LINUX Unplugged - 312: What Modern Linux Looks Like
Episode Date: July 31, 2019Manjaro takes significant steps to stand out, and the shared problem major distributions are trying to solve, and why it will shape the future of Linux. Plus macOS apps on Linux, and our first impress...ions of the Raspberry Pi 4. Special Guests: Alex Kretzschmar, Drew DeVore, Martin Wimpress, Neal Gompa, and Philip Muller.
Transcript
Discussion (0)
Now that you and I are ThinkPad users, I'm always on the lookout for like the ultimate portable ThinkPad.
Of course.
I love our T480s, but you just got to keep an eye out.
Well, how about the ThinkTiny?
It's a miniature laptop with a 0.96-inch display and a design that's heavily inspired by the ThinkPad.
It even has a track point in the little middle kind of fake keyboard area.
Yeah, it's exactly as cute as you
think. Now, to make it, Paul Klinger, the guy behind this, created a 3D printed case, custom,
and then crammed a 128 by 64 pixel OLED display inside there. As long as an ATtiny 1614 micro
controller and a 300 milliamp battery. Now, despite choosing all those small components,
he still had to make some modifications because even with the tiny screen,
the circuit board was still too big. Had to saw that right off.
Oh, sure.
You can find instructions, code, and all the design on his GitHub.
This is actually also his second tiny PC. The first was a miniature desktop PC replica with full RGB lighting,
of course. This is really pretty impressive. Now, I don't know that I could get very much
done on something that small. Maybe meet in the middle.
Hello, friends, and welcome into Linux Unplugged.
My name is Chris.
My name is Wes.
Hello Wes.
I didn't mean to catch you mid-sip.
You know.
I mean, I'm a thirsty guy sometimes, Chris.
We're working hard today because we have a huge episode.
Yeah, you're not lying.
Wes and I got a real early crack at it today because there's a lot to get into.
We've got some new hardware in-house.
We have a very special guest joining us and lots of news to get into, as well as a look at Fedora Core OS and Ubuntu Core.
And maybe we'll try to differentiate what those real products are. Sometimes we've made
the mistake here on the old show of conflating, Wes.
Oh yeah, but Drew's here to help us set it straight.
He's going to help us unconfigurate our conflations.
Finally.
And we'll be able to more appreciate how these two products differentiate.
You said it, not me.
So I mentioned we have a special guest this week.
Phil, one of the original founders of the Manjaro Project, is joining us in the virtual
lug today.
Hello, Phil.
Hello, everyone.
Hello.
Thank you for joining us.
And also, big time appropriate greetings to the rest of that virtual lug.
Hello, everybody.
Hello.
Hello.
Aloha.
Greetings.
Wow.
Impressive.
Impressive.
Wow, we got Brent and Byte and Niels in there, Cubicle Nate, Mini, Mech, of course, Phil,
CM, and Wimpy.
But joining us here on the studio line as well to help us break all this down is Alex,
Cheese, and Drew.
It's a huge show today.
Hey, guys.
Hey, what's up, dudes?
Hello.
It's a big show.
This is going to be a lot to pull off, but we're going to try to do it.
With everyone's help, I think we can.
So let's start with a warm-up story just to help us ease into this thing.
The Darling Project has picked up new contributors and is seeing new code contributed to the project.
You might remember this.
It's a longstanding, although it's been idle for years,
effort to allow macOS binaries to run on Linux.
Think wine, but for macOS applications.
Well, this summer, there's been a lot of progress moving along,
including some new developers. Yeah, that's right.
They just published their quarter two highlights.
And some of those highlights are things like new contributors coming on board
and making progress all over the stack.
They've been stubbing out more frameworks like AGL, Carbon, Address Book,
course services, and application services,
and improving app kit support, including stuff like nested frameworks.
There's also a fix for 32-bit application support
and a bunch of other low-level bug fixes. And maybe that doesn't sound exciting, but you need
all of that if you want any of this to work. Well, and you probably maybe were a little surprised
to your 32-bit support in there. Yeah. Yeah, that's actually what they think is going to be
one of their secret sauces, one of their selling points, it will be for people that want to continue to run 32-bit Mac software
after Apple goes all 64-bit with macOS.
Okay, hang on.
Right.
Roll back a second there.
So the 32-bit support in macOS,
is that not a constraint on the kernel?
It's purely a constraint on user space.
Yeah, Apple has not shipped a 32-bit Mac OS environment for several years now.
They dropped all the 32-bit Mac hardware support back in Lion, I think, so 10.5, which was
five or so years ago, I think.
It was when I was in college.
So they dropped it quite a long time ago.
And the version that's about to come out
drops the ability to even execute 32-bit applications, correct?
Right. They're dropping all the user libraries,
all the user applications,
all of the remaining shims that are in the kernel
for supporting 32-bit Mako ABI,
the binary loader for 32-bit Mako objects is getting removed.
All that goes away in Catalina 10.15.
Yeah, right.
Okay, okay, right.
So let me get this straight.
At a kernel level, executing 32-bit binaries is going away?
Right.
On a major commercial platform, correct.
That's fine.
That's their choice. But I'm just trying to understand how the heck this Darling project
then executes 32-bit applications when there's no kernel support for it. Darling runs on Linux.
It's akin to Wine. Oh, okay. There we go. There's my misunderstanding. So their bet is that people
will move to Linux as well, which is implicit in the statement that it is going to be their selling factor,
is that they'll also get people to switch.
Does this mean that it opens up the avenue for possibly having my Adobe suite on Linux?
I mean, that's the goal, right?
I mean, when you hear this, you think graphical applications.
As I recall, with the Darling project, initially it's just been command line applications,
but there are AppKit implementations happening
and they are working on moving things
like Carbon, core services, and support for AGL.
So actually it does seem to be
that the end goal is the full graphical stack.
I don't think it's possible.
There's just a lot to do there.
Yeah.
The second line on their GitHub page says,
please note that no GUI applications are supported at the moment.
Right.
And you can appreciate that Wine was successful to a degree
because Windows was stagnant for so many years.
And macOS doesn't really have that same characteristic.
It's a much more forward-moving operating system,
and they quickly deprecate support for things
that Microsoft is much slower to deprecate support for,
and Wine has been successful to a degree because of that.
And they built up the expectation, right?
They're just going to do it, and you have to live with it.
Yeah, it's a whole other ballgame.
As a former Mac developer,
one of the big things you kind of get educated on
as you do software development for macOS
is that you
expect apple to break you and so over time you just learn to adapt to keep rebuilding things and
changing things and updating things it's very similar to the linux world which is why you know
i get very confused when people are like well you know linux is changing so much and it breaks me
all the time it's like and as a mac it's like, but this is the same experience.
Similar on mobile as well, isn't it?
Less so, actually, because of the versioned SDK stuff.
So the versioned SDK stuff tends to hide that churn to a lesser extent.
Also, that SDK story on Android is fascinating, right?
Do you know what version of the SDK,
or rather, what version of Android
shipped the SDK that 99% of developers' target is?
Is it a 5 series?
You would be wishing it's Android 4.4 is the recommended target to reach 99% of Android devices right now,
as prescribed via Android Studio if you start creating a new app today.
I think that we should all make those businesses fail.
That's really awful. That's pretty awful.
Oh, man.
All right. Well, the ARM platform is, well, is maybe the future.
I don't know, actually.
That's remaining to be seen.
But I have in studio right now the Raspberry Pi 4 desktop kit.
I picked this up right when it was announced.
It comes with a Raspberry Pi 4, 4 gigabytes of RAM, a keyboard and a mouse,
the Raspberry Pi Edition. And they are cute.
And they are actually pretty great.
It comes with an official Raspberry Pi
beginner guidebook, which is a nice
book. I don't know if you got a chance to see it out there. Yeah, I was thumbing through
there. I'm impressed. And it comes with Raspbian
on a microSD 16 gig card,
so it's pretty tight.
And it comes with two HDMI
cables, micro to regular size, as well as the USB-C power brick.
And the whole kit is $120, I think.
And it arrived a couple of days ago, and we set it up this morning in the studio.
So we've been testing it all morning, and it's running a series of punishing Foronix benchmarks right now.
Oh, yeah.
We are doing all the different compiling, and we're going to test the disk IEO.
You did do some
early wireless testing.
Yeah, we just had it
set up on the wireless
here in the studio.
So we did some
basic iPerf tests,
and they weren't too bad.
Now, obviously,
this is all taking place
in our environment.
As a comparison,
I also ran those tests
on my ThinkPad,
which had about
double the throughput.
Yeah.
And we were seeing
around 80 megabits over the Wi-Fi,
which is a pretty nice upgrade from the previous.
And I found it sufficient to browse a large remote directory
and copy over a movie file and things like that.
Right, it may not be the fastest thing in the world,
but it was definitely workable as a desktop.
So I'm going to give it a go for a week, starting after this show.
I'm going to run it as my workstation for a week,
and then I'll report back next week
with how things went.
Because right now it's just really early days,
super early impression.
One thing that I guess I just maybe didn't expect,
but perhaps should have known,
is that the Raspbian install
that ships on that thing is 32-bit.
So that one of the selling points,
obviously, of the Raspberry Pi 4 is that it's got
that sweet 64-bit
4-core processor.
But Raspbian itself is 32-bit.
And I went through my land notes when
the story was announced, and I did have it noted in there
that the OS was 32-bit,
but I'd forgotten. So when it arrived
and I booted it up, I thought I'd be running
64-bit Linux because the selling point
is a 64-bit processor.
That is not the case.
And I think you might have figured out why.
Yeah, well, I found a great blog post by James Chambers
going through how to get Ubuntu server up and running on the Pi 4
while support is still improving, let's say.
And I guess there's a problem with the SD card driver.
So the RAM is limited to 1 gig in the 64-bit version.
It does otherwise work all right.
You can get it up and booting, but if you paid for four gigs, you probably want four gigs.
And, of course, 32-bit OSs can address four gigs of RAM, so you're fine.
I mean, it's functionally fine.
There's no real major downside as far as I can see.
It just wasn't what I expected because it was kind of advertised as a 64-bit product.
And it might be something I experiment with is trying other OSs on there. Wimpy,
have you done any experiments with Ubuntu Mate on the Pi 4?
Not enough to qualify a reasonable debate at the moment, no.
I'm looking forward to running that on there. I think ultimately that'll be my desktop for
that system. But right now I'll try Raspbian.
So I've had a number of projects going on at the moment. We spoke about them last week, which was mostly around the UMPC devices.
And we've got some news about another one of those tomorrow. And I'm about to head off on
two weeks of vacation and I'll be taking the Pi 4 with me. So in the evenings when I've
tucked everyone in bed, you know, I'll have a bit of a tinker
and hopefully I'll have something
when I get back in a few weeks' time.
Great.
Well, in the meantime, I'm going to enjoy Raspbian
and we'll have a link to the guide
to get 1804.2 installed on there,
including the information about that.
I guess you said it's a SATA driver bug?
SD card driver.
Ah, okay.
You wouldn't be using one of those, would you?
No, of course not.
No, how would I?
No, of course not.
I do love SD cards on the Raspberry Pi because it is 100% floppy disk.
It's so floppy disk.
It feels like it, yeah.
And the way you eject it from your computer and load it in, it's just kind of fun.
Ah, it takes me back.
It takes me back.
Yeah.
It takes me back.
It takes me back. Yeah.
Since the certain models of the Pi 2 and the Pi 3 have had USB booting,
I've kind of been persuaded by that I use USB devices now.
Yeah.
You were seeing that folks are using eSATA to USB devices to get large storage.
USB 3 to SATA and then adding solid-state drives.
At least that's what James Chambers suggests, in fact. Okay. And then they're using that on the Pi 4 to get decent large storage. USB 3 to SATA and then adding solid state drives. At least that's what James Chambers suggests
in fact. Okay, and then they're using that
on the Pi 4 to get decent large storage.
It's fast. There's some
options there. It's fun. Yeah, it'd be
interesting to see just how far you can push
this new beefier Pi. Yeah.
But also I wonder if I could get the thing
on wired Ethernet,
would it maybe be faster to use
iSCSI or something of that nature?
That's an interesting benchmark,
isn't it? Yeah. Right now, I'm
running just the standard Pharonix test suite benchmarks.
Love that tool. You know, we're going to have to do
some audio testing on there, too.
Oh, yeah. Because really,
that's what we're going to end up using these things for,
is solving different problems all around the studio.
Well, so let's
chat with Phil, because today, as we record, there's a pretty big story that's circulating around the Manjaro community and the rest of the web.
And that is Manjaro's announcement that they have struck a partnership to ship the closed-sourced free Office Suite by default in Manjaro Linux.
And I had a chance to take a look at it, because I'd never really messed around with it, so I'll share some thoughts on there. But I thought, let's get Phil on to just get an idea
of what's going on from the top of the project on this. So Phil, thanks again for coming on the show
and welcome for your first time on Linux Unplugged. No problem.
So in your own words, characterize today's announcement.
Well, it was announced that we have a partnership with SoftMaker, who is the producer of FreeOffice and a bunch of other open source products,
or in the case, also proprietary products.
They ship out in several systems like macOS, Windows, and also on Linux.
And it features pretty, at least from what I can tell,
pretty sweet compatibility with Microsoft Office and other commercial Office products.
And it has a similar look and feel to commercial Microsoft Desktop Office.
That's exactly because they started with Windows, then moved over to macOS, and finally got also to Linux.
And we got approached a couple of weeks ago from them directly to see how we can include free offers in our distribution.
What appealed to you? I mean, what was the big motivating factor?
Was part of it the better compatibility, the similarity to what users migrating to Manjaro might already be familiar with?
Well, it's a German company, for instance, and most of us are also from Germany.
So it's just a small hub to go to them directly.
And with the talks we had with them,
they offered to adjust the free office software
and even the other products they might sell
for our Linux distribution to make it more compatible.
And if we have some issues with that,
we can present to them directly
and have a direct line to their developers to fix issues if we find something.
Also, they are open to change whatever we might need.
So essentially there is a business relationship or a real actual partnership between you and the makers of FreeOffice in such that you have a channel to give
feedback and whatnot yes exactly so we start with a free office and then we see how that works and
it was a test balloon with the release candidate five and six to see a how does it work how does
it look so soft maker also checked it from their side if it fits their needs as well.
And then I announced it with RC6 to the public as well.
And now we are in a situation that we have A, people who like it, and B, people who don't dislike it.
That's to be expected, I suppose.
Was there a monetary exchange between the Manjaro project and Softmaker for this partnership?
No, it's just the beginning.
So they also have the Softmaker office, which might be then in consideration to see what we can do with that.
But for the beginning, it's just the partnership of we can say what is wrong or what we can improve and they will fix it and such.
And based on that, we can build on top of it.
That's really great that they are willing to work with you at that level
because your shipping,
Manjaro has always been a very practical distribution,
solving problems for the end users
that other distributions don't necessarily solve.
And this is yet another one, making that a smoother experience.
And the type of user that needs Office software isn't necessarily generally the type of user
that is going to be particularly savvy at solving compatibility issues and whatnot.
So this seems like a very practical choice for the end users.
And at the same time, you now get a channel, you get a
partnership with this company to help guide feedback, which seems like a very, very good
opportunity for the project overall. So I have to probably address what the big criticism has been
so far, which is, well, why do this when LibreOffice is perfectly capable and it's free software. So therefore, by default, that's what
we all should be supporting. Well, it's a default. And either way, we will stack out of the stack
and say, hey, we will be bold and try something different as the rest. So we will be noticed.
This is one of the considerations we did. And also the company is uh 30 years in the business and they
make a living out of the office uh sellings and they're willing to make it happen for several
platforms and willing to include external considerations to their products and that is
one of the consideration we also made in our decision-making,
that the world is changing.
And if a company sees Linux as a platform,
it's a plus.
Right, it's a recognition of desktop Linux
in a sense, isn't it?
Exactly.
And it shows also to us
that the Manjaro is on the map somehow
and widely used.
And we were really humbled
that a German company is approaching us
since we're still a hobby project. So not everybody achieved that.
Wimpy, as a district maintainer yourself, I'm curious on your thoughts about the inclusion
of LibreOffice as the default because it's the free version.
I'm somewhat surprised and a little bit dismayed at the community reaction to this, because there are many Linux distributions that ship by default, for example, Google Chrome and Steam.
NVIDIA drivers. That one, I'm not prepared to put in the same category, to be honest with you, because that's
all about enabling, you know, absolute functionality in the device that you own.
All right.
That's fair, actually.
I think that's a really good point.
That is a good distinction.
It's about fundamentally enabling the hardware that you paid good money for.
Whereas, you know, if you've got Firefox available, why Google Chrome?
you know if you've got firefox available why google chrome and if you've got super tux car available why steam you know or why you know why not hio also proprietary or why not um you know
games from humble bundle so i think there's a little bit of um hypocrisy going on here because there seems to be an application class which I'm not prepared
to accept that this is proprietary but then there are other applications which is oh that's
obviously going to be proprietary or that's fine I want the functionality that that thing gives me
and you get a free pass on that so I can understand where Phil's coming from.
Phil is trying to build the future for Manjaro
and try out new things.
And we should all be trying out new things
and feel that we're in a safe place to do that.
And we shouldn't be lambasted for experimenting
or trying to improve the Linux landscape.
Because after all, since 1994, what I've been seeking to achieve
is computing parity on my Linux desktop.
And we're very close to getting there these days.
And why stand in the way of achieving that?
Yeah, well said.
I completely agree.
You're right.
It has been.
And in the case of an office
suite, a word processor, spreadsheet, whatever, that seems like the quintessential get your job
done tool. The quintessential, I don't really care. Right, right. I, you know, for me, what
really, what I really want to be free is my core libraries, my kernel, my desktop environment. I always want to have a free browser
option forever. I always want to have something on my system that's a free browser. That's always
a big requirement of mine. I start to get less picky at the more like a layer seven type basic
applications like an Office suite application
or a VLC type video player.
I just don't really care
if that's not open source
as long as the core system,
as long as the fundamentals are free.
But we have it better than that, right?
Because we do already have
great open source implementations.
This move's not really going to,
is it going to,
are they arguing that LibreOffice
is going to be really hurt by this?
I don't think so.
I think the development is going to continue to be active and healthy over there,
and it's still a great choice if you want to use an Office suite.
Phil, this has to be a tight line to walk
because you've got two sort of conflicting positions here.
You've got folks that want to see you move the distribution forward,
and you said something there that really caught my attention.
It's a standout feature now, too.
It's a differentiator.
So it could be really good long-term for Manjaro
just because it could get more recognition.
It's going to be something that's standout different
in every single review of Manjaro now.
It's going to call that out.
When people are talking about moving over to Linux
and they ask, well, what distribution supports
my Office documents the best,
which is legitimately a workflow for people.
Manjaro's name may get mentioned now at the top of that list.
It could open the distribution up to new categories and types of users.
So it clearly could be potentially good long-term for the distribution.
At the same time, you've got people that say don't bite the hand that feeds you.
And it could upset longtime members of your community. I'm curious how
you handle that from a creator standpoint. How do you deal with that? How do you not get sort of
demotivated when these kind of conflicts arise? Well, I'm highly motivated with everything and
the team is strong behind us. The decision was not made singly by me.
It's also discussed with some of the team members,
which were included in the discussions made with Softmaker. Can you expand on that a bit?
How does that kind of process work inside the project?
Well, we are currently on the leap that Mancharo is recognized,
and that's why we will change the structure of mancharo itself because a lot of
projects are hobby projects most of us work on the side so everybody has a day-to-day job and
and top when we come home we have to work for mancharo as well So some of those aspects will change in a real manner real soon. And based on
that, we have more time to discuss with companies like SoftwareMaker and others approaching to make
Mancharo as a distribution secure for the future. I would love for you to come back and tell me more
about that when you can. Sure, we'll do that. So it's no problem for that.
The main reason we made the move with the software maker in general or general with
snaps we did in the past was canonical at Snapcraft Summit this year is that commercial
products have the problem to implement their software on the Linux basis because we are
so differentiated and splitted up. And Snaps, for example, is a baseline to make the publishers
available to say, hey, this is the standard library. If you go on that, your product will
work out. And Snapd is not only for Canonical and Ubuntu anymore.
It will work on other distros.
And we are humbled as Manjaro that we work with them
to make all of those things happen.
Sure, there's Flatpak and app images as well,
but the industry needs a standard,
and maybe this is one of the moves.
As a project, do you see the most desktop opportunity with Snaps?
It depends.
In our team and the discussion so far, Snaps has a little bit more features than Flatpaks.
Flatpaks are more from the Red Hat section.
Snaps more from Canonical.
So we are currently somehow in the middle of it.
And we have to see which side we go it's like in the
matrix the blue or the red pill any pill might be there or we don't take a pill at all because we
don't like drugs so it's always your choice yeah i mean there there is a lot of great applications
now shipping in both uh it is like where we're at now with this whole thing, Wimpy,
feels a bit like a whirlwind.
And there's more and more opportunity out there.
So Wimpy, I'm curious from your perspective,
where you think Snaps could fit in with Manjaro?
Obviously, as Phil has just alluded to,
he joined us at the Snapcraft Summit
earlier this year.
And personally, I spent a lot of time working with
phil and probably my defense of phil and manjaro earlier um betray my true feelings in that i feel
i've made a friend for life having met phil you know uh in montreal earlier this year. And that's what happens when you get a bunch of, you know,
free software developers in a room together working on stuff.
You do establish relationships and friendships.
And so, you know, I want to see Phil and Banjaro succeed.
And I would like to think that what we can bring through snaps and the relationships we have with significant ISVs will benefit Manjaro and other Linux distributions as well, not just Ubuntu. I'm in a weird position at Canonical in that I work in probably one of only two departments
where we don't actually care just about Ubuntu.
We care about the Linux ecosystem at large.
So I'm in an awkward place.
I really want to see Phil succeed.
I want to see Manjaro succeed.
And I hope that Snaps can bring something to that story for them.
Yeah, I think that's really neat
to see you guys working together.
Yeah, it is.
It is a great example of the groundwork
laid by Canonical that could,
in some scenarios, only be used
to benefit them,
but really does benefit all Linux users
who want diverse platforms.
Well, thank you for helping
just clear it up and coming on the show and talking about
it straight from the source.
I think that helps when there's a lot of different conversations going on.
I think it's good to get it directly from the project.
As things develop in the future, Phil, you are welcome to come back on here and share
how things go or other project news that comes up.
We'd love to keep the hailing frequencies open, as Wes always says.
Sure, why not?
And you're welcome to stick around for the rest of the show, too.
We will also, by the way, have a link to information about Phil
and as well as this release, which is available for testing right now.
We'll have the link in the show notes over there, which you can check out.
Rockin' XSCE 4.14 Pre-3.
Don't tell Joe.
Don't tell Joe.
He might have to go check it out.
Well, thank you, Phil, very much.
That was very insightful, and I think I'm very hopeful that it works out because it's a nice differentiator.
I like to see people changing it up a little bit, and it's neat to see differentiating in a way that's kind of new, kind of new.
I will say, too, that as a Manjaro user, personally, Phil, thanks for coming on.
First of all, thanks for starting the project.
And personally, I don't care what Office software is on my machine as long as it gets the job done.
After this news, I gave it a download, and I like it.
It's quicker than LibreOffice.
It feels lighter, and it does, you know, it has a ribbon.
It looks a lot like Office.
It looks good, yeah.
And so then we did a little digging into their revenue model.
It seems pretty safe.
It's not like creepy or anything.
They sell SoftMaker Office directly if you want to buy the commercial version, which adds a few features.
And they also have a special, like, fonts division that does a special typeface stuff that makes money.
And then they use the free Office to help people find out about the product and whatnot
in a very competitive space.
It actually makes a lot of sense.
So I think I'm going to leave it installed.
If I have to use a dock or a dock X or something,
well, I'll use this instead of Google Docs now.
You know, the funny part about all of this is
I almost never have an Office suite installed on my machine anyway.
No, I mean, I would legitimately just use Google Docs, and now I won't.
So in a way, it's sort of, I think, it's at least an upgrade in privacy.
You're back on desktop Linux.
So just a couple of items of note, as it were, in the housekeeping this week.
I want to get the word out there for Ubicon Europe,
which is going to be in Portugal in 71 days as we record.
Yeah, the October.
The October, Wes.
Sounds like fun.
10th through the 13th, 2019.
We'll have a link in the show notes if you want to go to UbiCon Europe.
I don't suppose you're going, Wimpy.
UbiCon Europe?
Absolutely.
No.
No.
I've got two talks accepted.
Yeah, I'm fully signed up.
Shocker, shocker that, shocker.
I haven't missed an UbuCon for four years.
Amazing.
Well, go say hi to Wimpy.
Go say hi to Wimpy.
We have a link in the show notes if you want to check it out.
Because you've got some time to book your trip, as it were.
Now, what you don't have time for, as you've got to just go right now as you're listening to this,
it were. Now, what you don't have time for, as you got to just go right now as you're listening to this, if you get on the day of, we very, very soon, July 31st, 11 a.m. Pacific, are launching
the next generation of free study groups. This one is the AWS Cloud Practitioner Exam Study Group.
It's a multi-part. By the time you're done, you can go take the test.
Wow.
For free.
Amazing.
Yeah. Meetup.com slash Jupybroadcasting for deets on that.
And if you want some stickers and you're going to be at DEF CON,
there is a sticker swap going on,
and rumor has it Elle will be at DEF CON with stickers.
Have you heard this rumor?
So not only do you get to hang out with Elle,
you also get some swag.
Some stickers.
Yeah, you'll get.
And there's a good chance
there's a good chance, addendum
to this rumor, it'll be show stickers.
I love those. Yeah, those are pretty good too.
Check that out. We'll have information about that sticker swap
in the show notes. If you're going to
DEF CON, you're like, what'd they just say? What about DEF CON?
We got the link in the show notes. I happen to
know that the Linux Unplugged sticker
fits very nicely on the back of
a Pixel 3, so just pro tip there.
Pro tip, yeah.
You know what we've been finding out is people love the User Air sticker
because it has a double meaning.
I mean, it's the podcast, but also it's kind of a funny saying.
It's a great-looking sticker because of Mr. Bacon.
So go get one of those Bacon-made stickers.
Yeah, we've got some new ones coming out too
that we will definitely have for the fest coming up next year
and some pretty cool stuff on the way, actually.
And we should mention, if you are in our Telegram channel
or if you're on Telegram,
Cheese has been making some, well, they're official,
but they're, like, not real stickers.
They're Telegram stickers.
They could be real stickers.
They may be real stickers in the future.
They're almost more fun somehow.
But they're Telegram stickers, and they're awesome,
and there's some really great
ones. So go get that sticker pack
too. Jesus. He just
like low-key updating it every now and then too.
He just like all of a sudden will notice like a new one will pop in
there. You're like, where did this come from?
Suddenly you can communicate all that
much better. Is there like a URL we can
how the hell do people get that? There it is.
A sticker URL that I will drop into
the show notes for everybody to grab them.
Yeah.
Boom.
There we go.
Solves that problem.
Thank you, sir.
Appreciate that.
All right.
Well, that, my friends, that is the housekeeping.
And now we're going to talk about something that's ripped right out of the headlines, as they say, Mr. Wes.
Our very own Drew has been cramming for this segment now for a couple of weeks.
Corleone Drew has been cramming for this segment now for a couple of weeks.
You may have seen the news that Fedora has announced Fedora Core OS, which is the merger of— What does that even mean?
It's kind of a confusing name, honestly.
I love it, actually.
I love it just because it brings back Fedora Core, which—
Oh, yeah.
I'm a nerd.
Yeah, so it's a new addition built specifically for running containerized workloads at scale.
That is right.
And not to be confused with Silverblue, this is not a desktop operating system.
Not at all.
And it's not meant to be.
Red Hat bought CoreOS back in early 2018.
And we've all been kind of waiting around to see what that merger is going to look like as they've been deprecating Project Atomic.
And they're getting close.
It's now in a preview.
It's a functional preview.
You can download it and run it.
And essentially, it's exactly what we thought it was going to be.
It is Core OS, which used to be Container OS, but, you know, fedorified with RPM OS tree on the base and immutable file system, automatic updating, integration with etcd, which, you know, isn't shipped by default.
You have to run it in a container, but it's built for orchestration platforms just like Container and core os after it were so alex i want to bring you
in just for a moment to help us appreciate where this sort of fits in in the product line alex's
day job is working with open shift at red hat so you probably have some good insights and where
we can kind of fit this into the product line the operating system matters less than ever so with
the world of containers you're treating your infrastructure
as cattle rather than pets and if you're not familiar with that phrase it's to do with
immutable infrastructure and essentially what that means is everything in your infrastructure
environment is treated as it can just be destroyed at any moment's notice like a cow i guess i mean it's a bit when you say
it out loud it sucks a bit but so rather than falling in love with your server and creating
a special little snowflake like you would a pet you treat it like a cattle and get rid of it and
so that ethos translates through to the the whole ethos behind core os and project atomic which is what this kind of new
version of fedora is born out of so it fits in very very closely with open shift which is red
hats enterprise distro of kubernetes and that allows um administrators of Kubernetes clusters, OpenShift in this case,
to have to basically have a turnkey version of Linux
that they can just deploy and not really have to worry.
So it has a lot of automation in it.
And essentially, it's just taking the best bits of CoreOS
and the best bits of Project Atomic,
slamming them together and
spitting out a container-friendly distro. Now, Drew, when you looked at this, was it more Fedora
than you expected? Because it sounds like a big part of this has been moving a lot of those tools
over. So I'm kind of curious about the practicals. Honestly, no, I don't really feel, you know,
it doesn't feel like a Fedora project to me. But part of it is because I had used CoreOS back when I was working in an MSP. And it really feels more like an extension of that ethos and just kind of Fedora-ified in that it is using familiar tools, right down to things like Podman and Builda.
So the thing that I really want to expand upon is Alex's servers as cattle metaphor here.
So one important thing to note is once you get one of these up and running,
One important thing to note is once you get one of these up and running, there is not really any reason to log into a server and make changes.
Instead of doing that, the intention is for you to spin up a replacement server and then kill the old server that no longer has that change. So it's very rare, I think, and it's going to be very rare
for people to use this on any kind of bare metal. It really is targeting virtual infrastructures
like KVM or AWS or anything like that, and even distributed across multiple virtualization hosts.
distributed across multiple virtualization hosts, because of the way it's built,
you can use many of these spread out across the internet and orchestrate them all together so that you have nodes in multiple data centers. And if anything, if any one service goes down,
you can automatically fail over to a completely different data center.
you can automatically fail over to a completely different data center.
It's worth mentioning that one of the core advantages, if you will,
is that it's self-updating, self-maintaining in a sense,
because the applications are all isolated in the containers.
And it uses etcd as a key value store that's shared across the cluster. I mean, it makes Beowulf clusters look adorable.
It really does.
And so it's like just at a whole new level.
All right.
Well, I'm interested, Drew.
You mentioned getting it set up.
I saw I had to use something called Ignition.
How do I go about actually using Fedora Core OS?
Right.
So when you go to do an installation,
like I'm going to use the QEMU example here,
they've got a pre-built QEMU image.
And what you can do is you can point your QEMU launch command at this ignition file,
which is not handwritten.
You've got to write it out in, I think it's YAML.
And then after you've written that,
And I think it's YAML. And then after you've written that, you use something called the Fedora Core OS Config Transpiler or FCCT to generate that ignition file.
So the ignition file is not meant to be handwritten at all.
It's really more machine readable.
And you can honestly just write it with very human readable text, which then just gets
kind of converted over. So once you spin it up, the QEMU image is going to pull your config
from that file and automatically set itself up. And that includes things like user creation, SSH key installation for that user, and systemd units.
And the whole thing is kind of developed around using systemd units to build things.
So say you want it to be a Docker host, right? Well, you can create a systemd unit in that
configuration file that calls Podman or Docker, if you want to use Docker, to pull down the image, build it, run it, the whole nine yards.
So just actually, can I touch on something there?
We've been using containers as a term,
but are they actually Docker containers,
or is this a Podman-based system?
Podman is Red Hat's thing, right?
They're a competitor to Docker?
I'm not sure I would say competitor,
because Red Hat and Docker are both in the Open Container Initiative, the OCI.
And what that is doing is trying to make a cross-compatible version of containers so that you could use Docker or Podman to accomplish the same goals.
Ah, okay.
to accomplish the same goals.
Ah, okay.
But they're using, in this case, they're using Podman,
which would essentially be the tools behind standing up the containers and managing the containers?
Well, both Docker and Podman are included in the base image for Fedora Core OS.
Now, Docker is not the preferred one.
They want you to use Podman
because Podman is not running Daemon as root the whole time.
The entire system can be run with only user permissions.
That's kind of the whole reason behind them creating Podman.
So I don't know if you are familiar, when you're running Cockpit,
I think you have to mount the Docker socket.
And that is considered a bit of a security no-no.
Because Podman is daemonless, you don't have to mount anything as root in order to
see what's going on. So it's a user space thing. That's the difference. I see. Thank you.
And the other thing is Podman was designed with Kubernetes integration in mind, whereas, you know,
Docker kind of predates the whole Kubernetes fanfare by a little bit and it's kubernetes integration is
more of an afterthought than built from the ground up would you agree with that alex yeah absolutely
i don't want to poo-poo any of the work that uh docker did in the early days because they they
closed that last 10 for containers i mean you look at look at what the FreeBSD guys had for ages with Jails,
and we started talking about containers years after they had them,
and they were smugly sat over there going,
you know, we've had this for years, guys.
But Docker closed that gap,
and now the industry as a whole has really woken up
to the engineering challenges around containers,
and Podman and Builder and scopio are the end result and there's there's a link in the show doc um
which talks about there's a there's a growing movement of moving away from this monolithic
demon which is docker which is all the plumbing and everything required um to run a container a
docker container as we always call them,
when actually they are just Linux containers.
They're not really Docker containers.
And now we're moving into a world
where you have Podman, which is Podmanager,
designed to execute containers.
You have Builder, Build A-H is how it's spelt.
And that's designed to just build containers and then you have scopio
which is designed to do things like tagging and moving images around all over the place and stuff
like that so it separates out the various concerns into separately maintainable tools which has a
huge benefit if you think about it of not needing to ship the entire thing when you only want to update your build tool.
Right, or the reverse.
If you don't need to build images on your production boxes,
you don't need that there either.
Oh, yeah, yeah.
That's a great point.
I hadn't thought of that.
So any other points, Drew?
Because one thing I have in the back of my mind is
I think I have made the mistake of kind of comparing this directly
with Ubuntu Core, and they're sort of solving different problems. They're entirely different. They are. And essentially, the way I look at
Ubuntu Core is more of an Internet of Things type workload than necessarily server host nodes,
which is what I think of for CoreOS. You'd agree with that, Wimpy?
nodes, which is what I think of for CoreOS. You'd agree with that, Wimpy? Yeah, Ubuntu Core is designed for IoT devices, and Ubuntu FireCloud Init is designed for containers and has been for
many years. So to quickly summarize it, the key difference is that Ubuntu Core is targeted more
at endpoint devices, and Fedora CoreOS is targeted more at server devices. That's essentially the gist of it.
Now, you can run Fedora CoreOS on a local device.
You can install it on bare metal.
There's just not much reason to.
It's really meant more for virtualization.
But when I spun up Ubuntu Core, the first thing that you notice is that you're not going
to use apt at all.
The first thing that you notice is that you're not going to use apt at all.
It's primarily for SnapD.
And one of the big use cases that I think is growing is for kiosks.
That's certainly one of the use cases. I mean, there is no apt in Ubuntu Core.
It's an entirely new Ubuntu for a new use case.
So everything in Ubuntu Core is a snap. The kernel is a snap,
the base operating system is a snap, and the applications that you deploy on it are snaps,
and they have to be strictly confined. So that's the only way you can go. And yes,
there are a number of digital signage organizations using Ubuntu Core to deploy their products.
Cool.
That sign may be running Linux, and we just don't even know it.
I think the thing where there is some confusion is there is, unless I'm wrong,
a common thread between Silver, Blue, Fedora Core OS, and Ubuntu Core,
and that is the updates are done in a way where you can roll back. It's sort of
self-updating. Those are sort of the common elements between all of those, right? Right.
What you're really looking at is the idea of an immutable file system, which is really becoming
an important aspect for security and stability across a lot of different distributions these days. And essentially,
what you're looking at is the base file system does not change during user operation. The only
way to change it is during a reboot by booting into a new image. And I was not finding strict
documentation on how Ubuntu handled this for Ubuntu Core.
So I'm curious, Wimpy, if you had any insight into how it does its image generation for updates.
For updates of the OS?
Yes.
Okay, so updates of Ubuntu Core behave exactly like Ubuntu updates for the cloud images,
which is every time we provide an update that patches a security vulnerability,
we automatically spin a new image.
So if you are downloading that image on that day, you get the current version.
And those images are then pushed down. So
in the case of Cloud Init, they're available via apt updates. And in the case of Ubuntu Core,
they're just atomic updates that just happen.
Right. Okay. That makes a lot of sense.
Future. Right now, Wes, you feel it?
Oh, yeah.
That's kind of great.
What's interesting is we're all arriving at the same solutions,
the same style of solutions, you know, to meet the demands of our customers.
So while we're all pursuing, you know, a slightly different tech here,
this is what modern Linux looks like now.
You know, it's not just Canonical and Red Hat doing this.
This is what enterprise and industry demands
of commercial Linux these days.
Yeah, and that's why also Manjaro is there
and looks which technology is the best to do that.
And strict confinement is also good
because most people say,
hey, in Snaps there can be malware.
But if you're in strict confinement, it's confined.
So it's only the application which might be affected,
but not the whole system.
Right. Absolutely.
The modern desktop has to have something in place for that,
especially if we really do want to have the regular, quote-unquote,
users join us over here.
We've got to keep them protected.
And how magical is it, Chris,
when you take down a server
and you have your state stored somewhere else,
bring your server back up somewhere else
and just point it at the state,
the volumes on a different storage somewhere.
And it just works.
The application runtime is separated
from the application state.
It's just magical when you do that, and it's as you left it.
It really is.
And it's neat to see things scaled down, and we're finally getting them scaled down to
work nicely, right?
So you can use them to just run a little infrastructure at home, still solving basically the same
fundamental problems that are needed when you are Facebook.
I mean, essentially what we've set up here at the studio is a little small business server
for a small team to work out of an office,
which is a studio. And so we have certain data transfer and storage requirements that might be
out of the ordinary than a small business. But what you're saying there is so true.
This enterprise scale technology that lets you deploy systems all around the globe on across
multiple different clouds is also now approachable by somebody like me. And Alex, you asked how does it feel.
I can't really equate it because you have to have been there,
but I imagine it must feel like somebody who was a professional chef
for their career, and they retired.
And a few years, like five, ten years after they retired,
the microwave was invented, right?
And then you watch how fast people can warm stuff
and it just blows your mind.
That's containers to me on
Linux.
When you got out of things, that was only just sort of starting, right?
I mean, you were in the height of the VM world.
Yes, oh, big time. Yesterday,
we destroyed all
of the containers on the server
here for a couple of minutes, and
then within minutes have them all
back up and running reattached to their their config and good to go and you weren't even worried
the first time you do that it changes your life i swear and we just did it yesterday just you know
i was like i gotta i gotta i can't even remember what we were we were oh we were uh we were doing
a dry run of the full upgrade and security update process on fedora and wanted to make sure like
everything worked and we did the SELinux troubleshooting
and we did a whole dry run on the whole thing,
which happened to be a hot run.
And it worked.
So these days, you're mad if you're not using virtualization at scale.
Ten years ago, you were mad if you were considering it.
I feel like containers are at that kind of same point, right?
Where it's reaching a critical velocity, a terminal velocity where
you're beginning to get to that point where you're crazy if you're not doing stuff this way.
Whereas, you know, five years ago, containers were just, you know, fairy dust.
One other thing I wanted to bring up is Silverblue. Silverblue as a workstation was never really meant for standard desktop work.
It was meant for people who are building containers with, I think, an eye towards the release of Fedora Core OS and the migration from Atomic into the new Core OS future.
and the migration from Atomic into the new CoreOS future.
So to me, them releasing the Silverblue workstation and getting that shipped in a very usable state
was kind of a precursor to what we're getting ready to see,
which exactly, as Alex said, is about to be all containers all the time.
Wow, you're right.
Alex said, is about to be all containers all the time.
Wow, you're right.
So Fedora Core OS will have developers running Fedora Silverblue, developing the software to run on Fedora Core OS.
That's right.
Ah, that puts it all in perspective right there.
Thank you, Drew.
Thank you for trying these out and helping us better alliterate the differences between
Ubuntu Core, Fedora Silverblue, and now Fedora Core OS.
Yeah, I mean, Chris can barely say them all.
There's no way he would have got it all right.
You know what, Drew?
We appreciate it.
Thank you very much.
Phil, thank you for coming on the show.
I look forward to hearing the future of the Manjaro project and, of course, having you back on here.
Phil, one of the founders of the Manjaro project.
Thank you very much for joining us.
Thanks.
Now, we'd love to have you join us too, jblive.tv.
We do this here show on Tuesdays.
Very soon, we're going to have a very special all-in-studio edition.
Oh, I'm excited.
I think that's in two weeks.
Is that right?
Two weeks.
So if you can, put that on your calendar.
Get the times over at jupiterbroadcasting.com slash calendar, and all of our links over at linuxunplugged.com.
Thanks for joining us.
See you next Tuesday! All right, we've got to go boat.
JBTitles.com. All right, so you got to go boat.
ChevyTitles.com All right, so you know what got me?
It's that sweet new intro Joe made me.
You caught me, I was like getting a sip of water during the regular intro
because my instincts kicked in and I thought it was longer.
Yeah, you do want to wet the palate, right?
I was trying to prepare for the show.
That's pretty good.
I resisted. I resisted. I wanted to do it.
I wanted to say one more thing.
Well, what did you want to say?
Get it in now.
There's actually a third piece to the CoreOS silver blue thing that was kind of glossed over, and that's that there's a Fedora IoT edition.
takes the CoreOS technology and puts it in an even smaller base, even more reduced and even more simplified for targeting the same use cases that Ubuntu Core does.
And again, it's a container-focused workflow because what they're going with is more of a model of,
well, you already have these OCI images.
They can be built multi-arch just like every other reasonable format can or unreasonable
format.
And so you can build these targeting ARM, targeting Power, targeting x86, MIPS, whatever
you like.
In this case, Fedora is supporting ARM and x86 with this and actually also i686 because
shockingly, and this depresses me to no end,
32-bit x86 is alive in IoT, and that hurts me really bad.
But that world is where a lot of this Fedora IoT stuff is going into,
is that they're building an even more reduced functionality,
targeted for bare metal kind of environment
for supporting these kinds of applications on there.
Because as it turns out, as Ubuntu Core also proves,
a containerized confined model for applications
can also work well in an IoT space
when the machine is sufficiently powerful,
as in the case of kiosk digital signage and stuff like that.
But the goal there would be, and who knows if it could be accomplished, but the goal would be that
the vendors could then actually maintain and update the applications and the operating systems
would be self-maintaining and we'd maybe actually live in a reality of standardized IoT devices that
have some level of security.
Bingo.
And in fact, next week at Flock,
the guy who is primarily leading the Fedora IoT edition,
Peter Robinson,
is actually going to be giving a talk about
enterprise-scale usage of Fedora IoT edition.
Wimby, I have a question for you.
How does Ubuntu Core boot from a kernel snap?
I don't, like, is there a bootloader
that can like mount a snap a kernel delivered as a snap is just a bunch of files on disk rate it's
no different from anything else it's not magical it's within the project we were a little bit
surprised when joey posted um a few months ago about the Snap Store was a snap and it can install snaps.
And everyone else was like, well, how the bloody hell does that work?
Snapception!
Yeah, right.
We were like, it's obvious how that works.
It was obvious to us, at least.
And then it was completely unobvious to everybody else.
So there are some things like this.
And it is a little bit of a paradigm shift,
but it's just, you know, snaps are not magic.
They're not voodoo.
It's just another way of getting bits on a disc somewhere.