky Linux, and their incentive here is probably to just say “hey, you know what, we’re making clones of Red Hat Enterprise Linux 8, we have to be fully compatible with it, so we have to have the same problems they have with their kernel”. So maybe if we tell people that “hey, you know what, this kernel is really buggy and problematic, maybe at some point it will change and we also will be able to ship better versions of the kernel”. And also they are advantageous for a distribution, to stick to a specific kernel version, you have a stable platform, your users know that their hardware support isn’t going to break from under them, a specific feature will also not break along the line of the distro. This is a good thing, sometimes when you update to the newer version of the stable kernel, some drivers are removed, some drivers just flat out don’t work with your older hardware, it happens sometimes, but also this research seems to indicate that the frozen kernel model is really not a rock solid option that most distributions want to emphasize. And it’s pretty understandable, if you have bugs that you have to backport fixes for every single day, you basically need as many developers as the Linux kernel has to maintain your own version of the kernel, you have to track every single day which patch has been applied and backport it, except if your current version of the kernel is end of life, you probably have to modify these patches as well and change the code, to make it fit to the older version that you’re using that is no longer officially supported. The new patches are for the newer versions of the kernel that you’re not using, so you probably have to rewrite a bunch of that stuff and that’s probably why a bunch of those bugs never get fixed on the kernel, they’re marked as low priority and they’re never dealt with and they will only concern themselves with real big security problems and bugs that impact a lot of their users, meaning that the end experience is still worse than just using the latest stable kernel. So I can definitely understand CIQ’s point of view, I can also understand why distros would want to stick to a specific kernel version, but personally I will always prefer using a distro that has the latest stable kernel or at least is not too far from the latest stable kernel instead of using a distro that has a fixed version that they backport fixes to because it’s just worse for everybody, except maybe for the distro.
Mozilla also published this week a roadmap for the features they would like to implement in Firefox over the course of the year and probably into the first half of 2025. From these features, there’s a big one which is tab grouping and vertical tabs, this is something a lot of other browsers have, you can just create tab groups, reopen those tab groups automatically, you can display your tabs vertically instead of on the top of the window which is also pretty useful in some use cases, those are good solid features, you could already do them with extensions but having the browser do it by default is cool. Now to balance these very useful features, they will also add a pretty meaningless one which is tab wallpapers, you will be able to change the wallpaper inside of your empty tabs I presume, which like who cares about this. Menus in Firefox will also be streamlined to make the more important items more visible and more accessible, I don’t really know if that one is very important because Firefox only has a hamburger menu unless you’re using it on Mac OS where you have a menu bar and that hamburger menu is already pretty bare bones so I don’t really see what needs to be improved here but why not, and more importantly they will focus on speed, performance and compatibility, they apparently already have something that should improve responsiveness by 20% although reading the article I’m not sure if this is supposed to already have landed in Firefox or if this is something that is ready but hasn’t been shipped yet, and Firefox has also been working with the Interop project which aims to make it easier for developers to make websites that work on more than just chromium but also support Firefox and Safari and other alternative browser rendering engines.
Finally and because it’s 2024, Mozilla and Firefox will also be working on AI features but fortunately they will run locally without infringing on your privacy, examples, one example at least that they give was generating alt text for images inserted into PDFs that you’re reading in your browser, this is a pretty useful accessibility feature for people who need what’s on screen to be described or read to them, having those alt text being automatically generated could be good for documents that haven’t had this alt text added manually. And it all depends on whether you stand on Firefox, personally I’m getting a bit tired of it being way slower than other things I talked about it in my weekly Patreon cast but basically on my gaming computer I used Nobara, it came with chromium by default and I was just blown away by the speed of that web browser compared to Firefox on the same machine, a lot of patrons and YouTube members told me they didn’t have any speed difference, I tried with a brand new install, with a brand new profile without extensions and yes Firefox is absolutely slower than anything chromium based on multiple computers for me, at least for the websites that I visit, so if Mozilla delivers this 20% responsiveness increase, this might be enough for me to stick with Firefox, if they already have delivered this 20% increase then I haven’t seen it or felt it at all in the past 2 years or 3 years that I’ve been using Firefox, no it’s way more than that, 5 or 6 years, I haven’t noticed it at all, so I don’t know, I hope this is something that is coming and not something that has already been shipped, I wasn’t really able to understand that from their blog post but let me know if you know. Apart from that, this roadmap is pretty limited, I mean I don’t know if Firefox really lacks that many features but the only big thing is like an accessibility feature which is certainly needed but not necessarily insanely useful for everyone and also tab grouping and vertical tabs, that’s cool but that’s it, it’s not a lot, it’s really not a lot, when you compare it with what the Thunderbird team has been doing on Thunderbird and how fast they’ve been moving with updates, I don’t know if Firefox is still a priority for Mozilla these days, it feels weird.
Okay, let’s finish this with the gaming news, first we have an interesting project called Netris, it’s basically Stadia but it’s self hostable and it’s open source, it’s focused on using Steam so it works with your existing game library, you don’t have to buy games from a specific store and you can then stream your Steam games to other devices either from a server that they host or from your own server because you can self host it. It offers Steam library sharing, it offers HD streams, parental controls and apparently low latency thanks to something they called Quick, QUIC, which apparently helps with streaming quality and with responsiveness. Now this solution doesn’t support games that use a third party launcher like EA Games, Ubisoft, Rockstar or Battle.net through their own launchers, I suppose Epic Games, they are not sighted but they are probably also not gonna work if you use heroic or something and they do use ProtonGE to run games so you will only be streaming and playing games that work on Linux obviously, it is still experimental and if you want to self host it, it does require an Nvidia GPU, probably because their streaming improvements rely on CUDA. But it is still an interesting project if you like cloud gaming but you would like to have actual control over the game library that you have, you don’t want to pay monthly for access to a catalogue, you want to own your games, you want to be able to play them on other devices when you’re not streaming, I think it’s a nice solution, the Nvidia requirement will probably be an issue for a lot of people, it is not the most loved GPU manufacturer out there on Linux but yeah, it seems like an interesting project nonetheless.
And since we’re on the topic of Nvidia, they released a beta for their proprietary drivers, it’s version 555 this time and it is a big one. It is not the version where they will switch to their open source modules by default for recent GPUs, this is probably going to be with version 560 but 555 is the version where they add support for explicit sync on Wayland, meaning that with these drivers, most if not all of the graphical glitches, slowdowns, artifacts, latency and performance problems you have with recent Nvidia GPUs on Wayland should have disappeared. At least for all Nvidia GPUs that will be supported by version 555, if you were already stuck on drivers 390 or 435 then obviously you are not going to get any improvements anymore, these are unsupported GPUs from Nvidia, it sucks but that’s just the reality of how hardware works these days, unfortunately. But if your GPU has access to the latest Nvidia drivers then you’re getting those improvements, meaning that you will get a decent Wayland experience at last. These drivers will also move to the GPU system processor firmware by default for RTX cards, that’s something I thought was already the case but apparently not, and you will need at least the kernel 4.15 to use these drivers which should not be a problem because if your GPU can use those drivers you probably want to run that on a distro that is not stuck to the kernel version 4, it’s already very very old. And the new drivers also implement support for 10 bit per component over HDMI and a few more Wayland related features. So obviously these are still beta quality drivers, if you absolutely need to run Wayland right now with Nvidia you can definitely get them from Nvidia’s website and jump through all the usual hoops to install those from this .run package. Personally I will wait for my distro to package and release them because I have not had many or any problems with Nvidia on Wayland but if you have and you really want to use Wayland well you can try and jump in on those beta drivers.
Okay so this will conclude this episode of the show, I hope you enjoyed listening to it, as always if you want to dive deeper into any of these topics all the links that I used are in the show notes and if you want to get a daily version of these news you can also become a Patreon member, at any tier you will get a daily show 5-10 minutes from Monday to Friday with basically everything that is in here is in the daily shows plus a few things that I just cut out of those weekly episodes because they would be way too long if not. So if that’s something that interests you the link to the Patreon page is in the show notes as well. So thank you all for listening and I guess you will hear me in the next one next week. Bye!
The AI-powered Podcast Player
