Untitled Linux Show 174 Transcript
Please be advised this transcript is AI-generated and may not be word for word. Time codes refer to the approximate times in the ad-supported version of the show.
00:00 - Jonathan Bennett (Host)
Hey folks, this week we're talking about quantum computing and a reported break of RSA. We're going to talk about NVIDIA, intel and the AI craze. We're going to cover the WordPress drama and a whole lot more. You don't want to miss it, so stay tuned. Podcasts you love From people you trust.
00:20 - Jeff Massie (Co-host)
This is Twit.
00:24 - Jonathan Bennett (Host)
This is the Untitled Linux Show, episode 174, recorded Saturday, october 19th. We always say it wrong. Hey, folks, it is Saturday and you know what that means. It's time to get geeky with Linux and hardware and software and open source and quantum. Yeah, we're going to talk quantum computing today too. It is not just me, of course. We've got here today. We've got Ken on this side. Let's see, I'm too used to the mirrored thing. We got Ken on this side, we got Jeff on that side and we're going to have a lot of fun today. And we're going to start by talking about something that we sort of talked about privately in one of our back channels over the week, and that is. I forget who it was that brought it up. It may have been Jeff that started it.
01:09
Did you hear that they broke RSA with quantum computers? I was like no, I had not heard about this. So I went and I looked into it and actually wrote about it for Hackaday. And if you go and read the Hackaday article, the one that I've got linked, you will quickly discover, especially if you read the comments, that quantum computing is not my area of expertise. But if you also read the comments, you'll find out that it's not any of the commenters' area of expertise either, and as you go and you read some of the coverage news coverage on this, you will find out that it's not very many people's areas of expertise. Because there's a lot of questions about did these researchers break AES? Did they break RSA? Did they break anything? What bit size did they break? What does this actually mean?
01:54
I spent some time this week reading into all of this and I actually went and I actually read the paper. I know this is kind of crazy and not everybody can say that they did this. The paper is mostly in Chinese, but they've got quite a bit of English at the beginning of it. That gives at least the overall journey of what they did and it is interesting. Like this is not nothing, it is interesting work, it's good work, it is work that moves forward the progress on quantum computing. So first off, I want to say that I'm not looking to belittle or demean the work that is done here. It is actually good work, but it is not the end of cryptography as we know it, as some people sort of seem to think that quantum computing is destined to be. So what they did is first off.
02:42
First off. We have to talk about quantum computing itself. You have two different types of quantum computers. You've got kind of this generic general purpose quantum computer and then you have something called an annealing quantum computer and they work differently. The general purpose quantum computer, that's the one where you can actually run like quantum algorithms. If you're familiar with the idea of Shor's algorithm S-H-O-R, apostrophe S, written for Dr Shor, the guy that invented this Shor's algorithm that is the one that predicts and I think it's been fairly well mathematically proven that this will actually work that you can factor numbers into their primes very, very rapidly, much more rapidly than what a classical computer can do, and that's something that runs on a general-purpose quantum computer.
03:33
The problem with general-purpose quantum computers is that they exist but they're extremely limited in the number of qubits is the term that you see used quantum bits, and there is some thought and this is kind of my thought personally that there will always be Okay. So with qubits, to be able to get them to do something useful, they all have to be in sync. The term coherence gets tossed around and all the qubits have to be like cohered together. They have to be in sync at the same time. Someone that I was reading after this week suggested. It was sort of like you have a handful of marbles and you drop them all and they all have to for it to do something useful. They all have to bounce at exactly the same time, and you can imagine that. Well, if you have two marbles, it's fairly easy to drop them at the exact same time and they all bounce. But as you scale that up, if you have like a hundred marbles, then that becomes an increasingly difficult problem. And so, just kind of an aside, my thought on this is we may find out that building a quantum computer and actually getting a quantum computer to properly cohere at the level that we need it to with large qubit size, that may be a hard problem. I don't mean hard like what you normally think of as difficult, but like mathematically speaking, that may be a hard problem. And so RSA and cryptography may, in the long term, actually be safe, because what you're doing is you're actually swapping one hard problem for another hard problem. That is a freebie. That's just sort of my opinion, and we'll see. History may prove me wrong. As we say, quantum computing is always 10 years out, so maybe in another 30 years we'll actually have something that works. So that is the classical form of quantum computing.
05:23
You also have this, what they call quantum annealing, which that's the idea that you shake it up so it has qubits as well, but they don't. They're like intentionally not all synced up. Quite the same way happens is the state settles, and so you get to this point to where, like in a quantum, quantumly speaking, um, all of these different bits kind of settle to their lowest possible state and you get some interesting quantum effects through that and so like that's that is in and of itself a an interesting thing that you can do with quantum. So that's what d-wave does. D-wave makes these quantum annealing machines and and they're fascinating and it's a different branch of study and one of the things that's being done right now, because the error states and the sink problem it's much easier to solve in a quantum annealing machine. It's easier to work around.
06:20
But what people are now trying to figure out? What practical purposes does a quantum annealing machine have? Because it can't run these general purpose problems like Shor's algorithm? And so what these researchers did is they said, well, let's look at RSA and find out. Is it even possible to break RSA with one of these quantum annealing computers? And the answer is yes, and they did it. Kneeling computers, and the answer is yes, and they did it.
06:51
The the catch is that they did it with a 22-bit key and the like the smallest, um. I made the statement in the article that the smallest key that you actually see in real use is 1024 bits and the commenters were like by the way, nist has already said you should not be using a 20 or one or 24 bit key. Actually, the smallest you're ever supposed to use now is a 2048 bit key. And the difference between a 22, a 22 bit key and a one or 24 bit key you have to keep in mind it's it's not a linear progression of difficulty, it's actually like it's. It's parametric, it, it, um, it. The difficulty ramps up dramatically Like for every bit. You're not increasing, but you're doubling the difficulty of doing it. So there's a huge difference between the difficulty of a 22-bit key and a 104-bit key. And when you're talking about quantum computers, especially with Shor's algorithm, the thing that matters is that when you're doing this factoring and you expect the difficulty to ramp, like that with Shor's algorithm, the difficulty does not ramp, it is more like a linear difficulty, and so you can get from 22 bits up to 1024 bits much, much easier than you would expect to with classical computing.
07:59
And the big question and this paper does not answer this question the big question with D-Wave annealing computers is, even with these clever ways that they have to do factoring and to do RSA, is the difficulty going to ramp linearly or is it going to be again a curve? Is it going to be an exponential curve where you can do a 22-bit key and that's fine, but you're never going to be able to do a 124-bit key with it, and that's unclear. Researchers have suggested that the math says that it is exponential and it's not going to be linear. People that are selling these computers say no, no, no, that's not the case. It's unclear. We're going to have to wait some more time, some more work's gonna have to be done on it.
08:44
Um, but again, to put this into perspective, you can literally factor. You can break a 22-bit rsa key by hand, on pen and paper, and in fact, on one of the reddit threads that I was reading, someone, someone did that. They're like here are the numbers that you have to know to be able to break this. Figure out which one your big key divides into and you've broken it, and it was. It was actually pretty impressive work. So, yes, neat stuff going on with quantum.
09:08 - Jeff Massie (Co-host)
No, you don't need to be worried about it yet well and that was pretty much when we were talking in the background that we, we, we agreed on, that it's nothing to worry about, yet it's it's a ways off quantum computers are going to be great for handling analog problems this one, this particular one, the the annealing computers are.
09:31 - Jonathan Bennett (Host)
They are basically analog computers. Yes, from what I understand, that is essentially what is going on and it's like I said, it's unclear to me that the sky is ever going to fall because of quantum computing. Obviously people have to prepare that it might, but time will tell. It's one of those things where it's always 10 years out, it's always 10 years away, and then I'm sure we're all going to wake up one day and it's like surprise. Here's the paper that just showed that it's possible. Or here's a new computer we built that is a thousand times more powerful than anything else ever has been.
10:07 - Jeff Massie (Co-host)
Those sorts of things sneak up on you, yeah, but I think we'll get into the 10 years away and then we'll get into a oh, it's three years away, it's one year away, but each one of those is going to iterate several times. So when we're even, oh, it's about a year out, okay, we'll see it in maybe a decade.
10:21 - Ken Mcdonald (Co-host)
I guess you could say a good analogy is we're back to the 8-bit days with quantum computing.
10:28 - Jonathan Bennett (Host)
If even the 8-bit days we're more like in 4-bit.
10:34 - Jeff Massie (Co-host)
And the thing with the computer stuff back then standard computer it scales so easily you can just keep adding iterations. But with the quantum computing there's so many of those qubits you have to have to basically double-check what your other ones did. You've got to have more marbles kind of double-checking what you did. That it's not a linear progression where that 4-bit oh you just take that and you can actually just stamp another one. Now you have an 8-bit machine. I mean a little more complexity but it's not that much greater.
11:15 - Jonathan Bennett (Host)
Yeah yeah, that's fair. So, like with designing processors, in some of those circuits you can literally just glue two of them side by side. You can take two 4-bit chips and just connect a couple of pins together and glue them side by side and you've got an 8-bit chip, and it's a very different ballgame in quantum.
11:41 - Jeff Massie (Co-host)
Yeah, I've had some processor design classes and I actually designed a processor in college. No, super simple, you know, just add, subtract some basic functionality. It wasn't you know nothing, nothing commercial, but yeah, that's a lot of it. You design one part and then you just start chunking them together and that's you know.
11:59 - Jonathan Bennett (Host)
So this is, this is a total sidetrack. This is a total rabbit trail. But if that sort of thing sounds fun to you, I have a book recommendation. Actually, I don't have it here with me because I've lent it to a teenager that is interested in this sort of stuff. Nand to Tetris N-A-N-D to Tetris is a book that is literally well, it's a college course, but it's also a textbook that is literally about assuming you only have NAND circuits, let's step through building your own processor, building your operating system, building a language and then making the game Tetris in that language. And then there's also some like add-ons to this where people have said, well, let's do it in a real FPGA, let's do it in a real FPGA with only open source tools, and it is an amazing. I've gone partway through the course and it is really really neat. So if that's something that really tickles your fun bone, nand to Tetris is something to look into. So that sounds pretty cool, it is cool, it's a lot of fun.
12:58
Let's talk about something else. Let's talk about Calibre. I have been told it's not Calibre, it is Calibre. I have been told it's not Calibre, it is Calibre. I think this is well. I think it's one of those words where there's more than one way to spell it. But, ken, you have the scoop on Calibre, don't you?
13:15 - Ken Mcdonald (Co-host)
Yes, I do. And this week Marius Nesta wrote about my favorite cross-platform ebook management software. He said developer kovid goyal has released version 7.20 of caliber and according to covet, on the caliber website it is pronounced as caliber, not caliber calibri. In fact I've got a link in the show notes to where he talks about that. But I am going to touch on some of the highlights of this new release. First, we have a new default dedicated PDF input engine that features automatic detection and removal of headers and footers based on document analysis. The old PDF input engine used by caliber is still available and you can switch between the new one and the old one in the pdf input section of the conversion dialog found through preferences. In fact, I was playing with that. The old one says pdf to html when you select it.
14:20
By the way, uh, caliber 7.20 also updates the Kobo driver to support the new firmware used by the Tolino Shine 5 ebook reader released last April. It also updates the trim image tool with a new control that lets users adjust the aspect ratio of and improves the read aloud feature by allowing users to configure an extra pause at the end of every sentence when using the Piper text-to-speech engine. I found playing with it, that the max increase that you can do is a two-second pause. I also played around with the read aloud feature using a section from the book by Mark Twain, Tom Sawyer. I actually found it hilarious to hear it read extremely fast, then go with a two-second pause between sentences and you do have several voices to choose from. For those of you that have a large e-book collection in Calibre, you may find it convenient to use this read-along feature to create your own audio books to take with you when you're driving. Just play with the voices there. Now for the PDF output engine. It has also been updated as well in Calibre 7.20, with support for the underscore width underscore, pixels underscore and the underscore height underscore pixels underscore. And these are variables, by the way, to know the width and height of the header and footer area in templates.
16:03
Several bugs were addressed in this new Calibre update To improve the content server. These include an issue where embedding the server HTML inside a third-party iframe caused an error, and it also addressed an issue where books with non-ASCII filenames couldn't be downloaded in the Kindle browser when using the mobile view. I can see where that could be very difficult. Calibre 7.20 also fixed an issue with Read Aloud Piper's backend preventing it from working on. What's this Windows system they were talking about, With voices that have non-ASCII characters in their names, and a regression from the previous release that broke the ability to copy a book to another library if the book author doesn't exist in the destination library. I'm glad I skipped over that version. If you use Calibre to download news from various sources, then I'm going to recommend reading Marius' article or the changelog for more details about the new, improved news sources.
17:13 - Jonathan Bennett (Host)
Yeah, it's an interesting tool. I like the fact that it has the voices and all of that. It's pretty cool.
17:22 - Ken Mcdonald (Co-host)
And it has a very active community of developers for its add-ins, plug-ins yeah.
17:33 - Jonathan Bennett (Host)
Yeah, definitely.
17:34 - Ken Mcdonald (Co-host)
Would love to hear either COVID or maybe one of the plug-in authors on Floss Weekly at some point in the future.
17:41 - Jonathan Bennett (Host)
That's a good idea. I can make a note about that some point in the future. That's a good idea. I can make a note about that. Um, I have to have to tease him about making program, a program that is called calibre.
17:52 - Jeff Massie (Co-host)
That'll be fun if one thing were consistent. We always say it wrong we always say it wrong.
18:01 - Jonathan Bennett (Host)
Well, there's the show title, right there. We always say it wrong. Well, there's the show title, right there. We always say it wrong. All right, jeff, let's talk about Plasma. We know how to pronounce this one at least, although apparently hardware is hard.
18:19 - Jeff Massie (Co-host)
Yes, that's according to Nate Graham. So KDE Plasma 6.2 is out, woohoo. So KDE Plasma 6.2 is out, woohoo. Well, okay, it's not really new news as it came out on the 8th of October with a bunch of bug fixes and polish. So it's not this week's main news and that was the plan that it was going to have all these bug fixes, polish. Well, 6.2 had some regressions and 6.2.1 and 6.2.2 are coming out, which helped fix some of the problems that reared their head with the latest release.
18:50
I don't know how widespread some of those 0.1, 0.2 are going to be or if they're going to jump straight to 0.3. We'll see what it is. I mean, it's in the background but I don't know if it'll actually hit official distributions if you're not running the beta. Anyway, the link to Nate Graham's weekly blog, you know, as Jonathan said, has the title of this week in Plasma Hardware is Hard. So that's what he said and you know he said the hardest part is the diversity in the hardware that people have. You know, if you look at the last, we'll just say seven years of CPUs, for example, and then you take all the motherboards that you know come along with all that with every GPU in that time range. You know discrete and internal. You know mix them all together. You know, and that's not even counting all the different distributions there are. So there's a lot of combinations and you know you realistically can't test that because you just have to. That's why they have beta, so you can just put it in the wild and just trying to get all the bugs out of it.
19:52
Well, plasma is moving fast and a lot of bugs are squashed in 6.2.1 and 6.2.2 and more features are coming and being worked on in 6.3. So you've got multiple versions building off of each other. So just to get everybody up to speed, first of all we have a few of the bugs that were fixed in 6.2.1. And they fixed three plasma crashes affecting the system tray and disks and devices widget under various circumstances. Fixed a case where plasma would crash in brightness-related code. Fixed a bug that caused configuration pages of the system monitor widgets to not be scrollable when needed. And that's just a. There's a ton more fixes in 6.2.1. I just touched on the major ones and a lot of the crashes.
20:42
There's a lot of little paper cuts they fix that I'm not going to go over. You know you open a drop down or something, and it's offset from where it should be or something like that. That's not a total deal breaker using the system, just a paper cut, as we call it. 6.2.2 brings the following fixes where they fixed a regression that would cause the cursor to misbehave in certain video games. Fixed two visual issues in Breeze GTK4 theming. Fixed a regression that caused HDR to stop working properly in games that request absurd brightness levels. You know like a billion nits of brightness. Absurd brightness levels. You know like a billion nits of brightness. I'm assuming that the the games either had bad coding or they said turn it up as bright as it'll go, let's just put in a billion. Fixed an issue that caused a visual distortion in the clipboard widgets config window when interacting with it in a specific way and you know, 6, 6.2.2 has more fixes but not any crashes and just more odd window behavior in specific situations that they fixed.
21:54
Now 6.3 has some goodies coming as well. So now it's possible to customize the pressure curve for drawing tablet pens. So you can you know you artists out there will you have that force uh that you can use on your pens. So you can, you can make your uh artistic lines better by by. You can tweak it now, uh, improving the way pop-ups using the sliding pop-ups effect slide out of floating plasma panels. The plasma digital clock widget now displays all events on days with more than five events, making it actually useful for that use case. Now, plasma's power and battery widget now shows better placeholder text when you're managing power using TLP instead of Power Profiles Daemon, or when Power Profiles Daemon is installed but not supported by the device's firmware. Many more features are coming, but if you notice, there are not any bug fixes yet in 6.3, and those will come later as we have to get the 6.2 series all released so they can see what needs fixed and what fixes took care of what issues.
23:02
Just to give a status update, the overall picture is currently and this is by Nate's latest blog. We have four very high priority plasma bugs, which are up from two last week, 33 15-minute plasma bugs, down from 35 last week. And for those that don't remember the 15 minute plasma bugs, those are bugs that a new user would would notice in the first 15 minutes of using the operating system. So that's, that's why it says 15 minute it's, it's the quick little. Hey, that doesn't seem right kind of things. But while the you know, it seems like there's a lot of bugs 143 KDE bugs of all kinds were fixed in the last week. So there's a lot of work going on to make this smooth, and you know. So while some people might think that, like I said, kde Plasma 6 is a little rough around the edges, they're working hard to make it smooth. You know they're really trying to put some polish on it while adding all these new features and getting it up to the speed of the 5 Series, which, remember, the 5 Series was out for years. So they've come a long way in a short amount of time.
24:15
If you're at all curious, I'd give it a shot Myself. I'm on 6.1, and while I have noticed a few paper cut issues, there isn't anything major stopping anything I'm doing on the computer. If you're really curious you know, for those that didn't listen, maybe you're new check out last week's show and I went over my adventure into KDE 6.1, you know a few little hiccups nothing, nothing major, and I don't have a link in the show notes. But I do know Neon rebased to 6.2, based on the 24.04 release of Ubuntu. So if you want to see what the KDE developers are playing with, that's a good way to take a look. But take a look at the show notes for full details on everything. But take a look at the show notes for full details on everything. And it's got links to all the bugs, the open ones, the fixed ones, all the logs, the notes, the code, everything you'd ever want to see. So give it a shot. Let's hear what you think.
25:17 - Jonathan Bennett (Host)
Yeah, I see somebody in our Discord is talking about how well, there's still bugs. That seems a little scary. Maybe it's not ready. I would say, give it a shot. How well, there's still bugs. That seems a little scary. Maybe it's not ready. I would say, give it a shot. Uh, I would even recommend something like the fedora 41 beta would be a great place to go give kde a test run, because you know they're going to be a very up-to-date kernel, very up-to-date kde release, um, and that's that's. I've run that on some machines and I've had great success with it, um. So I'd say, give it a shot and see if it works for you Try it on your daily system.
25:48
Well, I mean, if you don't want to jump in to the deep end of the pool, you can run it off of an ISO. You can put it on another machine, put it on a laptop you don't use very often. There's lots of options. Yeah, virtualize, although a lot of systems do not do well. A lot of new desktops do not do well with virtualization because so many of them need hardware acceleration these days. Several people ran into problems on Cosmic, the new Cosmic desktop, the beta that they put out of that, because you don't generally have any hardware acceleration in your virtual machines.
26:26 - Jeff Massie (Co-host)
Yeah, I like to run it on bare metal. When I play with something, I just either throw an old drive in or partition something, just a dual boot kind of thing and just have fun with it, and I wouldn't worry about it for a daily driver if it's not a mission critical machine or something that you're really panicked on. And if you, if you say, oh, I gotta roll it back, okay, it's not a huge deal it's, I would go ahead and go for it. If, if you're mission critical or you just like I cannot take any downtime, I cannot. It might be a little rough yet, but you're probably going to stay on lts anyway. You wouldn't go to 24 10 or you wouldn't go to fedora 40, you would. You know. You're going to stay away from that kind of stuff anyway. Yep, yep, that is fair all right.
27:15 - Jonathan Bennett (Host)
Well, let's talk about, uh, another release, this, this yeah, it was this week, I think, and that's ardor 8.9. No wait, 8.10. So we just talked about 8.8. What is going on here? Well, I did a little bit of digging into this and 8.9 got released because there were a couple of major issues with 8.8. People were getting crashes, random crashes, that were caused by 8. People were getting crashes, random crashes that were caused by it says in these release notes, a subtle change in how waveforms were drawn, introduced in 8.8. It also turned out that MIDI notes immediately at the start of playback were ignored. Both sort of problems. So both of those got fixed with 8.9. And then in 8.9 they found, uh, some more problems and they say here that there were a couple of new major issues. Uh, scheduling of disk input output threads, which tended to be very system dependent, but there were very real problems on the some systems where this particular problem occurred. Uh, and then, uh, importing smf mid MIDI via drag and drop now uses the file name Once again, content slipping a region now correctly causes an update of the playback buffers. So you can hear the results as expected. Just some some little irritating problems. I guess some of them were more than just little, but some irritating problems that slipped into 8.9. And so now there's an 8.10 and it's. It's funny Once again.
28:45
This is once again intended to be the last release of the 8.x series and the release notes here says that our Git repository is now at 9.0-pre-0. And you might be thinking what's going on with Ardour? Why have they had like four releases in a row now where they've had problems, they've had to do hot fixes. Why have we been on the 8-dot series for so long? And there is actually in the Ardour discourse in their kind of community forums, paul Davis throws out a reasonably long blog post talking about that. And there are a couple of things that have been happening behind the scenes and one of those is actually that they have been doing the sort of the commercial version Mixbus, and they've been working on I think it was a 10.0 release from Mixbus and there were, yeah, harrison Mixbus, 10.
29:46
And there were some big things that landed in that, one of which being Dolby Atmos support and not all of that stuff, particularly with Dolby Atmos because of licensing and patents and all kinds of stuff that doesn't make it back into Ardour, and so they sort of took a you could call it a break from Ardour. That's not exactly what it was, but they spent more of their time. So Paul Davison and X42, they both spent a good portion of their time doing this stuff. And then there's also things in the background of Ardour that they have been sort of refactoring piano rule, which they sort of took all of the MIDI stuff and abstracted it away so that they could use it anywhere without doing code duplication, and of course that sort of refactoring work always takes a while and doesn't have a whole lot of immediate benefits. And then one of the other interesting things that they talked about here was a new DAW Digital Audio Workstation that is focused on front of house operations. And so this is that is focused on front of house operations. And so this is.
30:51
It looks like that they are doing a sort of another version of Ardour that is going to be just for live recording, maybe live audio, live playback, but mainly live recording is what it appears to be.
30:59
It's not going to have any MIDI support, it's not going to do plugins, it's going to be much, much simpler than what all of Ardour gives you right now, because boy, there's a lot that's in Ardour the full-blown version of it, and so they're kind of looking at this and scaling it down to what makes more sense in a live setting. So live setting is front of house, live setting, yeah, front of house. So the term house is I think it's originally a theater term you might have heard the term stage left and stage right. Well, stage is the stage, house is the audience, and so when you're talking about front of house, you're talking about the sound system that exists at the front of the audience. So, like your big speakers that are pointed back towards the audience, that's the front of house system. So, like your big speakers that are pointed back towards the audience.
31:46 - Jeff Massie (Co-host)
That's the front of house system. I didn't know it. I've always heard it as a restaurant term.
31:52 - Jonathan Bennett (Host)
Yeah, different, different. So anyway, they're working on that, and so there are a lot of these things that have been happening in the background of Ardour and just have not paid off for everybody yet, but it sounds like they have made their way through that and now they are kicking it for 9.0 in earnest.
32:09 - Ken Mcdonald (Co-host)
I'm definitely looking forward to see how it handles MIDI after they finish that.
32:16 - Jonathan Bennett (Host)
Yeah, you know, midi has been good in Ardour for a long time, but they're making that MIDI editor a lot more usable in different places, which I think is going to open up some very interesting things in various places, like the clip. So one of the things that you can do that's reasonably new in Ardour is the ability to do clip editing and playback of clips, so you could put together a song that's entirely based on clips. So, like you could put together a. You could put together a song that's entirely based on clips, and so now you can grab something like I've got one here. Actually, you can grab something like this, which is just it's a pad with a whole bunch of different buttons on it and you could program one of these. So, okay, when I push this button, this particular drum fill starts playing, and then, when I hit this button, we're going to do a chord structure, and I hit this button and this particular synth riff will start. Well, then I hit this button and the synth riff stops. I hit this other button, we go to a different drum fill, and so you can literally put together a song with just the different combinations, and so there are YouTube videos of people doing this, by the way, just building a song based out of loops, right, and so it's that idea inside of Ardour.
33:25
One of the things that's been a problem it's been kind of clunky, I won't say it's a problem, but it's been clunky in Ardour is building it's particularly the MIDI version of those loops, of those clips, in real time, because what you have to do is you have to, like, go to your main editor, record the clip there and then grab it and move it over to your clips library and then put it onto one of the buttons and it's just, it's clunky, and so I'm hoping that's one of the things that they're looking to do with. This is give people the ability to work with all of this a bit more easily, and I'm excited for it. I think it's going to be cool. I enjoy fiddling around with our door. I get it from, you know, from time to time. I'll do all of the connections and hook it up to. I've got the modular synth with some MIDI stuff in it and all kinds of fun stuff. It's fun, pretty neat it is. It's fun. All right, let's talk about a new version of Clonezilla Ken.
34:16 - Ken Mcdonald (Co-host)
Yes, and I do recommend using this before you try out Plasma. Yeah, but with doing a first off start I out plasma. Yeah, but with do want to first off. Start off by thanking both Bobby Borisoff and Marius Nestor for informing us about developer Steven and I apologize if I mispronounce this Shial's release of CloneZilla Live version 3.2 this week. If you are unfamiliar with Clonzilla, bobby's article gives a brief description of Clonzilla and the tasks that it may be used for, so we'll just move straight to what is new in version 3.2.
34:56
Clonzilla Live version 3.2 provides a range of major enhancements and bug fixes. One of the key changes, according to Bobby, is upgrading the underlying OS. The new release is based on Debian SID repository, with the updates taken as of July 15, 2024. For those of y'all that aren't familiar with Debian SID, it's basically the test version of Debian and basically what Stephen's doing is just taking a snapshot and using what was in the Debian SID as of that time to base CloneZilla on. Now, clonezilla Live version 3.2 also provides users with better hardware support by bumping the kernel to Linux 6.11. And to include using ZSTD space dash capital T zero by default. I'll let you all research and find out what that actually does Now. This will make it easier for users to customize it using boot parameters parameters.
36:36
Clonezilla Live 3.2 also updates the OCS-Scan-Disk utility, tog utility to add the Messing-EDIO option, the OCS-On-the-Fly utility to disable the –J2 and –E2 options by default when running OCS-SR, and the LiveConfig package to support the user-crypted boot parameter.
37:09
Other changes include allowing the percent symbol in the auto-name format of the image name in the text mode user interface when saving an image. It improves the batch mode to quit when failing to restore a partition and replaces the reboot option with system control dash f reboot to prevent root over network file system from hanging when rebooting. Clonezilla Live 3.2 removed two packages that are no longer available in the package repository. The first is the WirelessTooths package and the second one is the Riser4Progs package. As some of y'all may remember, the Riser for Progg's package was actually deprecated upstream no-transcript. You can go back to the May timeframe and you'll find where we discussed that. As always, there are more details available from Bobby and Mario's article, as well as links to the official website, and you can also follow from the official website to the GitHub where, if you're really curious about all these options and functions that he's talking about in the article.
38:52
You can dig through there to find out more.
38:55 - Jonathan Bennett (Host)
Yeah, good stuff For those of us that run something like. My mind has entirely gone blank on the name of it the application that lets you put a whole bunch of isos on a flash drive. Ventoy, yeah, ventoy, thank you. Um, it is time to go and grab the new iso and give it a test, right. So whenever you get a new distro, like an update to your distro with ventoy, make sure and give it a test before you need it. I have been. I have been stuck a time or two because you know, in my case it was a new fedora release, did something new and fun and ventoy didn't have an update for it yet. On that particular occasion I had to boot into windows to fix my linux distribution, which nobody ever wants to have to do that oh for shame I know what I attested is to set up one of them for make two copies of it.
39:47 - Ken Mcdonald (Co-host)
Put one on a system that you've been thinking about moving to a VM and then another one on a system and actually have it booted up within a VM and run it from there so that you can just copy that. Use Clonezilla to copy from the uh bare metal into a vm yeah, yeah, that could certainly work.
40:12 - Jeff Massie (Co-host)
Yeah, I was gonna say too. I was like I thought we just talked about clonezilla, and so I I looked in the article and like, oh yeah, three, is it three? One three came out just three months ago, so they're they're really, uh, very active on the updates and the new releases.
40:26 - Jonathan Bennett (Host)
Yeah, pushing these out quite rapidly, which is good. I mean, nobody wants super old tools hanging around. It tends not to work on new hardware and all kinds of stuff.
40:36 - Ken Mcdonald (Co-host)
How long ago did 6.11 come out for the Linux kernel?
40:41 - Jonathan Bennett (Host)
That's been very recent. Within the last three or four weeks I think he's moving up there quick. Yeah was. I was impressed to see that. Now, one thing with 611 is it's not one of the long-term kernels. It's only going to be supported for a few months, but I guess at the speed that they do updates at clonezilla that's not going to be a problem. Correct, all right? Well, one of the things Ken mentioned was support for new file systems, and I think Jeff has the down low on yet another file system. It's not really new anymore, but what's new with OpenZFS?
41:15 - Jeff Massie (Co-host)
Yeah. So it's been a while since we've checked in with OpenZFS and I want to start with clearing up something in the name. So this is going to be a greatly simplified account of the history, but it helps clear some things up. So ZFS started with Sun Microsystems and later in the life, sun opened it up to open. They said it's okay, this is open source. Well, later, sun was purchased by Oracle and Oracle closed the code for ZFS. Well, later, sun was purchased by Oracle and Oracle closed the code for ZFS. So the last version of the open source code was forked and the OpenSolaris project continued to develop the code.
41:55
Well, there was OpenZFS and ZFS on Linux and, long story short, in 2020, zfs on Linux was merged into the bsd open zfs and now they share a unified code base for both platforms. So it when we say zfs, like on linux, it's open. Zfs is the official what it is, because oracle keeps developing zfs on their own behind closed doors. So the code base is diverged a little bit. I mean, maybe the features are the same, but how they're coded could be very different just because Oracle.
42:33 - Jonathan Bennett (Host)
Because Oracle.
42:34 - Jeff Massie (Co-host)
Yeah. So I just want to make sure when people like, well, wait, zfs, this is OpenZFS. What is OpenZFS is what we're running in Linux, so which brings us today where we have OpenZFS 2.3, release Candidate 2. And that was just released. Now it's only been a couple weeks since they had Release Candidate 1, which brought support for Linux kernel 6.11. Funny, we were talking about that.
43:04
2.3 in general brings direct IO to bypass the arc, which is adaptive replacement cache, which increases the efficiency of the transfer of data. It can also now handle directories and file names with 1,023 characters. It's also got RAID-Z Expansion support has been added, which allows a person to grow a RAID-Z pool without needing to take the RAID down, and fast deduplication has been added as well, and they're saying that's a major performance increase when you're trying to deduplicate your files. In the article to show notes, they have a table showing some of the improved performance where they optimize the kernel, same page memory and the optimizations, and optimize CPU pinning, because it can cause small delays when a task has to jump from CPU to CPU, and we've, in this show, have talked about how there's a delay with even one CPU if you only have one CPU in your system and you've got to jump from node to node, especially when the cache a core is using is attached to another node. Now it's all behind the scenes, but there's more latency when you have to jump to a further away uh section or when you're moving your thread from one core to another. So they try to optimize it and keep everything efficient by pinning down that thread. Now, sure, it's only maybe just a few nanoseconds, but when it's getting hit millions of times it slows things down.
44:44
They did run a benchmark with the ZFS test suite and the best improvement was from AlmaLinux 8, which took about 36 minutes out of the three hour and 42 minute test. You know there was a few that optimizations didn't improve the time, but they were worse by maybe a minute or two. I mean it very close. So you know, when you're running like a three and a half hour plus test, a minute difference or less than a minute, that's pretty much a wash. So it really isn't going to hurt you. The performance isn't going to hurt you, but a lot of cases it really helps quite a bit.
45:20
In the article, in the show notes, we have a link to the GitHub page where you can get the full list of the bug fixes that have been taken care of in the different RCs, like in the article in the show notes. It's for RC2. And just from RC1 to RC2, there's 21 fixes that they've got listed in there. So take a look at the article in the show notes because it's got details for the upcoming 2.3 release, the 2.3 release candidate one, release candidate two, and it's got all the OpenZFS goodness you could ever ask for in there.
45:56 - Jonathan Bennett (Host)
Yeah, so is OpenZFS, and I know we've talked about this. It's OpenZFS. That will not make it into the Linux kernel, right?
46:04 - Jeff Massie (Co-host)
Yeah, there's some licensing hiccups, so that's why it gets loaded in later, because it's I can't remember the license it's under, but it's not technically compatible, I don't yeah, there's.
46:20 - Jonathan Bennett (Host)
yeah, there's questions and I think there's also questions about whether it's like entirely freely under that license, because Oracle never entirely or Sun never entirely signed off on it Gets into complicated legal licensing stuff but it's not in the kernel because of that. But I know a lot of people really like it. I was doing a little bit of looking here and OpenZFS not only is a file system, it's also the logical volume manager. It will replace LVM and does all of that right inside of OpenZFS. Not only is a file system, it's also the logical volume manager. It will replace LVM and does all of that right inside of OpenZFS.
46:50 - Jeff Massie (Co-host)
And that MDADM as well, with the RAID.
46:55 - Jonathan Bennett (Host)
Because it does RAID stuff yeah.
46:58 - Jeff Massie (Co-host)
So it does a lot of heavy lifting in industry. So it's a very secure fi. I mean people.
47:06
People trust it, you know and it's, and part of it too is the amount of features that it has built in. You know, and with with the average user, I don't think you're probably going to notice anything. And if you have lvm or mdadm running a, yeah, it's not all combined into one program but you're not at the redundancy level that a lot of enterprise has to have. Because if you're like, well, I got to have the most secure, solid thing, well then you better be running multiple power supplies. You better have UPS generator backup. You know hardware that fails over to another raid, I mean that's. You know you can't. It's kind of silly to say I've got to have the most stable file system but you're running it on a cheap generic drive. That you know. You know you got some eBay thing you put together. It's like, well, you're not, you're not getting the full equation there.
48:03
And going back to your licensing a bit, I know in the article too they talked about how does the Oracle issue matter as much anymore? Because they've talked about we've had several years where the code base has probably diverged more because so much has been changed in the open versus what's in the Oracle. You know how close are they anymore Versus what's in the Oracle. How close are they anymore? I mean, nobody knows, but it was kind of thrown out that maybe things will change someday. But I wouldn't hold my breath.
48:32 - Jonathan Bennett (Host)
It works to load it in later I took a look on another project. Another project is GPLv3. And we looked into the idea of adding an exception For a particular use case and it's like, well, this is technically not something that it's not something that is already granted by the gpl. It would be an exception on top of the gpl. It's the linking exception. Right, it's the same thing. The kernel does, um, but it's like this is not something that the original author signed off on.
49:02
So start looking at, okay, what would it take to figure out whose source code is actually in here, and so you could do things like a git blame and find the different names. But then you get to looking at those git blame logs and some of those are where someone just added a tab to change the indention. It's like, well, that's not a transformative enough change for it to be a new copyright owner. And so all of that to say, trying to untangle the history of a project, to get the owners of it to sign off or to even audit it and figure out is there any Oracle code still in here? Is there any of Sun's code still in here that they still have a copyright? That would be a monumental task.
49:40 - Jeff Massie (Co-host)
That would be a huge task and and you know the kind of rub of it is it would have to go to the courts and you never know what they're going to find. Yeah, I mean, it could be a programmer's going to go. Yeah, it's black and white, they're not the same. Well, that might not be what the you know jury at 12 come up with, or whatever, or what kind of arguments are made or yeah, you just, and a lot of times nobody wants to pay the lawyers that much.
50:10
To sort it all out and correct versus. You know it works good to load it in as a module and we'll just deal with it there well, unfortunately, only other things.
50:22 - Jonathan Bennett (Host)
Unfortunately, I've got a story here where it looks like some people are dedicated to spending the money on lawyers.
50:28 - Jeff Massie (Co-host)
Oh no.
50:29 - Jonathan Bennett (Host)
Yeah, so we got to talk about WordPress. This is not the first show where I've talked about something and I've told you guys I don't want to talk about it. I don't want to be talking about WordPress Slightly different reasons, but I still don't want to be talking about WordPress. Slightly different reasons, but I still don't want to be talking about WordPress. I would prefer to be talking about how awesome WordPress is and how you know the people at Automatic are doing good work, being good stewards of it, and how that Matt Mullenweg is a show or is a friend of Floss Weekly. He's been on at least once, maybe a couple of times, and I think all of that is true. I think all of that is true. But right now there is some major soap opera level drama going on between WordPress and WP Engine, and it's unfortunate. So essentially, what has happened is WordPress like the WordPress automatic.
51:21
The company actually, uh, owns the trademark to wordpress, and trademarks are interesting because one of the things with trademarks is it's sort of it's sort of been decided by the courts mostly, but also by some legal things um, that in order to keep a trademark, you have to defend it and you can't just let people use your trademark, because if you just let people you know without any formality to it, can't just let people use your trademark, because if you just let people you know without any formality to it, if you just let people use it, then the courts will eventually find that you're not defending your trademark and therefore this trademark goes away. That's essentially what will happen. And so, on one hand, companies are required when they have a trademark, they are required to actually be somewhat assertive about the trademark and not allow people to just use it. There has to be trademark grants so people get to use it, and otherwise people really are not supposed to use the trademarks. Automatic has the WordPress trademark and the question here at the core of this is whether the WP in WP Engine well, obviously it refers to WordPress, nobody disputes that. The question is whether that is similar enough to WordPress for people to be confused over it. Is it a trademark violation? And in the past and in fact in WordPress's own guidelines, wp is not considered to be confusing, like it is not closely enough related to WordPress for it to be covered by the trademark.
52:53
I don't know exactly what happened to change this. I don't know that it's actually been public exactly why they have changed their stance. But Automatic has changed their stance and come out and said no, no, wp is now close enough to WordPress that you shouldn't use it. You should ask for permission before using it. And, of course, wp Engine uses WP in part of their naming and the argument that is being made is that people are confusing the WP Engine company with WordPress itself, and that would be a problem. That would be unfortunate, that would be negative and that would be something that they would have to go after for WordPress or for trademark use for the WordPress trademark. That rings hollow to a lot of people the idea that WP Engine is that confusable with wordpress itself. But automatic and mullowig have charged forward with this idea, um, and they they're threatening legal action.
53:57
They've done another couple of things that are really problematic, like just bad ideas. For example, one of the WordPress plugins that was made by people that have an association with WP Engine has been taken over on the WordPress store by Automatic and that has ruffled a lot of feathers. People are not okay with that. And then one of the other things that they did that I just I could not believe they made a. So when you go to log into WordPress. There are some buttons that you can press right, and they added a check mark to the WordPress login that says, essentially, I do not have a business, I am not affiliated, that's the way it says. I am not affiliated with WP Engine. And I believe if you check that box, you're not allowed to log in anymore. And I am just. I can't hardly believe that they did that. For one thing, I don't think that is legal, I think that is actually an antitrust violation and I believe that is literally illegal for them to do that. But just from, like the open source and the community perspective, I'm flabbergasted that they would go to that length.
55:17
They sent a, they tried to settle right, so, like Automatic, sent a letter to WP Engine and said, hey, we will let you use the trademark if you will do this. And they gave them some things that they wanted them to do and they want 8%. Right, they want 8% of their gross sales and that seems a little onerous, but then they have an option in there. It's like or you can dedicate eight percent of your sales to people working on wordpress, um, at the uh, basically what they work on, to be controlled by the wordpress organization, which is not the same as automatic um and and that seemed a lot more tolerable, at least to me. But then there's like another um, another stipulation in there that you won't fork, and then they listed a couple of other open source projects that are like under the gpl. It's like you will not fork these. It's like, guys, you, you, you can't. The gpl says that you can't do that. Like the gpl says that you cannot put that sort of restriction on your users. Like this is not okay. Like you're now in violation of the gpl as well.
56:26
Um, I am, I am flabbergasted. I've got a link to brody's video on it and, uh, he makes the statement that he's like, at some point mullenwig has to understand that what he's doing here is just totally insane. Um, but it's like I think the guy is in too deep now and can't turn back and I, I don't know the whole thing is insanity to me reading some of these things, like the depths that they have gone to, like not all of it is unreasonable. Like to say, hey, we think people are confusing wap engine and wordpress. Can we do something about that? Like that's a reasonable, that's a reasonable ask, but the, the place that this has gone to is not reasonable. It is. It has gone from a reasonable ask. But the place that this has gone to is not reasonable. It has gone from a reasonable ask to unreasonable repercussions.
57:10
So I don't know what the future is going to hold for WordPress. I hope they figure it out. I really, really hope that the people involved with this can sit down together and figure something out, because what is going on now is not good for WordPress, it's not good for the open source project and, genuinely, because there are so many websites run on the internet with wordpress like this is a thing that is not good for the internet. Um, it's not good. Let's figure it out. Let's do, let's be better, let's be better. Guys, come on, I don't know.
57:42 - Jeff Massie (Co-host)
I don't know what else to tell you well, and you know you kind of hit the nail on head too. Some of that is you got to protect your copyright because I remember your trademark, yeah, yeah, this is. This is probably gosh. Maybe even 30 years ago now, fmf, the motorcycle uh parts company, flying machine factory Factory they had an exhaust pipe that was called a Fat Boy. They got in trouble from Harley because they said well, we've already got a trademark, we got to defend this, you can't use this name. So they wound up changing the name, but it was like nobody's confusing a dirt bike motorcycle exhaust pipe with an entire Harley Davidson motorcycle. But it came down to well, we have to defend it, because if we don't, then it weakens the trademark.
58:30 - Jonathan Bennett (Host)
It weakens the legal claim to the trademark.
58:32 - Jeff Massie (Co-host)
Yeah, but, yeah, there could, I don't know. At the very least, I probably would have looked at the 8% and went, oh, maybe four, and we'll whatever. Yeah, I mean, that's what the lawyers do, right? Or a lot of those people making those contracts. You start going no, no, you don't take the first one thrown across the bow, you go well, tell you what? Let's blah, blah, blah, you know, and how about two percent?
58:54 - Jonathan Bennett (Host)
and we don't do anything else? How about four and a half percent? You do this.
58:58 - Jeff Massie (Co-host)
The other thing yeah, uh, four percent, but we'll link to your website and say we're not affiliated here. You can get all the little silly things.
59:12 - Ken Mcdonald (Co-host)
Any way for a lawyer to make money.
59:15 - Jonathan Bennett (Host)
Oh yes.
59:18 - Jeff Massie (Co-host)
Ego's getting away a lot of times.
59:20 - Jonathan Bennett (Host)
That's true too. That's usually not on the lawyer's side. No, Sometimes the lawyers get a bit more blame than they really should because they're doing the things that they're being told to by the people paying the money.
59:34 - Jeff Massie (Co-host)
Well, it's kind of unrelated to Linux, but just a quick little story was. I know of a big hydraulic dam that was going to get built. Of a big hydraulic dam that was going to get built and it was from a organization far away and they said, well, okay. The local people said, well, we want to put our name up here. And they said, no, you can't do that. And it shut the whole thing down. And another far away organization said, well, we'll let you put your name on there. You know, even though we own it, we get the power. Yeah, you, you put your name on there. You know, even though we own it, we get the power.
01:00:09 - Ken Mcdonald (Co-host)
Yeah, you can put your name on there.
01:00:10 - Jeff Massie (Co-host)
Okay, it went through and it has their name on it, even though it's technically owned and the power goes to a place far away. Yeah, and it was all just we want our name on there. That's the only reason it fell through.
01:00:23 - Jonathan Bennett (Host)
Oh, that's hilarious. All right, Ken, save us. Let's talk about FWEPTI something technical, something good.
01:00:30 - Ken Mcdonald (Co-host)
And something where they work together. Yeah Well, I'm going to start off by Bobby Barsoff. Marius Nestor, thank you again for another informative article. Thank you again for another informative article, this time about the latest maintenance release of FWAPD version 2.0.1. Allowing the GNOME firmware graphical front end to record devices for emulation, as well as support for saving the emulation-tag devices to the database instead of the configuration file, this makes simulating various device scenarios easier for users and developers alike. It also makes it easier to manage and track emulated hardware over time.
01:01:30
Fwapd version 2.0.1 ensures a more seamless experience by addressing several bugs affecting users. It also provides greater flexibility for package maintainers by allowing better control of the lib DRM dependency during builds. Fwpd now compiles more smoothly on Android and without Git installed. This makes it easier for developers and users on diverse platforms to install it and use it. Platforms to install it and use it. Fopd version 2.0.1 also speeds up the mechanism for fetching details about local firmware archives, as well as the FOPD tool command, by only loading engine features when required. Both articles contain a link to the project's GitHub page, where you can find more details about the changes. Also, look in our show notes for links to both articles, as well as to the Linux vendor firmware service where you can search for firmware you have so you can see if it is supported by FWAPD yeah, very cool.
01:02:52 - Jonathan Bennett (Host)
Um, I uh I'm particularly interested in that api. I think that could be pretty interesting. Uh, is it? Is it only for gnome? Is this gnome specific?
01:03:04 - Ken Mcdonald (Co-host)
uh, from what I was reading in the article and looking at through the github release notes, that's what it sounds like at the moment, so I can see where they could expand that later.
01:03:19 - Jonathan Bennett (Host)
I mean, I guess if they publish the API, there's nothing that prevents somebody else from coming along and using it.
01:03:24 - Ken Mcdonald (Co-host)
Like.
01:03:24 - Jonathan Bennett (Host)
KDE Discover, yeah, like KDE. Or, you know, even the store in Cosmic at some point, yeah. Or the Hyperland guys or whoever. Uh, so cool fun stuff. Um, jeff, let, okay, jeff I, I I've seen, I've seen stories about this.
01:03:45
Um, I have on my very long, very, very long to-do list. I have a very long to-do list, by the way. I have an extremely long, long to-do list. By the way, I have an extremely long project to-do list. One of the things on there is on my Fedora computer behind me with the AMD card in it, to try Rockium again and see if I can get it installed and see if I can get some general OpenAI stuff, llm stuff, like image generation for prompts. There are a couple of open source projects for doing that and on my very long to-do list is to try and bring that back out and actually get it working locally rather than having to kick it off to some service on the cloud. And it sounds like there's still being progress made on this AMD's Rock-M and all of that. It's still being developed and they're about to hit some big milestones, right game and all of that is it's it's still being developed and, uh, they're about to hit some big milestones, right?
01:04:36 - Jeff Massie (Co-host)
yeah, I. My list of stories here is the amd compute corner, just because there's been a lot of news lately, so I've got three stories and actually one personal uh tidbit in here as well, though. Uh, people on the discord, this won't be a surprise to, because I actually typed it out early in the week, but I'll just go ahead and throw it on the podcast for those not in the club, though you should be Join the club. Amd's been working on giving more love actually to their GPU compute stack and I have, like I said, three articles going over some of the improvements. The first one is RockM looks like it's going to be soon supporting OpenCL 3.0 soon. Opencl 3.0 came out in 2020, and NVIDIA's had support since 2021, and even Intel's been supporting it for quite a long time, several years. Currently, amd is supporting 2.1. But an AMD engineer put in the following on their GitHub ticket the ticket to support 3.0, and this is a quote from him. We are currently looking into upgrading to OpenCL 3.0. Please keep an eye out for announcements, updates in the future for more details, thanks. Now we don't have a timeline for when this is going to happen, but it is in the works. When we know, we'll let everybody know, because that's exciting news.
01:05:57
The second story is how AMD is working on compute virtualization support for Rock M slash HIP, which, just as like it sounds, they're going to allow pass-through support with the GPU to allow compute tasks with the GPU to allow compute tasks. Though looking at the article it sounds like they're going video and everything so you could do 3D gaming in a virtual machine with your AMD hardware. It looks like they're going to totally open up pass-through completely for the cards. At least that's the way it sounded in the article. Take it with a grain of salt, but it seems that's the way they're heading. They also mentioned this isn't just the OpenCL part of the spec, but the full RockMHIP compute stack, so they're not just focusing on part of it, it's the whole thing and they're currently working on getting this upstream. So again, I don't have a date, but there's a link in the article in the show notes to the presentation they gave at the XDC conference that just ended here and it was up in Montreal, canada. It gives a much more detailed explanation on how this is all going to work, how all the different levels of the driver currently work, how they'll work after the virtualization change. For those that really like to get into the details, take a look at the article in the show notes and follow the link to their presentation, because they have a lot of diagrams and graphical stack representations of how it all fits together, where everything is and how it communicates. You know, and the currently is this and will be this, and so for the for the hardcore programmers, it's all the details are there, or enough to at least get you started on a good general direction of it.
01:07:33
The third article linked in the show notes is about Red Hat and how they're going to use RHEL AI. So Red Hat Enterprise Linux AI. Now this is not the operating system itself, but an initiative about building a large language models for AI which can be used in enterprise software. So the RHEL AI now supports the RockM compute stack, but right now it's in a tech preview, so it's not ready for prime time quite yet. It's kind of this is the check this out. Let's put some final polish on it, but enterprise is always much more cautious than regular consumer. Now it supports AMD instinct MI 2010 GPUs for inference tasks and then AMD instinct MI 300 X GPUs for both training and inference AI workloads. So this is targeting more specifically their enterprise cards, the big, big server stuff. So up to this point, red Hat AI supported CPU loads and NVIDIA GPUs. So AMD is finally getting more into the game with this update.
01:08:46
Now, like I said at the start here, the final note in here isn't in the show notes but, as many longtime listeners have heard me talk, I support folding at home and, for anybody that doesn't know, this is where you take your spare CPU and GPU processing power and fold proteins to try and find cures for a wide variety of issues like many types of cancer, parkinson's, huntington's, many more, all sorts of stuff.
01:09:12
It's basically trying to find better medicine cures and understand why a lot of different afflictions happen, how they stop them and things like that. Well, I'm on another forum and I'm on on their folding team and their folding at home group and somebody from the folding at home organization was talking about the future of the project and they said that because AMD, they said, is throwing so many more resources at the Rock M slash hip software. They said they've noticed a huge increase, they're finding a lot of optimizations and they said it should double the speed that the AMD GPUs can fold Now they do mention. You know, right now AMD is still the best supported with their CUDA.
01:10:02 - Jonathan Bennett (Host)
But Rock-N-NVIDIA is still the best supported with their CUDA.
01:10:05 - Jeff Massie (Co-host)
Yeah, what did?
01:10:06 - Jonathan Bennett (Host)
I say you said AMD.
01:10:07 - Jeff Massie (Co-host)
Oh sorry. Yeah, nvidia, okay, yeah, and it's right here it says NVIDIA on the sheet.
01:10:15 - Jonathan Bennett (Host)
So I'm just so excited about AMD.
01:10:16 - Jeff Massie (Co-host)
I can't say AMD enough you know, but AMD is still the best supported with their CUDA, but Rock M, HIP is closing the gap, so it should be much more viable in the future. I mean, if you're getting double the performance out of it, that's a major step. And I'm condensing a lot of what he said, but I mean basically just reaffirmed from a different point of view that, yeah, AMD is finally taking Rock M and HIP a lot more serious and really deciding that they've got to develop that software if they're going to catch up to, you know, keep up with Intel and NVIDIA as far as, like, AI and other compute workloads. And now he didn't have an exact date again, you know just that stuff is coming and some of it's going to depend on how fast they can push out some of their updates. Some of it's going to depend on how fast they can push out some of their updates. It does look very promising though, and you know AMD, really that's the reason, you know, I catch a lot of flack from the other guys in the show here, because especially Rob, but Rob likes to give me flack, but you know I just ignore him. But that's one of the reasons I run an NVIDIA card is because of the compute and AMD can't be there. But if they can get this going, I would be more than happy to jump on an AMD GPU because I have currently my last couple CPUs have been AMD.
01:11:49
I'm not a fan. I'll just state this clear. I'm not a fanboy of any company. I kind of just look at the existing. You know, when I make an upgrade, who's got what? Let's see what's better, because I almost went Intel this last one, but I decided to jump on AMD. So it's looking good and we're going to level the playing field a little more. So definitely take a look at Folding at Home. If you're at all curious in that, Take a look at the articles and the show notes. We have a lot of details on the various things we talked about and overall the AMD future is just really looking bright.
01:12:26 - Jonathan Bennett (Host)
Yeah, absolutely. Like I said, it's been on my to-do list for a long time and hopefully I'll get time to go play with that and it'll be there. I asked a while ago the Fedora guys. I'm in a couple of chat rooms with them and I asked them like, is it possible these days to actually install HIP and AMD stuff and does it actually work now? And apparently yes, enough of it's packaged, because you can actually use a nude Fedora, hip and you know AMD stuff, and does it actually work now? And apparently yes, enough of its packaged Is you can actually use a nude fedora. So I think that's, I think that's pretty cool and yeah, it's definitely something to give.
01:13:03 - Jeff Massie (Co-host)
Give a shot and once they get even past the open CL, I mean open CL 3.0 will be good. But when they actually fully get on hip and all that and kind of go beyond open cl, well you'll you'll also see there.
01:13:17 - Jonathan Bennett (Host)
There are some projects out there to do things like run things written for cuda on open cl, and I think once those really come along and come into their own, that's when it'll it'll get even more usable. Um, I, I kind of I don't know we need Vulkan for compute.
01:13:35 - Jeff Massie (Co-host)
It's really what we need, which CUDA could turn out to be that, but I know NVIDIA is probably going to be really throwing up some walled gardens trying to keep people off of that, but the end users want a single, you know. Hey, I want to write this once and have it run everywhere.
01:13:54 - Jonathan Bennett (Host)
Nvidia would want to share nvidia does not want to share. They've proven that. I mean cuda is a closed source thing, it is an nvidia only thing well, and they actually were fighting amd a little bit about their translation code.
01:14:09 - Jeff Massie (Co-host)
Yeah, not surprised yeah, that is.
01:14:11
that is big money for NVIDIA, that is big money for NVIDIA and kind of on a side note, intel just had an announcement where they're really getting into AI, but they're not going to compete with NVIDIA at the high end, they're going for the mid-range. So they're not going to try to have the monster cards, they're going for the mid-range where N nvidia is trying to compete at the high end. So if amd wants to find a piece of that market, they've got to get their compute stack running and it all indication is that they're throwing a lot of resources at it to get it up to speed yeah, and it, from what I understand it, works like you can do things with it.
01:14:54 - Jonathan Bennett (Host)
It's just because cuda has been so uh it, cuda is the 800 pound gorilla in the room. I mean, it's been so dominant for so long. Um, it's just, it's hard for anybody, it's hard to convince people to write stuff for open cl and you know, amd has support for open, for a lot of open cl, not not 3.0 until here, recently, of course, but um, it's just, it's just such a challenge because they're coming from it's such the underdog position and and it's very optimized.
01:15:21 - Jeff Massie (Co-host)
But you know, interestingly, the folding at home person anyway now. So this is very limited use case, we'll say. But they said when nvidia went from opencl to cuda they didn't have that big of a performance increase. There. There was some, but it kind of tells you how they've been tuning their hardware and software for OpenCL already just to get the most out of it. Yeah, yeah.
01:15:44 - Jonathan Bennett (Host)
Makes sense. All right, let's see. I think I've got yeah, I've got one more story. This is something cool that is probably coming to the kernel and I thought it was really neat, so we're going to talk about it. Cool, that is probably coming to the kernel and I thought it was really neat, so we're going to talk about it. And that is proxy execution. And so you know, we talk a lot about things in the kernel and the idea of running with lower latency and being able to schedule things so that your tasks get done more efficiently. And, of course, the people working on the kernel are interested in exactly those same things and they've run into something where a task will get blocked. And so you know, you have something mutexes, which is it is a way to put a lock on something and say you know, process one locks this. No other process can touch it, because process one is doing something that has to be atomic. If you do all the entire thing, then process one will let go of it and then everything else can use it Mutex, whether that be a piece of hardware, a piece of memory. All right, so this is a thing that the kernel gives. It gives a way to do mutexes.
01:16:48
What happens when a low priority process gets a lock on something that a high priority process wants to be able to use? Well, up until this point, your system just has to wait until the low priority process gets finished with it, which in a lot of cases, is not what you want to happen. In some cases it's the exact opposite of the thing that you want to happen, because your low priority task is low priority and therefore takes it, does not priority, and therefore it does not get scheduled as often. So it may take quite some time for that to get done with what it's doing. Let go of the lock of the mutex and then your high priority task can come in and get it, so that results in things like higher system latency for the tasks that you actually care about.
01:17:31
So the patch, the thing that they're looking to add to the kernel and apparently this has been under development for a long time, because they're already up to version 12 of the preparatory patches for this Ah, what fun. But it looks like it's getting closer to actually landing and it may land in the 6.13, and what it does is it says we've got a high priority task, a high priority process waiting for this. We're going to automatically bump the prior, temporarily bump the priority of whatever is holding the lock it will inherit. It will become a proxy of the higher priority task. It will get a higher priority in the task queue to be able to get done with whatever it's doing and let go of this lock so that we can get to the high priority thing faster. In a nutshell, that is what this is Now it is apparently.
01:18:27
Of course it's getting into the scheduler code, which is always fun and exciting and very, very complicated. So there's a lot to it. But I think, just like from a system usability standpoint, I think that's really cool that they're looking at doing this, because it will. It will help things run faster on our machines. I mean, you could see feasibly things like even game frame rates get faster because of this. Obviously it's also going to make a difference for things like automotive Linux and real-time Linux. I think it's just going to be a win. I think it's pretty cool and it's a look into sort of the nerdy technical stuff that happens down inside of the kernel. So proxy execution hopefully coming soon.
01:19:12 - Jeff Massie (Co-host)
So proxy execution, hopefully coming soon.
01:19:20 - Jonathan Bennett (Host)
Yeah, the scheduling of work, it just has a lot to that.
01:19:22 - Ken Mcdonald (Co-host)
Yeah, oh yes, there's a lot to scheduling, no matter what kind of work you're trying to schedule.
01:19:25 - Jeff Massie (Co-host)
Well, and honestly, kind of going back to my last story when I said I had AMD CPU, the one reason I didn't go Intel was because they had the performance and efficiency cores and at the time the scheduler wasn't quite up to par, yeah. And so I said, okay, I want even core or I guess, one core type. So it was symmetric. You didn't have the, it didn't have to, the scheduler didn't have to make uh choices and that's why, also, I didn't get a uh, a 3d cache cpu like a 7900 or 7950 where they had half of it had cache, half had the extra 3d cache, half of it didn't, because I would want to avoid that kind of uh unevenness in the hardware. Now. Now, if it was the same thing, I very likely could have went 14 700k, just because the schedule now is pretty ironed out and they've got all the lookaheads and the feedbacks and whatnot.
01:20:25 - Jonathan Bennett (Host)
But something to think about you know the the at the time that you would have bought that intel chip. Is that one of the chips that could have eaten itself? Did you really dodge a bullet there? Yeah, you really dodged a bullet by going with AMD, maybe.
01:20:39 - Jeff Massie (Co-host)
Very possibly because I was looking at the 14700K and I wound up with an AMD 7900. So, depending on which one you looked at, I mean there were certain workloads, one did better than the other and whatnot, but it was one of them that ate itself. The 14700 and the 900. Both were eating themselves. Now, the 900 was worse, but the 700 was still in there, so I could have dodged a bullet.
01:21:08 - Jonathan Bennett (Host)
The same customer of mine that wanted to run their proprietary software on the M1 Mac also have a pair of computers from a very large vendor and we are still fighting with them to try to get them to RMA those computers that have dying Intel chips from exactly that. And they are dying. They're blue screening and that's the only thing we can figure out that would cause it. Yes, yeah, I think. I think the last email we got, the vendor did finally admit that yeah, that's probably what's going on. Yeah, like thanks, you couldn't have figured this out a year ago and gone ahead and replaced it then?
01:21:48 - Jeff Massie (Co-host)
yeah, it's only in all the headline. Well, but they couldn't have because even if they replaced it, they still had some of the faulty microcode. They had a lot of other problems, so it's only been in the last was it a month, two months that they actually have the fixes in there and there were several different fixes.
01:22:08
And they had that oxidation problem in there too, so they've had a rough go here lately. So and kind of a side note here, the editorial. I'm kind of a little concerned about the latest chips coming out, because they're already talking about the generation after that. So I'm wondering if these are going to be kind of like the AMD chips, where oh look we, we have way more efficient chips. It takes half the power, but it's five percent faster or seven percent faster, and people are just kind of going to go. I don't care about that and everybody's already looking to the next generation it's a tick instead of a talk.
01:22:47 - Jonathan Bennett (Host)
Yeah right, that's what intel. That's what intel used to do. They used to intentionally do that. They were on what they call a tick tock schedule, and so they would have a revision and then a big, big rebuild, and then a revision and a big rebuild.
01:22:58 - Jeff Massie (Co-host)
And it helps actually the development because you don't want to shrink your brand new design if you can help it. Because even having a standard design and then shrinking it, you're going to run into problems because of the geometry and the processing that you're doing at that tighter geometry and the smaller geometry. So if you can take the design issues away from the shrinkage issues, you know you kind of just spread the love out a little more on problem solving and it kind of makes a smoother pipeline. So that's a lot of times why they do that so that's a lot of times why they do that.
01:23:38
Would any generation they add, the things in the next generation is where they shrink it down. Yeah, yeah, literally. So. So you have the new logic on the same process node as the last generation and then, after you have that all worked out and the logic's good, it works. And plus you can then put in some improvements, like, oh we, we designed this bus, but we should have changed this circuitry so that we can have a little faster interface. And then you shrink it. You haven't made any major. You know you're not overhauling the chip like you you did from the last generation, but you refine it and then you can shrink it, because most of your problems then are going to be process related, not design related, not logic design. I mean you could have features that are too small, things that are too close that you have to change because because of the shrinkage. But I mean your, your, your design, how it works is solid.
01:24:29 - Ken Mcdonald (Co-host)
Yeah, yeah, if that makes sense interference from the electronics that's right next to it yeah, that's a big part of it.
01:24:36 - Jonathan Bennett (Host)
Oh yeah, um, it becomes a big deal. Just think about that. Like just just even a couple of years ago, but say three or four years ago, would you really have thought that amd and nvidia are the ones killing it and people are worried about intel like man that's such a departure I never would have owning intel stock right now right especially those people.
01:24:58
But I mean, you know people, we've got somebody talking in the, uh, in the, in our discord right now about how, like this, intel is in trouble and there are, there are things happening and it's like intel is not having a good time right now and there are some things happening, uh, goodness I don't know, is there in trouble?
01:25:15 - Jeff Massie (Co-host)
I mean because they they've got so much money in the bank they've got they're so big. I mean they could be in trouble for a long time before they're really in trouble.
01:25:23
Well, yes, that's true but I think now is kind of the the watershed moment where they're going to have to really and they've kind of been working on it's like okay, you've got to double down on the technology, you've got to really look at what you're doing. I mean the last ceo, it kind of interestingly he went to a different company and when he got hired on they were bragging him up and there was such a backlash against him that, uh, they had to disable comments and had to. I mean people were and they deleted the comments because they blamed him for Intel's failure. Setting it up of you know, not pushing the technology, just making a bunch of bad decisions. And now they're looking at, you know, splitting up their fabs like AMD did. Amd used to have fabs. They split them off into global foundries and now AMD proper is just a design house. They don't have the fabs and Intel's kind of looking to do the same thing.
01:26:25 - Ken Mcdonald (Co-host)
And here I thought I was thinking about starting that rumor.
01:26:30 - Jeff Massie (Co-host)
Too late. I don't know if it's a done deal yet or not, but they're if. If not, it's seriously being looked at. And and yeah, the company I work for, which you know, just a little little company I mean well, we're global, but our market cap, little compared to intel little compared to intel and our market cap is way bigger than Intel. And if you would have told me that 10 years ago. I'm like dude get help, stop drinking, stop doing drugs.
01:27:03
You need some medicine. But yeah, it's crazy and yeah, they're way bigger the amount of fabs that they have versus the company I work for. It's not even a competition, you know it's they're. They're still the 800 pound gorilla as far as what they physically own yeah goodness, yeah, all right.
01:27:27 - Jonathan Bennett (Host)
well, let's move on to some command line tips and we're gonna let ken go. And Ken, you either have something about checking the path or someone sneezes into the command line. I'm guessing we're checking the path.
01:27:41 - Ken Mcdonald (Co-host)
Yes, what Jonathan's talking about is actually another core utility. It's P-A-T-H-C-H-K. Let me go ahead and bring up my command line here. I actually just go ahead and pronounce it path check, since it checks paths and files to see if they will work on most POSIX systems. It is a great tool for checking the portability of path and files before you try creating them yes, I said before creating them.
01:28:15
For example, here I am in my temp directory and there's the file directories that are inside that I am going to do a quick check for this path using PathCheck. For those listening, I've just typed in pathcheck space a slash b, slash c, and if everything's good, you won't get anything back, which is what just happened when I hit enter. So this indicates that this path has no portability problems within Linux itself, or at least in my case Ubuntu 24.04. Now you cannot have some flags that you can use. One is a slash P, and here I'm going to type in another command slash capital P, this time, followed by A slash B, slash dash B, slash C, and when I hit enter it comes back saying that it has a leading dash and a component of the file name. So that's one of the portability checks that you can do.
01:29:42
It also will check for if you have a space character in it which can affect the portability issues in it, which can affect the portability issues and I bet y'all that are watching wondering how quickly I can type that but with the and this is a lowercase P, it checks for that in for that space. Those listening, what I've got is path check space dash, lowercase p and then in quotes a, b, c, d, e, f, g, h, a, I, j, k, l, m, n, f, u, z, slash d, e dash b, c, d, e, f, g, l, a, k, m, slash f space and then of course, the in quotes. A response to when I hit enter is path check non-portable character and then it has the single quotes around an empty space and file name and I'm not going to repeat all that for you.
01:30:51 - Jonathan Bennett (Host)
Thank you.
01:30:55 - Ken Mcdonald (Co-host)
Now you can actually combine both to do a portability check. Or you can use dash, dash portability, for example. Let's go ahead and get that space out of there and see what else it finds. When I do slash lowercase p, uppercase p, and now it tells us gives us a message that the limit is exceeded. The limit that POS posit puts is 14. In this case it's exceeded by a length of 26. And you can guess which part that has. So let's go ahead and shorten that part so that it stops at k, and then you've got the other part, after the slash is exceeded by a length of 15. So let's get rid of the de that that part starts with, and, yes, we get the and we get rid of that dash, and then it's fully portable. Yeah and all without having to actually create that directory.
01:32:25 - Jonathan Bennett (Host)
Yeah, that that can be, that's useful. Um, I, I had a. Uh, I had an issue, a mysterious issue that I had to go solve about a week ago and someone was like I create this file and then when I try to open it it says it doesn't have permissions to open it. And I got to looking at it and I looked at the file name and I looked at the folder name and it was like very long folder inside a very long named folder, inside a very long named folder. And then look at the file name itself and it's another very, very long file folder inside a very long named folder. And then look at the file name itself and there's another very, very long file name. It's like I bet this application has a 255 character limit.
01:32:59
Go looking at it and, sure enough, that's what it was, that this. This has been months that she had been having this problem with various files and we just couldn't figure it out. And finally I looked at it. I'm like that's a long name. I bet that's what's going on there. All right, let's see. Jeff, you want to take it and talk about? Well, this is the second part of your disabling package updates. All right, take it away.
01:33:23 - Jeff Massie (Co-host)
Yes, it is. Well, you know, if I'm giving you know the Red Hat, you know Yum, DNF some love, I got to give my old boy app some love too. I'm debbing at heart, but I try to show everybody some love. So for those that don't know what I'm talking about, last week we talked about holding back packages or even entire repositories from updating if, for some reason, you didn't want a package or several packages to update. There can be many reasons for this. Maybe there's a new version of a program but it isn't compatible with something else in your system, for example, maybe there's some library collisions. It can happen. So the specifics last week were for YUM and DNF, which is like Red Hat and I think SUSE as well, and this week we're hitting the Debian side of the house and APT.
01:34:16
So the first method, if you look at the article, is using apt-mark, which is used with the hold and unhold options. Hold does what it says Takes a package and it won't let it be installed. If it is installed, it won't allow it to be upgraded or removed. Unhold removes the hold and the package then can be just treated like any other package. You can use the apt-mark show hold option. So it's apt-mark space show hold no spaces option to see what packages are being held. So maybe you've been playing around with it and go wait a minute. Do I still have this on hold or not? That'll show you how to take care of it. The second method is blocking package updates using the apt preferences file. You create a no updates file and list the programs you want to put a hold on. I'll let the listener get some details on that, because there are some options you can set in the file as well which dictate how the hold is handled, how it can be customized with several different options. There's some priority ranking in there of what can override it, what won't override it. So the article goes into some nice detail about all the different customizations you can do and kind of a little too much for just a command line segment.
01:35:35
The third way is using the auto remove preference file so it won't let it be removed, so it can't be updated because it needs to remove to put the new version in. And the fourth way, excuse me, is by blacklisting the package sources list. Now this will stop updates from the entire repository. So probably not something anyone wants to do, unless you have a third-party repository that is specific to a single package and supporting files like maybe the, the chrome repository, for example yeah, I just want to lock it in and which I don't advise because you know security, updates and all that.
01:36:16
And well, I run Firefox anyway. But point you know, but something like that where you have a repository that's specific to a single, you know you don't want to pick one of the main ones and just not update half of your system. So only use that if you really know what you're doing. And the fifth and final way is when you manually update the software in your sudo apt upgrade, use the no upgrade option. So with the name of the package you don't want it changed, you don't want changed. So, for example, your command line might look like sudo apt upgrade dash, dash, no dash, upgrade space Apache 2, which would then upgrade everything except Apache 2. So take a look at the article linked in the show notes for all the details and specifics on locking a package and happy updating.
01:37:17 - Jonathan Bennett (Host)
Yeah, so Keith Keith512 in the Discord chat has a pretty clever idea here and that is that you can do apt mark, unhold and then a space and then the dollar sign parentheses to. You know you run another command and they get the output and you can just run the ARP, the apt mark show hold command there, and that'll let you go through and unhold everything, which is a definitely pretty interesting way to go about that.
01:37:46 - Jeff Massie (Co-host)
Yeah, that's, and that's a good idea of if you have, uh, you're, oh, I got a lot of them, I got to unhold and you don't have to put them in individually. So, yeah, good job, absolutely.
01:37:57 - Jonathan Bennett (Host)
All right, I'm going to talk very, very briefly about net cat, which is kind of a Swiss army chainsaw of doing stuff with the network. And uh, mine is is really, really trivial what we're doing with net cat and that is we're sending a packet with it and so I've got I don't remember if I mentioned this last week on the show, but I've got a new Bamboo mini printer over that way and been doing a little bit of printing with it and had a lot of fun with it. The software, the Bamboo Lab software that you run on Linux. While it's open source and it's great that they open sourced it it has some pain points and thanks to I think it was in our Discord one of the things someone said when I mentioned that I was getting it. Really it got me thinking you know, I should probably make sure and have that on a segregated network, and so I've got it on my guest network and then I've got my desktop can only get in through to it. Well, when you do all that, you make things like your MDNS, so your auto-conf stuff, the automatic configuration things, will tend to no longer work because you're not on the same network and in the software, in the slicing software.
01:39:10
It does not have a field to just give it an IP address. You can't say, hey, bamboo Labs, here's the IP address of the printer, go talk to it. It doesn't work that way. So what it does is it's waiting for one of these special MDNS packets to come in. This took me a while to figure out how to do it, and the answer is and I don't have the whole thing here, I just have a snippet from it but you can run a netcat command. So what you do is you echo your and your string is going to be something like, you know, http slash 1.1, 200, okay, and then the rest of it. You pipe that into NC, which is netcat, and you just say send this.
01:39:50
In my case it's UDP, here's the IP address, and then you give it the right port number at the end, and so it's using Netcat, just like this little connect point A to point B across the network Super useful for this and a whole bunch of other things, like the sky is the limit with Netcat. I think I could do probably a month or two's worth of tips on just Netcat, but I've got a link to it to some more information on netcat, some of the fun things you can do with it, and it is definitely a tool to have in your toolbox. Um yeah, good stuff. All right, I think we have covered the uh, the stories and the tips. We're gonna let each of the guys get in whatever they want to plug here.
01:40:31 - Ken Mcdonald (Co-host)
At the end we'll let ken go first well, I want to go back to my command line tip. I do recommend, uh, looking in the show notes, uh, linked to an article by robert arrow elder. Uh, that also goes over some things I didn't touch on as far as using path check. Also, I do want to ask if you like doubling your money and getting swag, then I do recommend reading the Foss Force article. I've got linked at down under the ending notes in our show notes. Then, after you've read it, share it with someone.
01:41:13 - Jonathan Bennett (Host)
All right, tor is looking to raise money. Yeah, good to know. All right, jeff. Anything you want to plug?
01:41:22 - Jeff Massie (Co-host)
Yeah, I got something. So sorry, no poetry this week, but you can catch me on this week's Floss or this past week's Floss. Weekly episode 805 Mastodon Bring your Own Algorithm. Week's Floss Weekly, episode 805 Mastodon bring your own algorithm. So Jonathan and I talked to Andy Piper from Mastodon and had a wonderful conversation and especially if you're not really into Mastodon or you're not quite sure about it, I would say check it out, because I didn't know a lot about it either. So I've got some of those basic questions and Andy did a great job of talking about it and where to go, how to get started, and I was going to plug. I've got a mastodon now, but I'm still waiting for the approval on twitsocial. So otherwise I'd say, yes, you can find me at. You know I'd have a cute, cute Rob segue thing, but alas I am not approved yet. So do you know what your handle is going to be? Uh, I think it's just Jeff Massey at twitsocial.
01:42:21 - Jonathan Bennett (Host)
Okay, we will. We will watch for it and as soon as that gets approved, as soon as somebody realizes that there is a request waiting for them to go, punch the button on. Uh, I'm sure that's all it is.
01:42:31 - Jeff Massie (Co-host)
Yeah, I figure in a few days if it hits.
01:42:33 - Ken Mcdonald (Co-host)
when it hits a week, I'll just go hey, you know, leo moved that server button into the house or it's on a vm somewhere.
01:42:42 - Jonathan Bennett (Host)
You know, maybe it's digital ocean, one of our sponsors. I think they're still sponsored. They've been a sponsor, at least you know. It may be on digital ocean, or maybe on aws, or maybe it's in the house, who knows?
01:42:51 - Jeff Massie (Co-host)
or worst comes to worst. If nobody gets to it, I'll just go over to hackadaysocial and I'll go hang out over there.
01:42:58 - Jonathan Bennett (Host)
That works. Yeah, that's the cool thing about Mastodon it doesn't matter all that much which server you're on, although we did there was a, so obviously we posted the links and Andy posted the links to Mastodon and people were talking about it. And one of the things that kind of humored me that people talked about about it is like you guys sort of downplayed which server you should be on and some people, because they are often targets of harassment, really need to be careful about what server that they get on. And I'm like, okay, yeah, that's probably fair. Um, that's not something that, like everybody needs to worry about. But if that is something, if you have a See, I should have put this If you either have problems with it or have a very low tolerance for a lack of moderation, then, yes, do be very careful about which Mastodon server that you jump on. So that is fair to point out, poggle.
01:43:48 - Ken Mcdonald (Co-host)
Or run up your own.
01:43:50 - Jonathan Bennett (Host)
Well, if the lack of moderation is what you're worried about, then hosting your own is probably not going to help, because then it's all you, man. All that moderation is on you. But if over-moderation is something that really drives you nuts, then, yes, host your own, and that will absolutely solve that problem for you. And that's the thing I love about Mastodon is if you get to control your own experience, you're not at the whims of anybody. You're not at the whims of so like. On Reddit, you're at the whims of the Reddit administrator for whatever group you're a part of, and the corporate structure of Reddit itself. On Twitter, you're at the whims of the people running Twitter, whether you like them or not. You're still at their whims To Facebook, every other social media platform. But on Mastodon, you can literally host it yourself and you are in charge of your own destiny.
01:44:36 - Jeff Massie (Co-host)
And it's just the coolest thing. That's why the title is Bring your Own Algorithm. That's what it was about. Because you're in charge or harassment-wise, just do like I do and not be famous, and then it just, you know, you just fly under the radar. It's okay, it can be safer sometime.
01:44:55 - Jonathan Bennett (Host)
Yes, that is true. All right, thank you guys both for being here so much. I appreciate it very, very much. I will mention Hackaday, right over there. We've got the security column goes live Friday mornings and, of course, floss Weekly live on Tuesday and the article gets posted on Wednesdays. And then, of course, if you're not a part of it, you should scan the QR code and join the club. Get in to Club Twit. It's about the price of a cup of coffee per month and it is definitely worth it. You get access well to the video of this show, if you don't have that already. You get the ad-free version of the shows, you get access to the Discord and you get to show your support for Twit, and that is definitely well worth it. So join the club. We appreciate everybody being here and we will see you next week on the Untitled Linux Show.
