Episode Transcript
Transcripts are displayed as originally observed. Some content, including advertisements may have changed.
Use Ctrl + F to search
0:00
I'm really excited. You know, I didn't realize that
0:02
we needed this hero. But Oracle
0:04
is swooping in to keep Linux
0:07
free and they say they can't
0:09
afford not to. You see, in 2006, out
0:12
of their goodwill and for the people
0:14
of this world, Oracle launched
0:16
a RHEL-compatible distribution that
0:19
offered support and lets you run your
0:21
applications like it were RHEL. And
0:23
they did that and they chose RHEL, they write on their blog,
0:27
because we did not want to fragment the Linux
0:29
community. And our effort to remain compatible
0:32
has been enormously successful.
0:35
And all along, Oracle has done this for us,
0:38
not so that way they could undercut
0:40
a competitor, not so that way
0:42
they could try to control the
0:44
stack in a free software world, but because they
0:47
wanted to prevent fragmentation and
0:50
they didn't want the user to suffer. And
0:52
while Oracle and IBM have compatible Linux distributions,
0:55
they have very different ideas about
0:57
the responsibilities and the open source stewards
1:01
for an operating system under the GPL version 2, Oracle
1:04
writes. And they're pretty upset about
1:06
the change that Red Hat has made. Of course,
1:09
they attribute it to IBM and they say, quote, why did IBM
1:11
make this change? Well, if you read IBM's
1:14
blog, which if I check,
1:17
links to redhat.com, but if you they say, they
1:20
write, if you read IBM's blog attempting to
1:22
explain its rationale, it boils down to, quote,
1:25
at Red Hat, thousands of people spend their time writing
1:27
code to enable new features, fixing bugs,
1:30
integrating different packages, and then supporting that
1:32
work for a long time.
1:34
We have to pay people to do that
1:36
work. Well,
1:38
Oracle writes, isn't it interesting?
1:40
IBM doesn't want to continue publicly releasing
1:42
rail source code because it has to pay its engineers,
1:45
they ask. That seems
1:47
odd, they write.
1:49
And given that Red Hat has a successful, independent
1:51
open source company,
1:52
they chose to publicly release the rail source
1:54
and pay its engineers for many years before IBM acquired
1:57
them in 2019. Something has changed. change
2:00
they imply. Something
2:02
is amiss.
2:04
But Oracle's going to stand up and they're going to fight
2:06
IBM. Oracle, they
2:08
write, will continue to release
2:11
a rel compatible distribution to the extent they
2:13
can.
2:14
They say in the path they had access to rel source code
2:16
because man that made it real easy. But
2:19
from a practical standpoint, they believe Oracle is
2:21
going to remain as compatible as they can
2:23
even through the
2:24
future. But you know, they
2:26
have some words of wisdom. They
2:28
write, if you're a Linux developer
2:30
who disagrees with IBM's actions and you believe in
2:32
the Linux freedom the way we do, we're hiring.
2:35
Come work for us, Red Hat employees, essentially what they're
2:37
saying. And then they have an observation
2:39
for ISVs. You know, places out there
2:42
that maybe run CentOS for
2:44
customers. They
2:46
write, IBM's actions are
2:49
not in your best interest.
2:51
By killing CentOS as a rel alternative and
2:53
attacking all my Linux and rocky Linux,
2:55
IBM is eliminating one way your
2:57
customers save money and make
3:00
a larger share of their wallet available
3:02
to you. If you
3:04
don't support your product on Oracle Linux, we'd
3:06
be happy to show you how easy it is. Give
3:09
your customers more choice. I
3:11
really love this paragraph. This is my favorite
3:13
in here, maybe my second favorite because it
3:17
is Oracle essentially advocating to
3:19
have everybody chip away at the very foundation
3:21
of which they built their product on top of. Stop
3:23
contributing to rel and come contribute to Oracle
3:26
Linux. Well, that's how
3:29
does that work when you base your product on rel? And
3:31
where does that lead? And then they wrap it up.
3:33
What they say a big idea.
3:35
Big idea for you, IBM, they say.
3:38
You say you don't want to have to pay all those rel developers.
3:40
Of course, they never said that. That's not what they said at all. It's
3:43
an obvious twisting of their words, but
3:45
they continue. Here's how you can
3:47
save money.
3:48
Just pull from us, become a downstream
3:50
distributor of Oracle Linux.
3:52
We'll happily take on the burden.
3:55
And that's the line that betrays their
3:57
motivation
3:58
behind this entire thing.
4:00
and shows you they are capitalizing
4:02
on the under-informed and the emotional reactions
4:05
out there. Because you couldn't do this.
4:07
There would be no Oracle Linux
4:09
to base on if there was no RHEL.
4:12
This is an absolutely preposterous
4:14
thing to say that
4:15
is leveraging low-information fools out
4:17
there to get them all riled up.
4:31
Hello, friends, and welcome back to
4:33
your weekly Linux talk show. My
4:35
name is Chris. My name is Wes. And my name
4:37
is Brent. Hello, gentlemen.
4:39
Well, coming up on the show today, the fallout
4:42
from the recent Red Hat Enterprise changes has
4:44
grown to a whole new level.
4:46
You might have just gotten a taste of that. And Fedora
4:49
is proposing telemetry collection for
4:51
Fedora Workstation 40 going forward.
4:54
Say it ain't so. It's true, Wes, but
4:56
never fear. Red Hat's Director of Software Engineering
4:58
will be joining us in a little bit to dig into the details.
5:00
And I do
5:01
indeed ask him the hard questions. And then
5:03
we'll round out the show with some boosts, some picks,
5:05
and a lot more. So first, let's say good morning to
5:08
our friends over at Tailscale. Go to
5:10
tailscale.com slash Linux Unplugged.
5:12
If you don't know, Tailscale is a mesh VPN protected
5:15
by Wirecard. You get it up and running in minutes,
5:17
builds a flat mesh network that you can
5:19
do everything on top of. No more inbound ports
5:21
for Wes, myself, or Brentley, probably. I
5:24
don't know.
5:25
Doesn't really count. Wirecard. Once
5:27
he has Starlink, though, then we're really going to find out. We love it.
5:29
It's going to change your game. So go say good morning to our
5:31
friends over at Tailscale. Tailscale.com
5:34
slash Linux Unplugged and get it for free for up to 100 devices.
5:37
And of course, time-appropriate greetings to our virtual
5:40
lug. Hello, Mumble Room. Hello, guys.
5:42
Hello. Hello, everybody.
5:47
Thank you for joining us today up there.
5:49
I think we'll need them. Now, a quick
5:51
reminder before we go too far into the show.
5:53
Brent's going to be in Berlin and a meetup on
5:56
Saturday, July 22nd. Meetup.com
5:58
slash Jupiter.
5:59
casting for that. We just want to mention it now because
6:02
it's coming up very soon by the time you hear this. Any
6:04
new details to share with us, Brent? I
6:06
think my new detail is that I'm really
6:09
highly considering doing two meetups again this
6:11
time. Last time we did that last minute, I think we
6:13
should just plan on doing that. So we've
6:16
got July 22, that's the Saturday.
6:18
And then I was thinking maybe the following
6:20
weekend, like the 20... Um, um,
6:22
Brent. The corner site. Yeah.
6:24
We don't want
6:29
to put this in the show, but I don't know if you should tell
6:31
them, right? It's like you tell the one meetup so that way
6:33
as many people show up and they're like, oh, surprise,
6:35
we got another meetup, you know, because otherwise
6:37
people might just punt the first meetup altogether.
6:40
So I don't know. Yeah, you want them to feel like they've
6:42
missed their only opportunity to see you and
6:44
then surprise. It's like the call for papers
6:46
has a deadline and they always get extended a little
6:48
bit. But like that deadline brings a lot of papers in.
6:51
So I don't know. All right. Well, anyways, back to
6:53
it. The bottom of silence. Yeah.
6:56
So meetup. Well, only one meetup for sure. Only one meetup.
6:59
Yeah. Com slash Jupiter broadcasting. You
7:01
better make it. I'll be there for two weeks, but I'm really
7:03
busy. So
7:04
yeah. Yeah. Yeah. He doesn't
7:07
want any food and chat with people and socialize.
7:09
This is no good. No good. All right. Well, let's
7:12
get into our first story this week.
7:14
And that is sort of the furthering
7:17
of the Clone Wars that we saw after
7:19
Red Hat Enterprise Linux announced that their source
7:21
RPMs would no longer directly
7:24
be available and that Red Rail source would
7:26
be available upstream via the Sentos stream.
7:28
Get hub.
7:29
We've been watching
7:31
Rocky Linux and Alma Linux and Oracle
7:33
Linux all have their responses
7:36
and we'll get into some of the updates there. But
7:38
I think one of the bigger things that happened this week is
7:41
SUSE made quite an announcement.
7:43
SUSE announced that instead of using open
7:46
SUSE, you should use their new rail fork.
7:48
And they have also announced they're going
7:50
to put $10 million of investment
7:52
behind it.
7:53
They write, quote, we have a responsibility to defend
7:56
these values. Talking about free software. This
7:59
investment will.
7:59
preserve the flow of innovation for years to come
8:02
and ensure that customers and community alike are
8:04
not subjected to vendor lock-in
8:07
and have genuine choice
8:08
as well today.
8:10
And of course, this is a hypocritical statement by SUSE
8:12
because they would probably trade their left arm
8:15
to have the market share Rell has and to have the vendor
8:17
lock-in situation Rell has. And
8:19
this is the only kind of position they have. It's
8:21
a position of weakness as
8:23
they want Rell's market share.
8:25
So they're going to start producing a Rell clone.
8:27
They write, SUSE is
8:29
committed to working with the open source community
8:31
to develop a long-term enduring compatible alternative
8:33
for Rell and CentOS users.
8:35
And SUSE plans to contribute this project to an open
8:37
source foundation, which
8:40
will provide ongoing free access
8:43
to alternative
8:44
source code. So they're going to set up a foundation,
8:47
Wes. So we're going to be hearing a lot more about this. I
8:49
see that they managed to get Edge and
8:51
AI slash ML into this announcement.
8:54
Nice work. That's the marketing
8:56
people at SUSE
8:58
getting their paychecks worth there. So
9:00
you remember a year ago or so, the
9:03
rumors of Liberty. Oh, right.
9:05
SUSE
9:06
and Liberty and limits. That was something
9:08
they were cooking up where they were considering making
9:10
a CentOS clone
9:12
sort of more akin to Rocky and Alma after
9:14
the original CentOS news. So
9:17
like, here's a path for you. Let's say, you know, you're using
9:19
SUSE, but you got some Rell boxes
9:21
there you're not really happy with. We've got an
9:23
offer right now. And they also have that patching
9:26
sort of
9:27
umbrella service that patches and covers
9:29
support for both SUSE systems and Rell
9:31
systems, which this would sort of snap in
9:33
nicely with.
9:35
And then the word that came
9:37
out from staffers that
9:40
could reveal with certain details on background.
9:42
But the word that came out was they
9:44
killed the project internally before the announcement,
9:46
like right kind of early
9:48
or kind of sort of at the late at the last
9:50
minute, I should say, not early. Because
9:54
they raised a concern that we raised on an episode of Linux
9:56
Action News when we heard the news.
9:58
And that was...
9:59
just going to create Red Hat's original problem.
10:02
They're going to have a Sentos clone that competes with their
10:04
bread and butter. And
10:06
now in this case, it's not even necessarily
10:09
going to be compatible with their main product. At least
10:11
with Sentos, there was application compatibility.
10:13
So if you wanted to
10:15
upgrade or whatever you want to call it,
10:17
step up to rel, you could
10:19
just move your applications. Hell,
10:22
there are scripts that just convert a Sentos
10:25
box into a rel box. I don't know if you should do
10:27
it, but they're out there. Right? I mean, that's the kind of compatibility
10:29
we've been talking about before. Now
10:31
we're going into an era where that won't necessarily be
10:33
the case. And with SUSE's offering, whatever it's
10:35
going to be, it's
10:36
going to be like
10:38
apples and oranges between their two
10:40
enterprise products now.
10:42
Would you wonder out of Oracle
10:44
or SUSE or some of the other ones, do any
10:46
of them have enough leverage to get some
10:49
of the value of having things certified
10:51
for their version of
10:53
the ecosystem? How do you even get it certified?
10:56
How do you do that? Well, I just mean, do
10:59
they have more of a at least background and staff
11:01
who can reach out to other places that
11:03
are certifying for a rel now? What is the certification
11:05
like
11:06
damn close to rel stamp? Or more like
11:08
more product saying works with whatever the SUSE
11:10
fork of rel is.
11:13
The other thing that leaves a bad taste in my mouth
11:15
is SUSE uses
11:18
quotes from Greg of CICU and Rocky Linux
11:21
to like emphasize their position.
11:24
And Rocky Linux doesn't
11:27
even really seem to be associated with this. In fact, Brian
11:29
Clams, the project manager at the Rocky Enterprise
11:32
Software Foundation
11:34
wrote on Hacker News that
11:36
they haven't teamed up with SUSE. They don't really know why they were quoted.
11:38
The announcement from SUSE included that CICU was
11:40
teaming up with them.
11:42
But
11:42
that doesn't seem to be the case. I mean, they're open, it seems
11:45
to using, of course, whatever SUSE
11:47
throws over the fence.
11:48
Of course they will
11:50
because then they're taking the legal liability.
11:53
And then he continues to write, Rocky is always
11:55
going to try to be a one to one compatible,
11:58
bug for bug compatible with rels that can.
12:00
Okay, we'll see how that plays out.
12:02
And then he goes on to say, and this
12:04
really shows you the group thing at Rocky in my opinion.
12:06
He says, quote,
12:08
I believe we Rocky Linux, we're
12:10
pretty specific about how we're going to get the source
12:12
in the announcement. And then he links to the announcement we've covered
12:14
two weeks ago.
12:15
We're going to get source RPMs from UBIs.
12:18
That's the cloud images that are rel
12:20
based.
12:21
We're going to spin up cloud instances, et cetera.
12:23
I always like it when
12:25
somebody's snarky about how specific they've been.
12:28
And then one of their specifics is, et
12:30
cetera.
12:31
We're going to vaguely figure out how to pull
12:33
things out of these container images.
12:35
We're going to find a cloud provider that won't shut
12:37
us down for doing this, but we don't really
12:39
know where, and we're not really going to say which one, but
12:42
I don't know why you're accusing us of not giving you specifics.
12:44
That's literally the logic here. That's
12:47
the Rocky team logic right there. Well, the next sentence
12:49
too, includes the only intentional admissions
12:51
are. Yeah.
12:53
It even says there's intentional admissions. I
12:55
know. This is the Rocky
12:57
approach.
12:58
You've got the SUSE approach. All of these
13:00
feel pretty sketch.
13:03
The SUSE one, it's a little unfortunate, right?
13:05
Because
13:06
you're never going to see Red Hat market like this. They
13:09
don't do this kind of marketing when
13:11
SUSE is down and
13:12
SUSE has been sold for the third time. You
13:15
don't see Red Hat coming out and joking about how,
13:17
or making light of the factor, doing blog
13:19
posts about how rel's been with one company this
13:22
entire time.
13:23
They don't really do that kind of marketing. So it's a little unfortunate
13:26
to see SUSE do it. And then of course, quoting Greg,
13:28
who's running Rocky Linux, who's just trying to find the
13:30
biggest loophole he possibly can to keep
13:33
this thing going.
13:35
Alma's going to take a different path. And I, I
13:37
fear, and I'm curious to know what you guys think
13:39
that perhaps, you know, they're going
13:41
to suffer in, in user share as a result.
13:44
Sounds like they're going to try to build
13:46
from SENTOS stream
13:49
and it's going to be a more active development process
13:51
for them. Looking at these different
13:53
takes, Neil, Rockies, SUSE
13:55
and Alma's, I'd be really
13:58
interested to hear your perspective.
13:59
Well, I would say that
14:02
I think people
14:04
don't understand exactly how close
14:07
Red Hat Enterprise Linux and CentOS Stream are.
14:11
I know Karl has made many statements about
14:13
this on the Fetaverse about
14:15
how close they are vis-a-vis
14:18
between a rel minor version and CentOS Stream.
14:21
Furthermore, I think he's also said at one point
14:24
that most
14:26
of the packages that are in CentOS Stream are basically
14:28
with the user space are identical to
14:31
what they're in rel. And the stuff
14:33
that changes tends to reconcile fairly
14:35
quickly between the two because
14:38
CentOS Stream is where
14:40
rel development happens. So
14:43
necessarily,
14:44
at some point, they've got to be similar,
14:47
if not the same. And more importantly,
14:49
they have to
14:51
be compatible with each other because CentOS
14:53
Stream is ultimately where rel
14:56
begins. Right, you can only
14:58
let them diverge so much before you're just making a bunch
15:00
of additional work for the next rel release, right?
15:03
Absolutely, right? And so,
15:04
I mean, the real sticking point
15:08
that I've seen from some
15:11
of the rocky folks
15:13
in particular has been the kernel, right?
15:17
And up front, yes. The kernel is
15:19
probably the one package where everything's going
15:21
to be a little different because
15:24
the kernel is very aggressively
15:26
actively maintained, even across
15:29
rel minor versions. But everything
15:31
that is in the rel kernel is in CentOS
15:33
Stream. They may not necessarily be in
15:35
the same order or the same sequence. But
15:39
if you know what to look at, it's not hard
15:41
to figure out what makes a rel kernel from
15:43
a CentOS Stream kernel. You
15:45
just have to kind of do the work. And as someone who
15:48
makes a derivative kernel
15:50
from CentOS Stream for a
15:53
distribution, the CentOS hyperscale variant,
15:56
right? I know how it works. And
15:59
I'm able to... work with it
16:01
fairly easily to
16:03
maintain a kernel for the CentOS
16:05
Hyperscale workstation and other
16:08
deliverables that are being made for CentOS Hyperscale
16:11
SIG. So I mean, if I
16:13
can do it, I'm just one person doing it on the side
16:16
in my spare time, I'm
16:17
pretty sure anyone else can too
16:20
if they paid attention to it enough. I
16:22
don't have a lot of sympathy for
16:25
those who
16:27
aren't willing to roll up the sleeves and contribute
16:30
to the community. And I
16:32
am really happy about
16:34
where Alma is going. I
16:36
mean, the situation sucks for everyone, of course, but
16:38
like, I think that this is going
16:42
to make Alma better long
16:44
term. And it's going to introduce
16:47
more participation
16:49
in the enterprise Linux ecosystem than
16:51
we've ever had in the past. And
16:54
the open contribution model within
16:56
CentOS stream, I'm looking forward to seeing
16:59
contributions come in from Alma Linux. I know for
17:01
a fact that at least one member of Alma Linux has
17:04
actually
17:04
been sending merge requests to
17:07
CentOS stream to fix
17:09
bugs and, you know, introduce
17:11
features and stuff like that. And
17:14
and we're just waiting to see them land. Yeah, what do you
17:16
think about my concern, though, that ultimately
17:19
why people have chosen these clones is because
17:23
it's the easy button and they can kick the can
17:25
down. They can have something that's rel compatible.
17:27
And that's truly what sells these clones is
17:30
that rel compatibility. That's why Rocky leans so
17:32
hard on the bug for bug compatibility. And
17:35
Alma is going to be able to say probably what application
17:38
compatibility, perhaps.
17:39
Well, so if you look at what
17:43
where these where these distributions tend to be used,
17:46
depending on what type of market segment you're in,
17:48
some of it may or may not matter as much. So
17:51
the main level of incompatibility
17:54
that Alma Linux will experience compared to
17:56
Red Hat Enterprise Linux.
17:57
Again, we don't we don't know how.
17:59
that's going to work out for them. We haven't
18:02
really seen them deal with it for the kernel,
18:05
and that's the kernel stuff. But
18:08
for most people, we're talking
18:11
about building open source
18:13
kernel module drivers for all the Linux
18:15
so that they can either enable hardware that's been disabled
18:18
in the rel kernel officially, or they
18:20
add additional features, or they're using add-on
18:22
components of some kind. These
18:25
are all able to be built against any
18:27
kernel and be tested against any kernel
18:29
for compatibility. I don't
18:33
foresee as much of a problem. The user space part
18:36
is much easier to deal with. There are very
18:38
few cases in which there's
18:41
a
18:42
binary interface breakage between
18:45
rel minor versions. Off the top of my head, I
18:47
know of two, and that is Qt, and
18:50
that is LLVM.
18:51
Most, if not all applications will work
18:53
just fine, even if they're
18:56
just primarily sourcing from Sentos stream. You're
18:58
deploying some particular proprietary app
19:00
that you can't really update or change that needs that
19:03
particular interface, and unless
19:05
you have that, maybe you're not going to run into too many problems.
19:08
Well, they have to reconcile anyway when you
19:10
go to a new minor version in rel anyway.
19:12
It's going to happen no matter what. When
19:15
you're looking at, say,
19:17
the main incompatibility issue that
19:19
people would have a problem with,
19:22
it's going to be proprietary kernel modules,
19:24
which Red Hat and the community's position
19:27
is proprietary kernel modules shouldn't
19:29
exist and aren't really supported. If
19:33
you're going to try to do weird stuff in the kernel, prepare
19:36
to get bitten. Again, I should remind
19:38
people, even Red Hat Enterprise
19:40
Linux does not guarantee ABI compatibility
19:43
at the kernel level between rel minor
19:45
versions since rel nine.
19:47
Even on rel minor versions, you have to
19:49
rebuild the kernel module. If you're
19:51
being compliant
19:53
with the kernel license, your
19:55
kernel module needs to be open source.
19:57
If it's open source, then somebody can build it. and
20:00
do their own things. So in practice,
20:03
I don't think it's as much of an issue as everyone is making
20:05
it out to be. All right, I appreciate that perspective
20:07
because that's,
20:08
I think, similar to how we see it. And-
20:11
It does make them feel much more like, you know, their
20:13
own entity. Now I'm looking forward to that.
20:15
Could be a space for some innovation there. And they're open to
20:17
that idea by reading their post. When
20:19
I look at Oracle's reaction, I look
20:21
at Rocky's, Alma's, and I look
20:23
at Sousa's. Oracle
20:26
and Sousa are standing out to me right now. Oracle
20:28
is so unbelievably disingenuous with their
20:30
posts. Like you're worried about preserving free
20:33
software
20:34
and you're worried about the custodians of the GPL2,
20:37
then open source ZFS.
20:40
GPL ZFS, if you're so damn worried,
20:43
what a bunch of liars. And
20:45
if anybody remembers what happened to Sun and
20:47
open Solaris
20:48
after Oracle bought them,
20:50
then it's laughable to say that they are the defenders
20:53
of free software.
20:54
There is so much irony and hypocrisy
20:56
in that post that I'm actually angry about
20:58
it.
20:59
And it's just unbelievable the way they're
21:01
like vultures taking
21:03
advantage of this situation. It's shameful behavior
21:06
from a company that's been around way longer
21:08
and has unfortunately never behaved the way they should.
21:11
Wouldn't it be great if this like, meant that the Oracle and
21:13
Sousa stuff in particular meant like, oh, a whole
21:15
new cluster of activity
21:17
in upstream,
21:18
you know, helping do these things, making
21:21
sense of the stream really great. Almost the only one. the
21:24
most clever way to kick the can without getting sued.
21:26
That's what they're all trying to do.
21:28
I mean, Rocky's at least being fairly transparent
21:30
about it in their kind of sales pitch way.
21:32
But almost the only one that's taking the reasonable approach
21:34
that's not only the most obviously legal one,
21:37
but may even lead to some future innovation
21:40
in RHEL because they're
21:41
gonna have their own bug
21:43
catchers now. They're gonna be doing their own testing. I
21:47
just, it's gross the way these
21:49
competitors kind of turn and capitalize
21:52
on this. And I guess that is what they do, but I
21:54
just,
21:55
Oracle has always disappointed me because
21:58
I've watched them for a while.
21:59
But Sousa. You know, with that whole preserving
22:01
software and the whole thing is just such
22:04
crap. They would love to be in Red Hat's
22:06
position. They would love to. And
22:08
I just, I don't know. Well, I think the biggest thing that
22:11
stands out for me is, again, Alma.
22:13
With Alma, they seem to be actually
22:16
showing some actions that support what
22:18
they're hoping to do, where everybody else is just kind of
22:21
making promises. And we've yet to see
22:24
it come to fruition.
22:28
Linode.com slash unplugged. Head on over there
22:30
to get $100 and 60 day credit. It's
22:32
a great way to support the show and
22:34
you can see the exciting news. Linode's
22:37
now part of Akamai. So all the tools that we love,
22:39
like the command line client that I use to snapshot
22:41
and upload to S3 object storage and more.
22:43
They're beautiful. Cloud manager dashboard, the
22:45
API that's well documented with libraries
22:48
ready to go for your favorite language. You
22:50
know, all the things that have made it easy
22:52
to deploy and scale
22:54
in the cloud. All that stuff is there. But
22:56
now it's combined with Akamai's power and
22:59
global reach and they're expanding services
23:01
to offer more cloud computing resources and
23:03
tooling while still giving you that reliable,
23:06
affordable and scalable solution for yourself,
23:08
for a community or a business of any size.
23:11
I have friends that have been using Linode for over a decade
23:13
and I've watched them scale their business and we
23:16
now have several years of experience and we've done the
23:18
same. We've been able to scale up and scale down
23:20
depending on listener demand and
23:22
everything that we put forward
23:24
is on Linode because we want it to come across
23:26
as professional and high performance and
23:29
we want to be able to compete with, you
23:31
know, larger distribution networks and now
23:33
that they're part of Akamai,
23:34
we're on that distribution network and it's part
23:36
of their global network of offerings. Data centers are going
23:39
to be expanding worldwide, giving you access
23:41
to even more resources to help you grow your business
23:43
and serve your friends, your customers, your community,
23:46
whoever it might be.
23:47
So why wait? Go experience the power of Linode now.
23:50
Akamai. Go to linode.com slash
23:52
unplug, get that $100 and learn how Linode,
23:55
now Akamai,
23:56
can help scale your applications from the cloud
23:58
all the way to the very edge.
24:00
Like I'm talking the edge, like,
24:02
you know, Brent's house all the way out there.
24:04
Go find out how and support the show. Leno.com
24:06
slash unplugged.
24:12
Well, moving from the world of Centaur
24:14
stream over to Fedora, we
24:16
want to get a little ahead of some potential
24:19
controversy out there as a proposal
24:21
to implement
24:22
quote unquote privacy preserving telemetry
24:25
in Fedora Workstation 40. So,
24:28
you know, to get the straight facts, we're
24:30
inviting Christian Schaller back to the show,
24:33
director of software engineering at Red Hat.
24:35
Well,
24:36
Christian, welcome back to the show. And
24:38
once again, you've joined us. It's been just about a year.
24:40
So it's about time. It's about time you
24:42
come back and tell us what's going on. So thanks
24:45
for being here. Well, thank you for having me. Much
24:47
appreciated. So the reason why we're chatting
24:49
this week is because there is a proposal
24:52
brewing for Fedora Workstation 40. So
24:55
in a couple releases to enable
24:57
what the project is labeling privacy
25:00
preserving telemetry.
25:02
And of course the Linux community is traditionally been
25:04
pretty sensitive when we're collecting telemetry
25:06
about our systems. So could
25:09
we maybe start with
25:11
why and then we'll
25:13
get to how you guys are going to do it differently and whatnot.
25:15
Yeah. So sort of why is that I
25:17
think we have very very often for many years now
25:20
come to conclusion that we are making a lot of decision
25:22
based on assumptions essentially.
25:24
So I mean, so I have discussions in Fedora about
25:27
like, hey, it's a time to turn off support for
25:29
an old generation of CPUs, for instance.
25:31
It's like, well, based
25:33
on, you know, what we see around us and what we
25:35
hear, we think this is fine because we
25:38
think there's nobody or not a significant
25:40
amount of people using this anymore, for instance, or
25:43
for that matter in terms of choices, right?
25:45
I mean, I think there have been so many times over the last
25:47
few years where I like, for instance, listen to your show and it's
25:49
like, hey, I use GNOME, but you know,
25:51
without that one extension is useless to me. That's
25:54
sort of like
25:54
a statement I heard quite a lot. And
25:56
then of course I am sitting there like, okay,
25:59
how representative. with that claim, how true
26:02
is it? And like, so I ended up having
26:04
to all this parse data in this, like,
26:06
and make guesstimates based on all the best experience,
26:08
like, is this relevant or is this because,
26:11
hey, like Chris really loves to do calls. So that's why
26:13
he has economic extension doing solving of Sudoku
26:15
or Persil. What about something more recent,
26:18
like the LibreOffice decision to switch from packaging
26:20
to Flatpak? Would, do you think there could be
26:22
a scenario where this telemetry data could
26:24
have been used to make a more data-driven
26:27
decision there?
26:27
Absolutely.
26:29
And LibreOffice was a hard one to really know
26:31
for because it's pre-installed in Fedora Workstation.
26:34
Right. So, so we don't know whether that's
26:36
in there. People never start it or
26:39
use it or if actually everyone was
26:41
using it. Right. So it's, and I mean, another
26:44
important thing for us, right, is that, you know, we have this long-term
26:46
vision of moving maybe towards more
26:48
something like SilverBlue, which
26:50
is an immutable core system. But
26:52
there's a lot of things in the way of doing that. Like
26:55
for instance,
26:57
there are certain printer drivers that you then have to
26:59
either do a complex job of layering onto the system
27:02
or whatever to get it in there. And
27:05
once again, with modern printers and printless
27:07
driving, no, driverless printing, sorry, that
27:11
might not be a problem. But once again, maybe there's
27:14
a hundred holes in Fedora users
27:16
who happen to have one of those older print models still
27:18
in use. So there's a lot of these things
27:20
where we're getting a system for gathering
27:22
metrics can make us help us make better
27:24
decisions basically for the Fedora community.
27:27
What about the art of it all?
27:29
Like, you know, I'll, I'll, I'll, I'll steel
27:31
man this for just for a moment. I think Fedora
27:34
workstation has turned out pretty great
27:36
and has become one of the absolute
27:38
leading edge distributions. That's really kind of
27:40
shown the way forward for a bunch
27:42
of other distributions and has made a bunch of decisions
27:44
that seemed
27:45
semi-risky at the time.
27:48
System D, pipewire, ButterFS,
27:51
would a data driven distribution have
27:53
made some of those same decisions. What about the
27:55
art of it?
27:56
Well, I mean, obviously there's certain decisions
27:59
that a few. metrics probably can't help
28:01
you with. It's like, is
28:03
making a choice to do something?
28:06
For instance, it will bring in a whole new group
28:08
of users.
28:09
The other metrics on the current
28:12
user base probably will never help you answer that question.
28:14
And
28:15
at the same time, I don't think about at least the way
28:17
we look at metrics. You
28:20
know, it would affect the things like our investment
28:22
in PyPyr as a good example. I mean,
28:24
that was partially thinking, okay, hey, how can
28:26
we make life easier for pro audio users? And
28:29
how can we stop having the
28:31
audio stack being such a
28:34
mess, basically, and at the same time also get
28:36
similar line of support for video? So
28:41
obviously, certain things, metrics will not help
28:43
us with. And that's, I guess, where the art of it comes in, where
28:45
we have to sort of make
28:47
some call about strategic and
28:50
educated decisions about where we're heading that
28:52
we can't necessarily prove a front with data.
28:55
So there'll still be some gut involved, of course.
28:58
Okay, so let's talk about, let's talk about how you're
29:00
going to do this. Because one of the things
29:02
I've noticed reading through the proposal and
29:04
then the discourse conversation,
29:06
it feels like there's
29:07
a lot of thought about how
29:10
to do it in a way that doesn't use a
29:12
third party system like Google Analytics. Sounds
29:14
like there's been a lot of thought about doing things that
29:17
don't capture identifying information,
29:20
even like things like, you
29:21
know, not looking for looking at proprietary
29:23
installed apps, because maybe somebody has
29:25
some sort of custom binary they run that might
29:28
hint at who they might be like, you guys have thought a lot
29:30
about that stuff. But I see
29:33
less thought on what you actually want to collect
29:35
as the conversation has gone on. It seems like there's kind
29:37
of been a realization of,
29:38
okay, we probably do want to track that, but we have to do
29:41
it right.
29:41
It seemed like in the initial proposal,
29:44
there was this attempt to be like, well, we think we want to watch
29:46
these things. But as the conversation has gone on,
29:49
that window seems to be expanding a little
29:51
bit. What are your thoughts on the critical
29:53
things and just sort of the way this might go? And my
29:56
initial criticism of there's been a lot of
29:58
thought put into the technical implementation. presentation,
30:00
but not a lot of thought into the actual practical
30:03
collection
30:04
so far. When we put all the proposals
30:06
over to our thought was,
30:09
let's start agreeing upon the process for
30:11
how to do it before we start digging into those.
30:13
And for us, we thought, like, you know, instead of having
30:15
a predefined list of these are the things
30:17
we have been thinking about. Let's instead focus
30:20
on the fact that let's have a community driven process where
30:22
we have elected community members who will be
30:24
part of decision making for what metrics to gather
30:26
or not so that it becomes more about the process for
30:28
how we do it, as opposed to
30:31
individual items. I mean, I agree, of
30:33
course, based on feedback, we realized that we needed to provide more
30:35
examples because people like, okay, how are you
30:37
going to check my browsing history? Or I mean,
30:39
what are you trying to do here? And so
30:42
we're trying to add that now and bring that in.
30:44
But it, you know, it's also this sort
30:46
of weird situation
30:48
when you do development in open is that on one side, people
30:50
say, Oh, why isn't your proposal more fleshed
30:52
out? But at the same time, if you come
30:55
with something that's super fleshed out, like, Oh, the decision is
30:57
already made. It isn't involved with from the beginning. Fair.
31:01
Yep. Yep.
31:02
And it seems like there
31:04
is a lot of technology
31:06
available that's thought through this a lot.
31:08
Reading through the proposal, it seems like a lot
31:10
of the tech that's going to be the back end is
31:12
going to be stuff that was created by endless
31:15
for their platform with some
31:17
modifications, it seems. And can you talk a little bit
31:19
about why endless is implementation
31:22
and how it's going to work a little bit differently for Fedora?
31:25
Yeah, so the background for it was that
31:27
quite some years ago, no, I was a goad in
31:30
Greece, and Rob McQueen, who's the head of endless,
31:32
he was actually talking about their, their
31:34
metric system. And the
31:38
topic of talk was like, how do we do metrics in a way that
31:40
isn't sort of murky to users?
31:42
And one of the things he pointed out was like, hey,
31:44
let's make sure the data captures open
31:46
source, so people can all get that code. Let's make
31:48
sure the server is open source, so people can all get that
31:50
code, and make sure you
31:53
know, it's easy to turn on and off for people so they can, you
31:55
know, he's locked out. So I thought like, oh, yeah, that's actually
31:57
so reasonable, because I think a lot of the gift reaction
31:59
people have against any kind of metrics gathering is,
32:02
I don't know what's happening on my system. There's some people looking
32:05
at what I'm doing and taking
32:07
some data out and I don't know which data this or what's
32:09
going to be done with it. So by trying to
32:11
be sort of super transparent about it and having
32:13
like everything being open sourced, everything being
32:15
hosted in a clear way and having
32:18
even a community process for deciding what to gather,
32:21
we're sort of hoping to alleviate
32:24
those fears, right? And I
32:27
think at the same time, we did some changes because
32:30
Fedora is different from Endless in certain
32:32
ways. So we felt we wanted to err
32:35
even more on the side of caution when it came to
32:37
how we did it. So for instance,
32:39
like, so that's why we did some changes to
32:41
the endless call. But at the same time, we're trying to basically work
32:43
with Endless to have a shared call base so that as
32:46
we improve this thing, it's
32:48
useful for everyone. And of course, also other distros, we
32:51
might want to
32:52
start using this can also take the same system.
32:54
One of the things that I thought was really kind of neat
32:57
and gives me a bit of peace of mind is it
33:00
sounds like it would be theoretically
33:03
possible for the user to point the metrics
33:05
collection at their own server back
33:07
end and then they could actually look at the exact
33:10
data being sent to the server.
33:12
That's neat. That's a neat implementation. So okay,
33:14
on the client side, it is going
33:16
to be on by default
33:18
for brand new installs so far that
33:20
if the proposal goes forward with workstation 40. How
33:24
will that practically work? I've got a brand
33:26
new install. I've just done my first
33:28
login. Are you collecting information about
33:30
my system? What's being sent at that point?
33:32
Yeah, so not nothing is being sent. But
33:35
the system is collecting data and we
33:37
are discussing whether it's better to wait. I
33:39
mean, it's also about like, if you wait, then maybe we
33:41
miss critical information that could be useful
33:44
to let's say, a figure of the bug that's happening on the
33:46
system, right before the first login as
33:48
an example. So what's going to happen is
33:50
that the system looks by default. And
33:53
then once you get the initial
33:57
system setup screen, it will ask you whether you want to do metrics
33:59
or not. And if you say, they know that it'll just turn off everything
34:01
and delete whatever it collected so far. Could
34:04
be, as I said, based on discussion that we end
34:07
up turning it on at that point as opposed to
34:09
having it on from the get-go. I
34:11
feel that's a minor detail, but I mean, I think for some
34:14
people it matters more. But to be
34:16
clear, the collection is on.
34:18
Right now, the proposal is the collection will be on,
34:20
but the uploading is not on
34:23
and does not get enabled until the user clears
34:25
the privacy screen where they have the checkbox to
34:27
turn it off.
34:28
Yes, correct. And
34:30
then once it's on, or say you're on a 39
34:32
system or a 38 system
34:35
and you upgraded to 40
34:36
and it's not on yet,
34:38
there's going to be a new UI created
34:40
in GNOME settings?
34:42
Yes, correct. So there will be
34:44
a UI there, people can go, I mean, let's say
34:46
you did a fresh install and then you regret
34:49
your choice later on, be that you turned it on or
34:51
off, you can go into GNOME settings and change
34:53
it. But also for people who did upgrade, they
34:56
will then be able to go there and turn it on. And
34:59
we are talking about like, whether on the road you can come
35:01
up with a good way to present the UI after
35:03
upgrade. I mean, part of the problem is at the moment we have
35:05
nothing in Fedora that would sort of easily
35:08
enable us to present the question
35:10
to users, say, do you want to open into
35:12
this
35:12
now since we have this new thing? So
35:14
that's why at the moment it's only
35:16
for fresh installs. But I mean, maybe with the online, we come
35:18
up with a way to do that in a non-annoying
35:21
way. Right. Interesting.
35:23
I mean, because I think it would be kind of,
35:26
I think it'd be kind of maybe jarring for users
35:28
that upgrade that had it off to have it turned on by default.
35:30
So I think that probably is a safe way to at least start. Okay.
35:32
So I want to, I want
35:34
to kind of talk about the overall idea and
35:37
goals here, because we have some implementation details.
35:39
They seem pretty solid. They're going to follow
35:41
a lot of the endless stuff, which they've done a really
35:44
comprehensive blog post on that is
35:46
fantastic,
35:47
that I'll put a link to in the show notes if people want those details.
35:50
So my question to you is,
35:51
if this goes forward, what are like the must have
35:53
things from a 50,000 foot level
35:57
managing the project view that you'd really like to have,
35:59
like the data points. that you don't
36:01
have that are your pain points right now?
36:03
I think it's definitely, I would like to
36:05
be able to see more about,
36:07
I guess, certain technology
36:09
choices, as we talked about, right? I want,
36:12
for instance, know, hey, when is it fine
36:14
to stop caring about a certain class or hardware, for instance.
36:18
I also want to see, you know, what shell
36:20
extensions are popular, so we can see, I'd like, do we
36:22
need to try to integrate them closer? I mean, it doesn't
36:24
necessarily be that, hey, okay, it turns
36:27
out that 60% of Fedora users use a specific
36:29
extension, so let's package that in Fedora. I mean,
36:31
that might be the output, but it could also be that Alan Day,
36:33
who's overhead designer, maybe he will
36:35
take that as input to go back to the gnome community and say,
36:38
hey, what if you try to integrate this feature
36:40
into, you know, myself, because it turns
36:42
out, you know, there's a huge subset or
36:46
even superset or users who wants
36:48
this specific feature. Another
36:52
thing could be, you know, mention
36:54
Butterfest. You know, I don't know how
36:56
many Fedora users are still on XFS versus
36:59
Butterfest, so I mean, in terms of, for instance, starting
37:01
to try to take advantage of certain Butterfest
37:03
features, I need to know certain
37:05
that, okay, no 99% are over in Butterfest,
37:08
so I can start sort of assuming almost that that's the
37:10
default for a Fedora install. There
37:12
are also things like, when
37:14
you, for instance, want to convince a new hardware partner or
37:17
not even a partner, if you want to convince a hardware maker to
37:19
start supporting the LVFS for firmware,
37:22
if I can say, hey,
37:24
I know for sure that there's 50,000
37:26
Fedora users that are going to have your hardware
37:28
and they would love to get firmware updates for that. That's
37:31
a lot stronger argument to make, both
37:33
for me to them and for them internally, than if I say, I
37:36
think there's quite a few somewhere out there who probably
37:38
uses for stuff. That data opens doors, I would
37:40
imagine, starts conversations for
37:42
sure. So, okay, so one of the other areas I saw
37:44
discussed is,
37:46
and I suppose you could look at this with a perspective
37:49
of, I'm sort of shocked this wasn't already
37:51
there, of course it's not, but I'm sort of shocked
37:53
because any other commercial software packages would
37:55
be there, GNOME Software.
37:57
There's discussions in their tracking about, well, when
37:59
does somebody actually...
37:59
click on a banner in Nome Software,
38:02
do the people actually use the featured app section
38:04
and just really, really
38:06
basic information about
38:08
just the standard stuff, how users are using
38:11
Nome Software. I never even thought that that
38:13
information wasn't available. I for some reason
38:15
thought maybe there would be like analytics
38:17
in there, but of course there's not.
38:19
And to me, it seems like
38:20
obvious that how do you really improve
38:23
something like Nome Software without having an idea
38:25
where the rough parts are. But what if, now let's just,
38:27
let's play just fun here.
38:29
What if something comes back, and I doubt this is gonna be the case, but
38:31
what if something comes back and says like,
38:33
4% of Workstation users
38:36
use Nome Software more than twice or something. Like
38:38
it's such a, could there be a moment
38:41
where
38:41
upstream at GNOME or
38:43
the Fedora project goes, you know, we're actually just gonna
38:45
stop working on something, could there be that
38:48
kind of radical shift? What do you think the ultimate
38:50
end result could be for something like that?
38:53
Well, I think definitely it could be cases
38:55
where we, for instance, realize that, hey, this
38:57
tool that we thought were something Kotash
39:00
Mark everyone was using, nobody is using. And
39:02
it would at least trigger some kind of rethink, right? I
39:04
mean, there's no point in, I mean, we are investing
39:07
quite a bit in GNOME Software in
39:09
terms of having like an engineer more or less full-time on it.
39:11
And if it turned
39:13
out that nobody was using GNOME Software, then A,
39:17
the question is, why is that? I mean,
39:19
of course, maybe we would start probably by figuring
39:21
out what, because to some degree it would feel naturally
39:23
like, okay, people probably need some software tool to
39:25
use, really, everyone just used the NF command line or
39:27
what's the story there. But
39:30
of course, worst case scenario, we might decide, okay, well,
39:32
if nobody's using it, why are we, maybe that engineer
39:34
can instead go and work on this other feature that, you know,
39:36
we've seen from the data people are desperate for, but
39:38
we currently have nobody working on.
39:40
Right. I mean, I'm sure it does, right? I'm sure
39:42
it gets used. The popup comes up, scissors updates, people
39:45
click it, it launches GNOME Software, I'm sure it gets
39:47
used. But I just thought it was an interesting thought
39:49
experiment. I guess my other thought that
39:51
I just
39:52
generally, and this is kind of going for your personal
39:54
opinion on this one, I've seen kind of a trend
39:56
where there's people and groups out
39:59
there
39:59
to capitalize and take the opposite
40:02
position of anything
40:05
associated with Red Hat does. You know what I mean? So
40:07
if Red Hat comes out and says, we're going to do this, they'll come out and
40:09
say, we're going to save the users and we're going to do
40:11
Y and come join us. Are you concerned
40:14
at all about this
40:15
metrics telemetry stuff getting
40:18
spun as anti-free software or anything like
40:20
that?
40:21
Well, yes, I
40:23
guess I am. I mean, obviously I do
40:25
not want that to happen. At the same time,
40:28
I always feel that you need to be sort of confident
40:30
enough about that you're doing the right thing. That
40:34
fear doesn't deter you from doing it because
40:36
I mean, at the end of the day, I believe that this data
40:39
will allow us to make Fedora significantly better. And
40:41
I think that will pull in a lot more users. So
40:43
for me, I think that
40:45
always, you know, any noise on
40:48
the internet, basically around it. Yeah,
40:51
it's generally noise. I think you're right. I think if
40:53
Fedora stands out and if Fedora has gotten better over
40:55
the years, and I think
40:57
we've seen it, right? It's grown in
40:59
popularity and its praise and
41:01
rightfully so, because it earned that through it through
41:03
just being a good product.
41:05
And this really is just we're talking workstation,
41:07
right? We're not necessarily talking some of the other flavors
41:09
or spins. This is just workstation right now. Yeah.
41:12
So I mean, and this, of course, is a gnome
41:14
only distribution. They designed this specifically
41:16
to track data and gnome. But that
41:18
said, of course, big chunks with are usable.
41:20
So what we have said is that if any other spins wants
41:23
this, they are free to, of course, to work on
41:25
making sure that this is integrated into
41:27
their spin and works for them. But
41:30
at the same time, I want to keep the data separate because
41:32
I guess as an example,
41:34
I mentioned in one of these discussions we had internally,
41:36
I said like,
41:38
you know, I want to, for instance, know whether, well,
41:40
I mean, what is an interesting data point for me
41:43
is, for instance, how many gnome users are
41:45
using on terminal versus console,
41:47
for instance, what is not from
41:50
a Fedora workstation perspective is important for me is
41:52
like, is whether KDE users are using console
41:54
or not. And at the same time, with that,
41:57
again, might be useful data for the KDE spin, right?
41:59
They want to verify. Oh, people are actually using over the fault
42:01
terminal application. Um, so, so
42:03
I think we will probably come up with a way that if, if
42:06
someone, the spins decides to do this, we will have
42:08
a sort of separate repository for them because some
42:10
data of course is valid across the spins,
42:12
but
42:13
you know, KDE probably couldn't give care
42:15
less about which, you know, extensions people use
42:17
or not. No, maybe they'd like to know
42:19
what plasmoids I'm using though. Cause I do have a couple.
42:22
Um, okay. So really my last question for you is
42:25
just again, to just sort of steel man this a little bit,
42:27
it's actually in totality. I think this is a great thing. And
42:30
I think you guys have a really solid implementation that respects
42:32
privacy.
42:33
But let's just say like you say you're getting a really,
42:36
there's a few data points that are coming and you got some great signal
42:38
and somebody at some point by auditing
42:41
this discovers.
42:42
Actually, I do have a way to deduce
42:45
somebody's potential identity from this
42:47
metric. Have you thought at all about a process
42:49
for how you'll
42:50
wipe that data out or if you will, if you'll,
42:52
if you're going to keep taking that signal, if you'll get rid
42:54
of that, that particular telemetry point,
42:57
what are your thoughts there?
42:58
Yeah. No, I mean, I think our thought is that if
43:00
that happens, we will basically on the database level need
43:02
to do go and do some data laundry. I have
43:05
a hard time imagining the scenario where
43:07
it will be possible. I mean, like I was
43:09
trying to think about it and like say, Hey, like if
43:12
you're happened to use an Acer laptop,
43:14
maybe you're the last person in the world to do that and
43:17
if I know you and I know we use an Acer laptop, I know
43:19
that we got that data from you, but
43:21
that means I have to already know
43:23
that you're using an Acer laptop for me to know
43:26
that that was your data point and you need
43:28
to be the last person in the world using
43:31
that Acer brand laptop. Sure. Yeah. Yeah. So,
43:34
I mean, if it happens
43:36
to you, of course, jump in and then do surgery
43:38
on the database to, to purchase, but I have a,
43:40
I have a hard time imagining the scenario
43:42
where, where this comes together like that. Yeah.
43:45
I mean, reading through it, it really seems like the number
43:47
one priority is avoiding user identifiable
43:49
information, even at the cost of losing out
43:51
on useful information.
43:53
Yeah, correct. Yeah. Well, Christian,
43:55
thanks for coming on and answering some of my questions and
43:58
keep us updated on how it goes. We'll keep an eye on.
43:59
of the discussion and Fedora Workstation 40 as
44:02
well as it develops. Thanks again. Can I
44:04
actually add one last thing before we leave? Sure,
44:06
absolutely. So one thing that came up a lot
44:08
in the internal discussion and I want to address it is people
44:10
are like, because a lot of things in Fedora
44:12
Workstation has been about should signing
44:14
up for this be something that has a, we have a, you
44:17
know, this privacy panel saying, do you want
44:19
to join or not? And then the question is, what's the default?
44:22
Should the default be joined or not to join? And
44:24
a lot of people feel that if you even suggest something,
44:27
you're doing something wrong. But I, I
44:29
ended up actually reading some
44:32
studies from Cambridge University today, who had
44:34
actually done studies on like
44:36
opt-in and why do people behave the
44:39
certain way when they see opt-in things. And the
44:42
primary thing to point out was that when people see, for
44:44
instance, a question asked, Hey, do you want
44:46
opt-in and it's the false? So yes, they take
44:49
that as a recommendation from the people making the
44:51
survey. And of course, if we
44:54
are asking people to sign up, obviously, we want them
44:56
to sign up. And that's implicit. And
44:59
then of course, the two other things I mentioned was like ease
45:01
of choice. I mean, of course, not time to go and click no,
45:04
it's a little harder than just clicking
45:07
next. But it's it
45:09
was more about fact that people like I don't care
45:11
enough. So I'll just click next as opposed to
45:13
I don't understand it. And I click next, right.
45:16
And an interesting of the last thing I mentioned was
45:18
that people tend to choose to
45:22
default just because they consider it the
45:24
status quo, basically. But
45:27
I mean, to me, the most important thing is like the
45:29
thing that really struck me was this whole thing about like people
45:34
see that choice as a recommendation from the people
45:36
asking. And I feel that like, if you're going
45:38
to ask people or have a telemetry system,
45:40
obviously, you need to believe enough in it to
45:42
feel confident about recommending our users to use
45:45
it. And so
45:47
I think we are looking at some, you know,
45:49
design tweaks to try to compromise
45:51
with people on that. But at the end of the day,
45:54
in order to also get the volume in it, because once
45:56
again, right, I don't want people if it's
45:59
sort of only the hard.
45:59
core and sub signing up, for instance, then you
46:02
sort of like, oh, so the most popular ID
46:04
is still Emacs. I never realized, you know, well,
46:07
everybody's on I3. Wow. Who knew they
46:09
switched over. I didn't know. Yeah, absolutely.
46:12
They're, yeah, they're all using tiling extensions. That's,
46:15
and I think that is a good point is you want to try to
46:17
cast a widest net as possible. And I
46:20
think the thing that gives me some peace of mind
46:22
as a user, and I don't know if I ever would, but
46:24
if at some point
46:26
I felt like I wanted to opt out,
46:28
it's just right there in gnome settings. And I can just go
46:30
check that box and that's
46:32
nice that I don't have to like invoke some sort of command
46:35
line
46:36
magic, you know, to get it to do, to turn
46:38
off some sort of service or something like that. I just go check
46:40
a box. Yeah.
46:41
Yeah. And it's worth mentioning that, you
46:43
know, Fedora used to have this opt-in tool back in day
46:46
called smolt. I never tried it. Yeah. I
46:48
was going to mention that I forgot, but yeah, I
46:50
remember it almost had like a,
46:53
God, it like, it had like a previous
46:55
era UI to it. It was really, it was an, it was
46:57
a different time back then, but it was kind of a
46:59
more primitive version, right? It was not, not
47:02
nearly as, and yeah, I imagine. And
47:04
it sort of also proved that opt-in doesn't
47:06
necessarily work because it turned out that only people who were motivated
47:09
to run it were the people who like, Oh, my system
47:11
doesn't work correctly. So I'm going to run smolt to
47:13
submit that and hopefully I'm going to fix my system. The
47:17
problem with that is that you go to database that made it look
47:19
like the leanest community was only using weird systems
47:21
basically. Right. Yeah. Yeah. I would
47:23
skew the results I would think, or you got guys in
47:25
there like my system so awesome. I want to submit
47:27
it with my six monitors. Yeah,
47:30
exactly. And then, and
47:32
then if you're the sign new system, assuming everyone alter has six
47:34
monitors, you're probably missing part of the market. I
47:37
mean, I imagine there's two, there's going to be a lot of software you'll have
47:39
to create on the backend to visualize all this and view
47:41
it and understand it. Right. That's going to be a whole project
47:43
in itself. So yeah, I mean, of course, as
47:45
long as we have it, I mean, you know, one of the restrictions
47:48
we're putting is that it's all going to be
47:51
independent data. We're not going to create like a user
47:53
profile
47:53
with like linking things
47:55
together. So it makes, of
47:57
course parsing it a little simpler in some ways,
47:59
because it's like. okay, so 40% of
48:01
Fedora users like to use Inkscape.
48:04
It's like I cannot necessarily that easily correlate it
48:06
with it, like, you
48:08
know, and they use GIMP too, kind
48:10
of, right? Because they're just two separate data points.
48:12
I mean, maybe if the percentage is very close, I would assume
48:14
that, you know, maybe these are some of the same
48:16
people who are artistically inclined.
48:20
But it
48:21
means,
48:23
you know, the sort of analysis we can do, at least
48:25
as long as we keep that model of having
48:28
everything completely independent is a little
48:30
bit more limited, but, you know, still a
48:32
lot of useful data to be had. That's an example
48:34
of the privacy trade-off that the project's going to have to make,
48:37
yeah.
48:37
And I could, you know, of course there's
48:39
commercial vendors that are going to go way farther and go
48:41
way beyond, but I could see wanting to walk that
48:43
line very carefully and respectfully
48:45
as a free software project, because you'll probably
48:47
be setting an example for other projects.
48:50
Yeah, and hopefully, and I mean, you know, we obviously have
48:52
learned from what and this they are doing, and I think they did
48:54
a stellar job with their tooling, and we are trying
48:57
to take it and sort of adopt it for our use case. And
48:59
I think that's the other thing is like,
49:01
I feel this is trying to do, you know,
49:03
metrics gathering the right way, as opposed
49:05
to, like, people probably think about, like, as
49:07
I said, like, you know, weird hidden things in your system
49:10
that gathers hidden data in a mysterious
49:12
way. Right. It's in there. You
49:14
know it's doing something, but you just have no idea what. That's
49:16
the Windows experience, and we don't want that. No,
49:19
exactly. All right,
49:20
Christian, thank you for coming on and answering my questions and
49:22
explaining it to us. Yeah, well, thank you for having
49:24
me. Let's start to make it less than a year
49:26
next time. Sounds good. Thank you.
49:32
collide.com slash unplugged.
49:34
If you work in security or IT, and
49:36
your company is Okta, you need to listen to this.
49:39
If you've noticed that over the past few years, the majority
49:41
of data breaches, you
49:43
know, hacks that maybe you read about in
49:45
the news, they seem to have something in common, don't they?
49:48
The attack vectors often the employees.
49:50
I mean, sometimes their device gets hacked
49:53
because of unpatched software. I'm looking at you old
49:55
Android devices. Sometimes,
49:57
unfortunately, employees leave sensitive data in
49:59
an unsecured place.
49:59
Sometimes they just don't know better. And
50:02
it seems like every day a hacker, you know, breaks
50:04
in using credentials they got from some phished
50:07
account or something that leaked online.
50:09
Just, it's a huge problem to deal with at scale.
50:11
And the problem here isn't really end
50:13
users, is it? I mean, they're
50:15
not, they're not doing it intentionally. And
50:18
it's
50:19
the solution so far that enterprises have been provided.
50:21
They just haven't really prevented these breaches.
50:23
They just don't do enough. It doesn't have
50:25
to be that way though. Imagine a world where
50:27
only secure devices could access your apps. In
50:30
this world, phished credentials are useless to hackers
50:33
because you can manage every OS, even the Linux
50:35
machines from a single dashboard. And best
50:37
of all, you
50:38
get employees to fix their own device security
50:40
issues without creating a whole bunch of work for IT.
50:43
And the good news is
50:45
you don't even have to imagine this. It's actually available
50:47
with Collide. That's what Collide does. It's a device
50:50
trust solution for companies with Okta. And
50:52
they ensure that even if a device isn't trusted
50:54
and secure, it won't log in to your
50:57
apps. That's powerful and it stops
50:59
problems before they happen. So go to collide.com
51:02
slash unplug
51:03
to watch a demo and just see how this works. It
51:05
isn't magic. They've really figured it out. So you go
51:07
to K O L I D E.com
51:09
slash unplugged. That's
51:11
collide.com slash unplugged.
51:17
As usual this week, we got some great feedback.
51:19
Thank you very much for that. If you want to leave us
51:21
some, maybe about what Christian and Chris
51:23
just shared, we would love to hear it.
51:26
Linux unplugged.com slash contact
51:28
for that. And, uh, I'll even read it personally
51:31
for you. Now, Matias
51:33
sent us a little note and I thought, Chris, you
51:35
might add this to your, maybe your wishlist. Matias
51:38
says,
51:39
I have a book suggestion about IT
51:42
security.
51:43
I do have to warn you though, after reading this, you
51:45
will feel a strong need to move
51:47
out into the woods in
51:49
a Faraday cage. The book recommendation
51:51
is this
51:52
is how they tell me the world
51:55
ends. The cyber weapons arms
51:57
race by Nicole Pearl.
51:59
Oh man. And we have a link to a
52:02
Goodreads insert on that, so
52:04
have a look. And Chris, are you gonna get this one? Boy,
52:07
I already have run out into the woods tendencies.
52:10
I don't know, is this... Next to a Faraday cage, I
52:12
think. Hm,
52:14
can I build a Faraday cage into the RV
52:16
bedroom? Is Jube's a Faraday cage?
52:19
Maybe sometimes. She's got plenty
52:21
of radio signals inside.
52:23
Interesting. Okay, alright. This
52:26
is how they tell me the world ends, the cyber
52:28
weapons arm race. You know,
52:31
I always find getting information
52:33
on this fascinating, even if, you know, maybe it
52:35
only affects a small percentage of those
52:38
on the internet doing, who knows, maybe
52:41
nefarious dark secret things. But
52:44
it's always kind of interesting because even if it increases
52:47
your own, you know, mindset
52:49
or protection or the steps you take for yourself
52:51
by, I don't know, five, 10%, well, that's already
52:54
a win for you personally. So I say go
52:56
for it. I mean, if it's an audio book,
52:58
which I don't know if it is, but if someone finds it, please let
53:00
us know.
53:01
Why not? Yeah, you got me.
53:03
You sold me. All right. Thank
53:05
you for the... I really appreciate the book
53:07
recommendation.
53:11
And now it is time for the
53:13
boost. A delicious name, Noodles with
53:16
the Z rights in, and they are our baller
53:18
this week with 93,650 sats. Hey Rich
53:20
Luster! Sent it in via
53:22
the podcast index. They write, hey guys, just sending a boost
53:26
after a long kerfuffle with strike and getting sats because I had
53:28
my identity
53:30
stolen in the past. Oh, so KYC that stands for know your
53:32
customer, which is sort of law here in the West, makes
53:35
it near impossible to get sats for guys like me. Anyways,
53:37
I've been listening since around episode 250. So
53:40
a long time listener, but a first time booster. I just want to
53:42
say love the show and I hope it goes on for infinity.
53:46
Or near it. Well, as best we can. Right? I
53:48
mean, look at, look at these two
53:49
guys. They look like they got years left on them still.
53:55
can
54:00
save these sats for robot bodies? Yeah,
54:02
definitely. You know, the robots are going to
54:04
be taking sats for sure.
54:06
I actually would love to know more information
54:09
about the whole like identity
54:11
stolen, how you got the sats after that and the troubles
54:13
you ran into because I got to imagine that could happen to anybody.
54:15
Thanks for jumping through all the even more hoops.
54:18
You know, it's also a zip code boost
54:20
there, Wes Payne. Yeah, looks
54:22
like it's coming in from Fresno,
54:24
California. Hey, not
54:27
too far away.
54:28
They're also noodles.
54:30
One, two, three, two on Matrixorg,
54:32
if anybody wants to chat from Fresno.
54:35
Very nice. Thank you. Rotted mood boosts
54:37
in with 50,003 cents. Oh.
54:41
Hey, Rich Luster. Hey, Rich
54:43
Luster.
54:45
I was wondering where you all get the most value
54:48
for the show, the membership or
54:51
the boosts. If one were to have to do
54:53
one or the other, which provides the most
54:55
for the network? Also, here's another
54:57
boost just to say I'll miss Linux action news
55:00
during its break. It's what brought me in
55:02
to the network. Oh, that's always interesting. I always
55:04
like to hear that.
55:05
Thank you. Yeah. So what
55:07
I don't know if I would say one, particularly
55:09
over the other, I think the memberships I
55:11
look at as a nice, predictable
55:13
kind of this is what we know we can afford to
55:16
produce on
55:16
an ongoing sort of. Yeah. And
55:19
so, you know, that's
55:21
pretty important, a pretty important metric. I
55:24
look at the boost as a way to
55:26
try to finance that individual episode
55:28
as sort of
55:29
like so
55:30
if the membership covers
55:32
the baseline production,
55:34
the boost is sort of covering our
55:37
time and a little bit extra because some of it goes
55:39
to Drew, some of it goes to developers.
55:41
So it's kind of like a thank you for that particular production
55:44
and the value you got from that particular episode.
55:46
So I kind of see them as sort of two separate things.
55:49
But my answer is, is whatever works best for you,
55:51
right? Like if you don't want to mess with the Sats,
55:54
totally understand where the membership would go. If you're ready to
55:56
experiment with something that's pretty fun and technically interesting.
55:59
than the boost there. And they seem to be really,
56:02
really popular for people that are on new podcast apps
56:04
too, because if you think about it, you're just listening along
56:06
and the feedback button's just right there. You just hit that button
56:09
and boop, boop, boop, boop, boop, boom, you send it in.
56:11
It's a nice way to say thank you if there was an episode that you thought
56:14
was really useful.
56:15
We also touched on this topic quite a bit,
56:17
Chris, in episode 32 of Office
56:20
Hours. You can check that out in officehours.hair where
56:22
we're trying a bit of a new model
56:24
if you haven't heard of it yet. Chris, can you give us like
56:26
a one sentence
56:29
what we're doing over there? What are we doing over there? A
56:31
bounty for production?
56:32
You know, it's a community show and we're setting
56:34
a modest bounty.
56:36
And when we reach that or get near it, we start producing
56:38
the episodes
56:39
with short, quick, like little updates on what's going
56:41
on in the feed in between from time to time.
56:43
So it's a new model we're trying. So the idea is
56:46
that we have, we're
56:48
not kind of like putting the show production on credit and
56:50
then hoping we make it back maybe, which
56:53
we probably wouldn't. And so what we're gonna
56:55
try instead is we'll do a bounty. And when that bounty
56:58
is getting close or reached, then we will
57:00
begin production on the episode. And we know we can safely produce
57:02
the episode and not put the network at risk. Office
57:04
Hours Pre-Pay Edition. Yeah, you could say it's
57:07
a new experiment we're trying. You know, we'll see, we'll
57:09
see. What
57:10
about LAN? Well, it's the reason why we didn't
57:12
do it with LAN is because our
57:15
goal with LAN is to really keep it lean and tight. You
57:17
know, I mean, if we can get really solid
57:19
information in under 15 minutes, that's a win for us.
57:22
And so if we, you know, the only way that I think the
57:24
value for value model really works is it's not,
57:27
it's not a payment technology.
57:29
It is a format. And so
57:31
you have to do the shout outs.
57:33
You know, the community comes
57:35
up with like the row of ducks and the rich lobster
57:38
and you
57:38
know, like that's all community driven and
57:41
it's part of the value for value format. And
57:43
I don't think LAN, not all shows, at
57:45
least I think right now,
57:46
are particularly well formatted
57:49
for that. If you want something that's 10 to 15
57:51
and 25 minutes is a long episode, I
57:53
just don't know if it works there.
57:55
Maybe, I mean, maybe you could do like a short version
57:57
of like the top two or top three boosters or something. But
58:00
that's maybe why we start with try things like
58:02
that in office hours. And then, you know, if it's wildly
58:04
successful, we can say so office hours is
58:06
that's just a good that's a great point. That is our test ground
58:08
show where we just whatever like the wildest thing we're trying
58:11
on the network,
58:12
we're trying some new technology, we're trying some
58:14
back end thing,
58:15
we're trying out a new format. We do it in office hours.
58:17
It's sort of a skunk works.
58:19
So the reason I ask about this is because
58:21
I remember when Linux action show split
58:23
into LAN and LOP, right? And the idea
58:26
back then was the two
58:28
components of Linux action show would be decomposed
58:30
into
58:31
tighter, more focused shows.
58:34
But,
58:34
you know, LOP is designed to respond to LAN.
58:37
And if you don't have LAN like so.
58:40
So where does the where does
58:42
the inflow come in for being
58:45
able to have the commentary of the news from from
58:47
the JB perspectives? How do I
58:50
guess I don't know. Does it mean that you want to pull pieces
58:53
of land back into into LOP and
58:55
make the show again? It's very
58:57
rare that we do the same topics in both shows.
59:00
It's probably one out of every four episodes or
59:02
so of LAN.
59:03
OK, we're basically doing what we just did
59:05
today is
59:06
we're pulling in
59:07
the topics that are worth discussing.
59:09
The reality is, and, you know, outside
59:11
of this Red Hat news, there's not a lot there's
59:14
not a lot of news during the summer in Linux. Yeah,
59:16
it's usually a dead time. Yes. And Thanksgiving
59:18
are pretty dead. Right.
59:20
So it's sort of like the time to do this for
59:22
LAN. And we can take that extra load
59:25
on now.
59:25
When summer ends, new cycle picks back up, developers
59:28
get back to work.
59:30
You know, we'll see. And maybe the advertising market will be changing
59:32
by then, too.
59:33
That's a possibility.
59:35
Cyber Gray boosted in two, three,
59:37
four, five, six satoshis.
59:41
Started listening during 2020 and have been
59:43
an avid listener since all of Jupiter's
59:45
shows are awesome, fun, and the entire
59:48
JB team is a delight to hear from.
59:50
Oh, geez, I don't know if
59:52
I could go on, guys. I'm blushing so much over here. Yeah.
59:54
I would love to attend a meetup. Any chance
59:57
there can be more over by, say,
59:59
Alex?
59:59
or in the general East Coast
1:00:02
of the US generally. The
1:00:04
boost is a nearby zip code, by the way.
1:00:07
Virginia Beach, Virginia, or
1:00:09
there were bounds. Wouldn't that be great? Yeah,
1:00:13
travel is tricky because A, it's
1:00:15
expensive,
1:00:16
and B, I got three young kids,
1:00:18
and C, I got a tiny farm. So it's tricky right
1:00:20
now,
1:00:21
but it is absolutely one of the highest goals
1:00:23
I have, is I'd really love to be able to travel more and get out
1:00:25
and do
1:00:26
more meetups and more live shows, because when
1:00:29
you do something for a really long time,
1:00:32
going out and getting that kind of recharge, it's
1:00:34
like, yeah, anytime you get a chance to do that, it's
1:00:37
so great.
1:00:38
Every time we have a meetup, we're like, let's go! And
1:00:40
it really gives us a nice boost because there's such
1:00:42
great people, great conversation, and
1:00:45
all of that. So that's a long way
1:00:47
of saying, inevitably, yes, I don't
1:00:49
know when, we really had a great
1:00:51
crowd in the Raleigh area.
1:00:54
So that I could see doing again soon.
1:00:56
If we could find really cheap flights,
1:00:57
or if we have a listener out there, they can get us a cheap flight.
1:01:00
Or if
1:01:01
I could get a source of a lot of gas,
1:01:03
and I could just drive there, I'd do that too. One
1:01:06
way or another. We take the train and we podcast
1:01:08
the whole line. Oh yeah. Oh yes. Can
1:01:11
you take a train all the way there? Can we do
1:01:13
that? Let's find out. That would be great.
1:01:16
V, it probably almost be as long as taking the RV.
1:01:18
Lebann comes in with 20,000 Seth using
1:01:21
Castamatic, which I
1:01:23
hear is great on iOS. Interesting to see
1:01:25
Oracle bashing IBM and Red Hat for the recent
1:01:27
rail decisions.
1:01:29
Yeah. He says, it's one Titan
1:01:31
fighting another. However, should Oracle
1:01:34
be the one to throw the first stone? I know.
1:01:36
I thought Oracle was just not gonna say anything. I
1:01:38
think that's why I was taking so aback by such a tacky
1:01:41
post. They would just keep quiet, do what they're gonna do.
1:01:43
They're like a huge tech Titan.
1:01:46
Lawyer driven.
1:01:47
They don't need to say anything. And
1:01:50
then to have a post that sounds like it's written
1:01:52
by a tech bro who's cock of the walk of his
1:01:54
department and nobody can tell him to say anything different.
1:01:57
It's like, wow, I can't believe this is coming out.
1:01:59
This is coming out of a hundred person company. It's unbelievable.
1:02:03
Uh, anyways, that's my take.
1:02:05
Uh, sorry, I didn't mean to go on. The
1:02:07
band goes on to say, but alas, Microsoft course
1:02:09
corrected years ago
1:02:10
and arose from everyone's S list, pushing
1:02:13
Google down in the process.
1:02:14
Perhaps it's time Oracle do the same.
1:02:17
The easy way to do every, oh, I know. He
1:02:19
says the easy way to everyone's heart is for them to
1:02:21
GPL ZFS. I
1:02:23
agree. I will have a thank you appreciative
1:02:25
Oracle episode
1:02:27
if they just GPL opens ZFS
1:02:29
or whatever Oracle hearts open
1:02:31
source. The interesting thing about Oracle's
1:02:33
relationship with open sourcing things people want.
1:02:39
I will make an observation and you can make
1:02:41
for what it's worth.
1:02:44
Um, Oracle relicenced D trace,
1:02:46
which had been wanted for a decade in Linux
1:02:49
from right. Yeah. Harris
1:02:51
license to the GPL after nobody
1:02:53
wanted it anymore.
1:02:54
And after at that point, you know,
1:02:57
a BPF trace that came on board and
1:03:00
system taps a D trace portability
1:03:02
framework was reused for a EBPF
1:03:05
and things like that. Yeah. At
1:03:07
that point, Oracle relicenced D trace
1:03:10
into GPL and incorporated it into Oracle
1:03:12
Linux. So I
1:03:15
don't know what that says about how Oracle
1:03:17
does stuff, but it is an observation
1:03:20
that I've had about how
1:03:22
they seem to approach taking Solaris technologies
1:03:25
and making them available
1:03:27
for Linux. All right, quick, everyone pretend
1:03:29
CFS is trash. Yeah. Pretend
1:03:31
you don't care about it and then Oracle will open source it,
1:03:33
you know, because they don't want to fragment Linux and they don't
1:03:35
want to cause any heartburn for Linux users because they
1:03:37
care so much about free software. Hey, maybe they
1:03:39
could be a Linux unplugged sponsor. Yeah.
1:03:42
Wouldn't it
1:03:43
Oracle go get Oracle Linux. It's just like rel,
1:03:45
but now not quite as much. J-moon
1:03:51
boosts in with 7,777 sets. Boost!
1:03:56
With the simple message. Boost!
1:03:59
Thanks, J-moon.
1:04:00
Love to know you're out there.
1:04:03
Pascal comes in with 2000 sats from
1:04:05
the Konshacks web app. Hey,
1:04:07
oh, that's a great app. Hey,
1:04:11
Linux team. Thanks for using the podcaster
1:04:14
support page for your podcast. Time to tell
1:04:16
your audience how easily they can post
1:04:18
a message and send you some support.
1:04:21
Also, if you have any suggestions, please
1:04:23
let us know how we can improve Konshacks.
1:04:25
All the best from Pascal. How great is that? What? They're
1:04:28
kind of like passing on. It's like passing on the message
1:04:30
about the service via boost and that's
1:04:32
great. So Konshacks is a great dashboard.
1:04:35
If you launch a value for value podcast that uses
1:04:37
boosts,
1:04:38
you can. Pipe it into Konshacks
1:04:40
and then it also generate a support page for your
1:04:42
audience so they can go and see like if it's
1:04:44
something we could potentially use with office hours. So I'm going to, I'm going
1:04:47
to look more into it.
1:04:48
It could be something there. Radman comes
1:04:50
in with 5,000 sats is looking forward to this every
1:04:53
single week. Thanks guys.
1:04:55
Olliewyn comes in with 4,000 sats
1:04:57
just to say great show. Also from
1:05:00
Konshacks web. Look at that. Hey,
1:05:02
okay. Might be part of the Konshacks crew
1:05:04
there.
1:05:05
User 56823579 comes in from Fountain with 2021 satoshis.
1:05:14
First time booster from NL. Now I'm
1:05:17
just going to pretend that's Newfoundland. A
1:05:19
postcode boost with Fountain earns sats.
1:05:21
See postcode. I'm telling you, it's Newfoundland. Listening
1:05:24
to most of the JB shows.
1:05:27
Got a postcode boost. Ah,
1:05:29
sorry, Brent. It's from the Netherlands in
1:05:31
Harlein. Harlein? Hey,
1:05:33
that's great.
1:05:35
Thank you for boosted in and thank you for the
1:05:38
zip code. Are you thinking West when you
1:05:40
see these, you thinking maybe I should just build into the script.
1:05:42
When somebody says something about a postcode,
1:05:44
it just like looks at the satom out and go look that postcode
1:05:47
up. I hadn't, but now I am.
1:05:50
Coming in hot with the boost. Jean Bean
1:05:52
comes in hot with 8649 sats. You've
1:05:55
all been talking a lot about XMPP. Is
1:05:58
there an end to end encryption available? for
1:06:00
it. I think many years ago end to end was
1:06:02
not there. That's what led me to other places.
1:06:05
XMPP has a hot, it's hot again. It's
1:06:07
people are talking about it again. They've looked at the alternatives
1:06:09
and they go back and they say to themselves,
1:06:12
I wasn't so bad. But
1:06:13
I do think the end and encryption must
1:06:15
have been, I don't
1:06:17
know. I mean, that must have been client side because I could
1:06:19
have sworn it was part of off the record conversations.
1:06:21
Oh, that's right. I knew there was something.
1:06:24
Yeah, back then it was called OTR, off
1:06:26
the record.
1:06:27
And it encrypted the messages
1:06:30
and then threw away the decryption key
1:06:32
after a period of time. So you couldn't go back
1:06:35
and see the messages. That's what led them to be
1:06:37
quote unquote off the record. That's cool.
1:06:39
I think we should, I think we should set up
1:06:41
just a really, really early Snapchat.
1:06:44
Oh, matrix. Yeah, of course. Yeah, you could do it with
1:06:46
matrix as well. The difference between matrix and XMPP
1:06:48
on this is that the decryption key is preserved
1:06:51
by default. You could destroy the decryption
1:06:53
key whenever you want and make all of your messages
1:06:55
unreadable. Yeah,
1:06:58
all our all our Star Trek chats destroyed
1:07:00
like that. So sad. Bear 454 boosts in
1:07:03
with 10,000 sats. Oh,
1:07:06
boost! From the podcast index.
1:07:09
While I get your takes on the Red Hat
1:07:11
source decision, I don't agree.
1:07:14
For me, Red Hat has broken the promise
1:07:16
of providing support for a free product.
1:07:19
The product is now only freely
1:07:21
available after you've paid for support. And
1:07:24
while it may be technically legal under the G.P.O.,
1:07:27
it definitely is not free Libre
1:07:30
open source software. This is
1:07:32
something that I've been that has kind
1:07:34
of been a peave of mine for a long
1:07:36
time. OK, what they think that it's
1:07:38
not G.P.O. anymore.
1:07:40
No, no, no, the support thing like what they're defining
1:07:42
as support. Right.
1:07:44
When when you look at open source stuff
1:07:46
and you look at what is actually involved
1:07:49
in making I mean, even in proprietary software,
1:07:51
like what support entails is
1:07:54
not just hey, you file a
1:07:56
bug and we and someone responds
1:07:58
back and does.
1:07:59
something with it. Or
1:08:01
it also includes basic
1:08:04
things like being able to do
1:08:07
the development to make the software work,
1:08:10
add features and security
1:08:12
fixes and stuff like that. It's engineering
1:08:15
effort. It's customer experience
1:08:17
effort. It's documentation effort.
1:08:21
It's life cycle. It's certification
1:08:23
effort. All these things require
1:08:26
people to spend time
1:08:28
and energy on. A lot
1:08:30
of this is difficult to do on
1:08:32
a volunteer basis. Some of it is straight up impossible
1:08:35
on a volunteer basis. Are you going to
1:08:37
ask somebody who can only spend 30 minutes
1:08:39
a week on a project to do ISO
1:08:42
certification or common
1:08:44
criteria or FIPS? No.
1:08:48
That's unreasonable. Those
1:08:50
kinds of things require
1:08:53
people to be paid to
1:08:55
follow through
1:08:58
with everything to deal with all the gnarly
1:09:00
bits of these things. Also
1:09:03
to keep the software working for a longer
1:09:05
period of time than anyone really should. These
1:09:08
kinds of things take effort
1:09:11
and time and money.
1:09:14
Nobody can live off of, you know,
1:09:16
you can't live off of, in
1:09:19
the art world and in the actors world,
1:09:21
you can't live off of exposure. In
1:09:23
the open source world, you can't live off of GitHub
1:09:26
stars. You can't live off
1:09:28
of appreciation. Appreciation
1:09:31
is great. I barely get any of it for the open source
1:09:33
work I do. But
1:09:35
I can't live off of it. If I got
1:09:37
all the thanks in the world, it doesn't change the fact that if
1:09:39
I have no money, I can't live in my house and
1:09:41
I can't get food and I basically would
1:09:44
be a dead man. Like, no, you
1:09:46
got to eat. You
1:09:48
need to survive somehow. And by the
1:09:50
way, Rel is for businesses. It's for enterprises
1:09:52
and businesses that make money.
1:09:54
That's its customer base. That's
1:09:56
truly its customer base. And I think
1:09:58
where I can kind of
1:10:01
jive with bear here is like, I
1:10:03
would, I do wish all of these
1:10:06
CentOS changes and RHEL changes over the years
1:10:09
were done on an enterprise timescale
1:10:11
where it's like at the end of the supported release, we're
1:10:13
making this change.
1:10:15
I, you know, I've been, I've been informed
1:10:17
that, you know, it's like,
1:10:18
this has sort of had to be done because the
1:10:20
infrastructure that we use to produce those SRPMs
1:10:23
needed to be addressed. And so
1:10:26
there was just several factors at play, but
1:10:28
I do wish the time, I think with where I kind
1:10:30
of agree with bear there is like, I do wish the timeline
1:10:32
could have been better
1:10:33
and more user-friendly. It's complicated to, you know,
1:10:35
weigh the interactions of all
1:10:38
entities, corporations, people with their own agendas
1:10:40
and goals in the open source, you know, mixing
1:10:43
bowl. And this is just one factor I think to
1:10:45
keep weighing as we watch what happens and in the future
1:10:47
choices and where you want to be in the ecosystem. You
1:10:49
know, I want you to think about this,
1:10:51
the viewers, the listeners, everyone.
1:10:54
How bad is it to pay $400 a year for
1:10:58
all the software, all the open source software
1:11:03
that, that underpins your applications,
1:11:06
your workloads
1:11:08
to run essentially forever,
1:11:10
right? How bad is it to pay $400
1:11:13
a year to pay for the engineers
1:11:16
that spend all that time to work on it? Well,
1:11:18
especially if the software you're running makes you a million
1:11:21
dollars a year. Right. Like, and
1:11:23
you know, in the desktop space, you know, we've talked about
1:11:25
how there's this general assumption that,
1:11:28
you know, there's no money in the desktop, but you could
1:11:30
buy
1:11:31
desktop Linux. You've always been able to buy desktop
1:11:33
Linux. Uh, Red Hat
1:11:35
offers Red Hat Enterprise Linux workstation for,
1:11:38
I think it's like $200 a year per system. Uh,
1:11:43
like you can go to redhat.com slash store and
1:11:45
you can just go see it for yourself as an individual. You
1:11:47
can buy workstation. You can buy
1:11:50
server and use
1:11:52
it on your own machines. Red Hat is
1:11:54
gracious enough that for individuals, you
1:11:57
can get 16 instances of RHEL with all
1:11:59
the content available.
1:11:59
available for free through
1:12:02
the Red Hat Developer subscription for individuals
1:12:04
that you can get at developer.redhat.com. But
1:12:06
if you want, you know, be able to file support
1:12:08
cases and stuff like that,
1:12:10
you could absolutely buy it as an individual. And
1:12:12
if it's something you depend on and you want
1:12:15
reliability, unmatched reliability,
1:12:17
pay for it. I mean, it's not even just with
1:12:19
RHEL, right? Slezz has had, SUSE
1:12:21
Linux Enterprise has had paid off options for
1:12:25
decades. Like it's been there for both
1:12:27
distributions. You go to SUSE.com
1:12:30
slash shop, I think, and they have them there.
1:12:32
I used to buy the box version of SUSE all the time.
1:12:34
Loved it. Just pay for something like paying
1:12:36
for a subscription. You know, we
1:12:39
talk about how doing code contributions and documentation
1:12:41
and stuff like that. My philosophy here
1:12:43
is that you contribute and support
1:12:46
open source with either
1:12:47
your time and effort or with money.
1:12:49
You could do a mix if you'd like, if you're
1:12:52
not skilled in everything, but like those are usually
1:12:54
your choices. And I tend
1:12:57
to choose more often than not that I spend
1:12:59
my time and effort helping open source projects,
1:13:01
but where I can't, I obviously give money instead.
1:13:04
Or sometimes I do a mix if I can't help a lot,
1:13:06
but I can try. Like if you
1:13:08
care about the success of the
1:13:10
Linux desktop or the Linux server and the Linux
1:13:12
ecosystem,
1:13:14
start by trying to pay for the
1:13:16
stuff that you consider
1:13:18
that gives you value, that enriches
1:13:22
your life, that makes things great
1:13:24
for you.
1:13:27
It's been a real problem for, we've
1:13:30
been talking about it for as long as, I don't know, all
1:13:32
the way back to freaking Linux action show way
1:13:34
back in the beginning of how there's this
1:13:36
problem about funding open source work.
1:13:39
Red Hat is the largest company that
1:13:41
I know of doing open source work, and
1:13:44
it's the largest pure play open source
1:13:46
company I'm aware of.
1:13:48
You don't want to support and enable
1:13:51
them to do more of that. That
1:13:54
money helps make all that stuff happen
1:13:56
and improve it and make it go better. If
1:14:00
you want to have open source be the
1:14:02
default, if
1:14:06
you want open source to win,
1:14:09
you can't take it for granted and you can't
1:14:11
just not be
1:14:13
willing to support it. Like
1:14:16
that's my piece on it. We agree. Listener
1:14:19
Brent boosts in to say, thank you, Neil, for
1:14:21
everything you do for open source. And
1:14:24
we've also got a boost from
1:14:26
listener SWAT, 2,317 Satoshis.
1:14:31
Long time lurker and first time booster
1:14:34
here. B-O-O-S-T. By
1:14:36
the way, the Satoshi amount is half
1:14:38
of my Dutch zip code.
1:14:40
Good luck Wes.
1:14:41
Please keep newbie content available in
1:14:43
the show, even if it's just a remark here or there.
1:14:46
I noticed that the younger sis admins or programmers
1:14:48
don't always know or remember the
1:14:50
old stories and don't have the knowledge we
1:14:53
older folks take for granted. Apparently
1:14:55
there are still some people that we
1:14:57
need to onboard.
1:14:59
All right, Wes, track them down. It's
1:15:01
somewhere in the Netherlands. Oregon,
1:15:04
all zoom, one strict.
1:15:07
Boost in and let us know maybe a little
1:15:09
more closely where you are. Yeah, and how good
1:15:11
he got that. Yeah, we are still taking feedback
1:15:14
on technical content versus newbie content.
1:15:17
We're always trying to get that balance right. We,
1:15:19
I think, personally like to skew a little more technical, but
1:15:23
also we don't want to be unavailable. And newbies too,
1:15:25
we want to be approachable by them.
1:15:27
Sort of a hard Venn diagram
1:15:30
to actually nail, but that's why your feedback
1:15:32
kind of helps us nail it down. Unknown
1:15:35
sender, a mysterious sender came in with 6,777 sats
1:15:39
with no message, but we just wanted to say thank you. And they're
1:15:41
using Castamatic, which
1:15:44
runs on iOS.
1:15:45
Zagatec comes in with a Rorodux.
1:15:49
A little behind on the boost, but wanted
1:15:51
to thank you for your perspective on the whole red hat
1:15:53
soap opera.
1:15:54
Your coverage of it in 517 really brought
1:15:56
a different perspective. I appreciate it.
1:15:58
Well, we appreciate that.
1:15:59
And it does register with you when you get a little value
1:16:02
from it, you thought to boost in. Thank you,
1:16:04
Zach Attack, for that row of ducks. Thank you, everybody.
1:16:06
It really does feel like a soap opera at this point. I
1:16:09
know. I know. I know. We
1:16:12
got a boost too from Dan Johnson in the Mumble Room who
1:16:14
boosted in their
1:16:16
noster earnings in the last week.
1:16:19
Man, Dan, I
1:16:20
got to get my noster
1:16:22
game up. I got to figure out how to link my noster
1:16:24
identity. And I've been
1:16:26
enjoying the discussions over there, but I'm just a lurker.
1:16:29
I'm just a total lurker. Well, some
1:16:31
of those sets are yours. Oh, OK. Yeah,
1:16:33
I did. I did give Dan a zap. So
1:16:36
you boosted into your own show. Bonus boost
1:16:38
that's relevant from a Debris 17,000 sets.
1:16:42
Should JB continue to be technical? This is response
1:16:44
to 503. Yes. That's one of the main
1:16:46
reasons I love these shows. Can I go deep and break down
1:16:48
what's being talked about in a way that's understandable? I'll
1:16:51
explain it along the way. Well, we appreciate that.
1:16:53
And I'd still like to get some signal on that and
1:16:55
see if we can continue
1:16:57
to get there. I think everybody who did boost in,
1:16:59
not all of them do make it on the air, but we try to get
1:17:01
most of them. We had 18 total boosters
1:17:04
this week, which
1:17:04
is awesome across 20 total boosts. A
1:17:07
couple of people boost in a couple of times.
1:17:08
And we earned a grand total of 241,232 sets. It's
1:17:12
a bit low this week, but
1:17:15
still appreciate every one of them.
1:17:17
And we'll put them to work.
1:17:19
Thank you very much, everybody. And I had a question for the boosters
1:17:21
next week. Oh, you got a question. I'll ask you live
1:17:23
if you noticed any particular habits that your
1:17:26
dogs have developed due to like your podcast
1:17:28
listening or just listening to content on your phone.
1:17:30
Here's an example. So Levi
1:17:32
likes to stay in whatever room I'm going to be
1:17:35
in. And if I leave
1:17:37
a device playing a podcast in a room
1:17:39
and I step out, he stays in that room because he
1:17:41
knows I'm coming back. But if I take the phone with
1:17:43
me, that's playing audio and I leave the room, then
1:17:45
he follows me. That is clever.
1:17:48
I'm going to have to look out for that. No, I've noticed mine,
1:17:50
you know, when I put my wireless earbuds in, they're
1:17:52
like, oh, going for a walk. Oh, yeah,
1:17:55
yeah, yeah, yeah. That kind of stuff.
1:17:57
If your dog does that, boost in and let us know because
1:17:59
I love dogs. Uh, you can boost in by getting
1:18:02
Albie. Just go to get Albie.com.
1:18:04
Little web extension that brings the lightning network to your
1:18:06
web browser. You top that off either directly, or
1:18:08
you can use something like strike or the cash app. And
1:18:11
then you just go find Linux unplugged on the podcast
1:18:13
index. We have that linked podcastindex.org
1:18:15
and then
1:18:16
you can boost right there from their website or go
1:18:18
get a new podcast app at podcast
1:18:21
apps,
1:18:21
podcasting, to no revolution podcastapps.com.
1:18:25
Fountain, Castomatic and Podverse.
1:18:27
I think are the favorites among our audience. And
1:18:29
they're constantly getting new updates.
1:18:31
Podverse just got a great update for Android, really
1:18:34
solid.
1:18:35
Um, and Fountain has been a rock on my
1:18:37
iPhone for a while now. They really nailed it over
1:18:39
there. And I hear good things about Castamatic.
1:18:41
So thank you everybody who boosts in. And of course we'll be
1:18:43
thinking our members later as well, but
1:18:46
we are thinking about you too. We've
1:18:49
got not one, but two
1:18:52
fantastic picks this week.
1:18:54
I just wanted to give a shout out to Thunderbird115,
1:18:56
the
1:18:57
supernova release.
1:19:00
It looks so good. I think I'm going to give it a go. Uh,
1:19:02
they have redone the UI quite a bit. I
1:19:04
really like it. They've
1:19:06
made it super fast. They say a bunch of new
1:19:08
stuff in here. Got
1:19:09
a new logo,
1:19:11
improved calendar design, new
1:19:13
menu system, new unified
1:19:16
folder system, I mean, it's just checking
1:19:18
boxes for me, Westpain. Pretty sure the,
1:19:21
uh, updated versions already in Nix packages
1:19:23
too. So it should be easy to try. Love
1:19:26
it. I know they've been working on that. So
1:19:28
congratulations to them.
1:19:30
And then our second
1:19:32
pick this week,
1:19:33
I was so blown away. We had to have two picks.
1:19:35
I can't believe it. Maybe it's, maybe it's
1:19:37
because we shamed them on air.
1:19:39
I doubt it. I doubt it.
1:19:42
A while ago, I don't know if it was in the members stream
1:19:44
or if it was
1:19:46
actually in the show, but I mentioned as little
1:19:48
disappointed in RustDesk because they
1:19:50
were really dragging their feet on Wayland support
1:19:52
and RustDesk is a fantastic
1:19:54
true team viewer alternative. I'd say
1:19:56
superior to team viewer in every way. Been a really quality
1:19:59
tool. I mean, Nivia's bunch for doing live production.
1:20:01
Oh man, back when all of our systems
1:20:03
were on X11, I would do
1:20:05
entire shows remotely produced, routed
1:20:08
through the studio, and
1:20:09
Rust Desk was helping me activate all that software
1:20:11
based stuff. It was so solid.
1:20:13
Then I made the transition to Wayland, and when I
1:20:16
installed Rust Desk, it came up with a little box
1:20:18
and it says, oh, Rust Desk doesn't work on Wayland.
1:20:21
Would you like us to fix it?
1:20:23
And if you say yes, what they do
1:20:26
is they go alter your system and change
1:20:28
it to run X11, and they were Wayland. That's
1:20:31
what their fix is. Right. I was so incensed by
1:20:33
that. I just stopped using Rust Desk.
1:20:36
Well, Rust Desk 1.2
1:20:38
is out. It's completely rewritten in Flutter.
1:20:41
What? Maybe that's going to be a good thing.
1:20:43
Beta support for IPv6. I
1:20:45
thought I'd never see it.
1:20:46
Hardware encoding for
1:20:49
H.264, H.265 in beta.
1:20:51
AV1 codec support. And
1:20:53
the big one, it is a beta, but a big one.
1:20:57
Wayland support. No way.
1:20:59
Headless Linux support.
1:21:00
Resolution adjustment.
1:21:02
I did see a comment here. It looks like this is pretty new.
1:21:05
Surely there's bugs. Ha. Yeah.
1:21:07
Just noticed KDE Neon doesn't seem
1:21:09
to work. So far, we only are
1:21:11
testing it. This is a quote on their GitHub.
1:21:14
Only tested on mainstream stable Ubuntu 2.204, Arch,
1:21:17
and Fedora.
1:21:18
All right. So it's a mixed bag on Nix
1:21:20
probably. We'll see. Sounds
1:21:23
like it's not going to work so well on my Neon system. But
1:21:26
I can wait. This is progress. Just
1:21:29
use Fedora. Yeah, there you go. That's all
1:21:31
of it. This is fantastic progress.
1:21:34
And I want to express a lot of gratitude for
1:21:35
the Rust Desk team for taking this
1:21:38
work on. Rewriting into a new toolkit.
1:21:40
That's no major job.
1:21:42
Hardware encoding.
1:21:45
Huge thing.
1:21:46
And then Wayland support.
1:21:47
Yeah. Maybe a little word, you know, like not knowing what
1:21:49
the plan for Rust Desk was. Was it ever going to continue
1:21:51
to be a good player or something I could recommend? Even
1:21:53
if we're not using it, something to recommend for other people because
1:21:55
it was so easy to get. I mean, I've used it
1:21:58
to support, you know, my mom's laptop in the past.
1:21:59
Yeah, so be nice to have it around We'll
1:22:02
see how it does. I'll be curious to see exactly how they're
1:22:04
implementing it, too Oh, man, oh,
1:22:07
we're just gonna put it back on the kids machines. That's
1:22:09
gonna be so nice The RDP
1:22:11
stuff's been a little rough.
1:22:13
Oh, I'm excited about that.
1:22:14
Well, if I if I if I do it I'll report back. I'll
1:22:17
tell you how it goes
1:22:18
I'd also like to put a request out to the audience for your
1:22:20
thoughts on these most recent developments in
1:22:23
the rel soap opera that we've
1:22:25
been witnessing unfold This
1:22:27
feels like we've entered the beginning
1:22:29
of a really new era and
1:22:31
Alma's the player that I'm personally
1:22:33
the most interested in but I also worry
1:22:35
and I'd like to know what the audience thinks That
1:22:37
this strategy may cost them those
1:22:39
users that are truly looking for that kick the can
1:22:42
bug compatible I don't want to pay rel
1:22:44
solution and Will Alma
1:22:46
kind of get punished for that strategy or will they be successful
1:22:49
because they're kind of doing it the legit red hat
1:22:51
way Which
1:22:52
fire an enterprise I'd be wanting to work
1:22:54
with a distribution that's doing it the most legit sustainable
1:22:57
way possible Which seems like what
1:22:59
almost doing so I'd like to put a request out there
1:23:01
for the audience's thoughts on that
1:23:03
Now we're gonna have to switch the studio to Alma I
1:23:05
know like you know Like two episodes
1:23:07
ago when we talked about all the rail stuff. I felt
1:23:09
like I had a really fine resolution on my thoughts
1:23:13
On all these most recent developments. It feels a lot
1:23:15
more vague
1:23:16
and a lot of promises and just I don't
1:23:18
know Yeah, it's a little awkward something
1:23:20
we get to watch for I suppose a while to come
1:23:23
enterprise timelines after all. Yeah, that's true It is
1:23:25
gonna and the whole foundation has to get spun up.
1:23:27
It's it's gonna take a while
1:23:29
Maybe by the time Linux fest Northwest comes
1:23:31
along. We'll see some serious progress. Hey, I
1:23:33
hope we see everybody at Linux fest Northwest
1:23:36
Details at Linux fest Northwest org. It's gonna
1:23:38
be in October. We're gonna have lady joops
1:23:40
there We're gonna have some cooking going crews gonna
1:23:42
be there It's gonna be a great time
1:23:44
If you want a little more show remember we do get together every
1:23:46
single Sunday at noon Pacific 3 p.m.
1:23:49
Eastern over at JBLive.tv See
1:23:51
you next week same bad time
1:23:53
same bad station
1:23:55
and a very special Thank you to our members
1:23:57
unplugged core comm where you can support the show
1:23:59
you get an ad for
1:23:59
free version or you get the bootleg version,
1:24:02
which we really try to add some value to that. So I feel like
1:24:04
you get like double the show, which could be
1:24:06
if you want more Linux show, maybe you're going to go on
1:24:08
a long ride. I don't know.
1:24:10
Get the member feed, get the bootleg feed. Jordan Pennelyn,
1:24:13
Jordan's Dad RV approved. Yeah, that's for sure. Road trip approved
1:24:15
for sure. And of course
1:24:17
we appreciate that because that support
1:24:19
maintains production and you can support all the shows
1:24:21
if you really want to go all in at jupyter.party.
1:24:25
Links to what we talked about today are over at linuxunplugged.com
1:24:28
slash 519. That's
1:24:30
the website with all the links and deets, including
1:24:33
that blog post from endless OS
1:24:35
that goes into very
1:24:37
expanded detail about how that telemetry
1:24:39
collection is going to work. So if you really want to be informed
1:24:41
on the nuts and bolts, the
1:24:44
endless implementation is pretty solid and they documented
1:24:46
it over there and a good portion of that will be used by
1:24:48
Fedora and love your thoughts
1:24:51
on that as well. linuxunplugged.com
1:24:53
slash 519. You got it. And of course
1:24:55
the contact
1:24:55
page is over there as well as our subscribe
1:24:57
page and all that other good stuff. Thanks
1:25:00
so much for tuning this week's episode of the Unplugged program
1:25:02
and we'll see you right back here next Sunday.
Podchaser is the ultimate destination for podcast data, search, and discovery. Learn More