1
00:00:00,000 --> 00:00:04,094
Audrow Nash: Hey there, robot
enthusiast. Welcome to the Auto

2
00:00:04,094 --> 00:00:08,456
Nash podcast. Today we're trying
something a bit different. We've

3
00:00:08,456 --> 00:00:12,483
got a video demo packed episode
that's perfect for those who

4
00:00:12,483 --> 00:00:16,644
love robotics and are tired of
dealing with cumbersome package

5
00:00:16,644 --> 00:00:20,872
management. Joining me are Wolf
Vollprecht, the founder and CEO

6
00:00:20,872 --> 00:00:24,765
of Prefix.dev and Ruben Arts,
who's a software engineer at

7
00:00:24,765 --> 00:00:28,926
Prefix.dev. They're here to talk
about Pixi, a package manager

8
00:00:28,926 --> 00:00:33,154
that aims to make your robotics
development life easier. If you

9
00:00:33,154 --> 00:00:36,846
are a ROS developer or just
someone who's interested in

10
00:00:36,846 --> 00:00:40,671
learning more about how to
streamline your workflow, this

11
00:00:40,671 --> 00:00:44,564
episode is for you. We'll dive
into the motivations behind

12
00:00:44,564 --> 00:00:48,658
Pixi, how it works, and even get
hands-on with some demos. So

13
00:00:48,658 --> 00:00:50,000
let's get started. ​

14
00:00:50,000 --> 00:00:53,533
Audrow Nash: Hi Wolf. Hi Ruben.
let me start by having you guys

15
00:00:53,533 --> 00:00:56,000
introduce yourself Wolf. Would
you go first?

16
00:00:56,000 --> 00:00:59,961
Wolf Vollprecht: Definitely.
Thanks Audrow for having us. I'm

17
00:00:59,961 --> 00:01:04,052
Wolf, I'm the founder and CEO of
Prefix.dev and also one of the

18
00:01:04,052 --> 00:01:07,948
initiators of the Robo Stack
Project, which is partially why

19
00:01:07,948 --> 00:01:11,000
we're here today, to talk about
Pixi, but I'll.

20
00:01:11,000 --> 00:01:14,904
Audrow Nash: Hell yeah. And
Ruben, wanna tell me about

21
00:01:14,904 --> 00:01:17,000
yourself? Introduce yourself.

22
00:01:17,000 --> 00:01:21,462
Ruben Arts: you. I'm Ruben. I
work at Prefix. I previously

23
00:01:21,462 --> 00:01:26,231
worked as a software developer
for a robotics company where we

24
00:01:26,231 --> 00:01:30,385
experienced a lot of,
deployment, problems that we had

25
00:01:30,385 --> 00:01:34,846
to solve. And, going from that,
I got some experiences and

26
00:01:34,846 --> 00:01:38,000
joined prefix to, to make it
even better.

27
00:01:38,000 --> 00:01:42,500
Audrow Nash: Oh yeah. And we're
about to learn what Pixi is and

28
00:01:42,500 --> 00:01:44,000
what Prefix is doing.

29
00:01:44,000 --> 00:01:50,685
Audrow Nash: So Wolf, would you
tell me about the motivation?

30
00:01:50,685 --> 00:01:52,000
Behind Pixi.

31
00:01:52,000 --> 00:01:53,000
Wolf Vollprecht: For sure.

32
00:01:53,000 --> 00:01:56,598
Audrow Nash: just in general,
what problem are you solving?

33
00:01:56,598 --> 00:01:58,000
Why is this a big deal?

34
00:01:58,000 --> 00:02:01,883
Wolf Vollprecht: yeah, so as
developers, we're all needing

35
00:02:01,883 --> 00:02:05,497
software to build our
development environments, right?

36
00:02:05,497 --> 00:02:09,112
And it's been pretty hard to do
that consistently on a

37
00:02:09,112 --> 00:02:13,329
cross-platform manner. And Pixi
is package manager that aims to

38
00:02:13,329 --> 00:02:17,480
solve this issue. So makes it
really easy for your projects to

39
00:02:17,480 --> 00:02:21,228
set up the dependencies and
packages and things that you

40
00:02:21,228 --> 00:02:25,512
need, and then make it very easy
to run those too. So really the

41
00:02:25,512 --> 00:02:29,395
vision for Pixi is to be like
the missing companion to git

42
00:02:29,395 --> 00:02:32,943
where you git clone a
repository, and then you can do

43
00:02:32,943 --> 00:02:37,093
something like Pixi Run Start,
and it would set up everything,

44
00:02:37,093 --> 00:02:41,042
all the dependencies, and run
the project and maybe do some

45
00:02:41,042 --> 00:02:45,260
compilation, run some cmake, run
some whatever, Python and just

46
00:02:45,260 --> 00:02:47,000
work. That's the big goal.

47
00:02:47,000 --> 00:02:50,609
Audrow Nash: Hell yeah. And so
with that you get, you mentioned

48
00:02:50,609 --> 00:02:53,990
CMake and Python and things like
this. You're getting cross

49
00:02:53,990 --> 00:02:56,911
language support, which is
really cool. And you are

50
00:02:56,911 --> 00:03:00,635
focusing mostly on robotics. I'm
sure there's plenty of other use

51
00:03:00,635 --> 00:03:03,729
cases for this. but tell me
about what this is a thing

52
00:03:03,729 --> 00:03:07,396
that's needed in robotics. Maybe
Ruben. 'cause I know you have a

53
00:03:07,396 --> 00:03:09,000
lot of experience with this.

54
00:03:09,000 --> 00:03:12,762
Ruben Arts: Yeah, so one of the
biggest problems we had to

55
00:03:12,762 --> 00:03:16,329
tackle is that you are dealing
with both the ROS Index,

56
00:03:16,329 --> 00:03:20,285
software stack. You're dealing
with apt, you're dealing with,

57
00:03:20,285 --> 00:03:23,982
PIP for some things, and then
building that together with

58
00:03:23,982 --> 00:03:28,133
something like colcon or catkin
for those coming from the ROS 1.

59
00:03:28,133 --> 00:03:32,219
And doing that all together is
something that you have to learn

60
00:03:32,219 --> 00:03:36,111
and something that you have to
teach your colleagues if they

61
00:03:36,111 --> 00:03:39,743
come in with a not non ROS
background. So we really went

62
00:03:39,743 --> 00:03:43,764
into like how we're gonna make
this easy, how we're gonna make

63
00:03:43,764 --> 00:03:47,591
this fast, and how we're gonna
make this, into a deployment

64
00:03:47,591 --> 00:03:51,677
workflow. Because if you develop
robotics, you also need a wait

65
00:03:51,677 --> 00:03:55,633
to get it to the customer and
make sure you can update it. So

66
00:03:55,633 --> 00:03:59,460
from that angle, we tried to
figure that out and we found a

67
00:03:59,460 --> 00:04:03,157
solution in the Conda world,
which is what we're building

68
00:04:03,157 --> 00:04:04,000
Pixi for now.

69
00:04:04,000 --> 00:04:08,000
Audrow Nash: Oh yeah. Okay. Now.
So I feel like the motivation is

70
00:04:08,000 --> 00:04:11,508
clear. We wanna distribute
robotics software, and make it

71
00:04:11,508 --> 00:04:14,954
easy. So you wanna do it on
different operating systems.

72
00:04:14,954 --> 00:04:18,154
It's been very painful
historically. Let's make that

73
00:04:18,154 --> 00:04:20,985
better. it uses lots of
languages. That's been

74
00:04:20,985 --> 00:04:24,923
difficult. Typically, let's make
an approach that works with all

75
00:04:24,923 --> 00:04:28,246
this, all these different
languages. So let's get into

76
00:04:28,246 --> 00:04:32,000
kind of the meat of what this is
and how to think about this.

77
00:04:32,000 --> 00:04:35,405
Audrow Nash: And I think it
starts probably with package

78
00:04:35,405 --> 00:04:39,236
management. would you tell me a
bit about the parts and package

79
00:04:39,236 --> 00:04:41,000
management? let's go to Wolf.

80
00:04:41,000 --> 00:04:44,519
Wolf Vollprecht: Yeah, so for
sure. So package management is

81
00:04:44,519 --> 00:04:48,155
something that I've been dealing
with for the past five years,

82
00:04:48,155 --> 00:04:51,615
and the idea of a package is, a
software package is just an

83
00:04:51,615 --> 00:04:54,782
artifact, like a tar ball,
essentially. sometimes it's

84
00:04:54,782 --> 00:04:58,124
compiled, sometimes it's not
compiled, but we are dealing

85
00:04:58,124 --> 00:05:01,408
with compiled packages, so
binary artifacts essentially.

86
00:05:01,408 --> 00:05:05,162
And then the rest of the package
is just on metadata, what's the

87
00:05:05,162 --> 00:05:08,505
name, what's the version, what
are your dependencies? And

88
00:05:08,505 --> 00:05:12,141
dependencies are important of
course. yeah. And then a package

89
00:05:12,141 --> 00:05:15,659
manager is just a tool that
makes sure that, you get all the

90
00:05:15,659 --> 00:05:19,413
packages you need, install them
in the right order, and have the

91
00:05:19,413 --> 00:05:22,873
right versions that you want.
And, Pixi is basically one of

92
00:05:22,873 --> 00:05:26,567
those package managers. And Pixi
operates on packages that come

93
00:05:26,567 --> 00:05:30,145
from the Conda ecosystem. And
Ruben briefly mentioned it, and

94
00:05:30,145 --> 00:05:33,781
Conda existed for a long time
before us, basically. and so it,

95
00:05:33,781 --> 00:05:37,241
there are already quite a few
packages available and that's

96
00:05:37,241 --> 00:05:39,000
what we're building on top of.

97
00:05:39,000 --> 00:05:43,908
Audrow Nash: Okay, so Conda is
where all the binary files that

98
00:05:43,908 --> 00:05:48,025
you would execute on your
computer for your specific

99
00:05:48,025 --> 00:05:52,933
operating system live. and Pixi
is a way of getting those in a

100
00:05:52,933 --> 00:05:58,000
nice way. what are the problems
that Pixi is solving to do this?

101
00:05:58,000 --> 00:06:01,906
Wolf Vollprecht: Yeah. we, so
Conda historically is

102
00:06:01,906 --> 00:06:06,579
implemented in Python and we
completely rewrote everything in

103
00:06:06,579 --> 00:06:11,174
Rust as you do as a startup.
And, we made it much faster, of

104
00:06:11,174 --> 00:06:15,540
course, and everything is like
thread safe and you can do

105
00:06:15,540 --> 00:06:20,060
things in parallel that you
could never dream about before.

106
00:06:20,060 --> 00:06:24,732
And that's enabled by choosing
Rust, I would say. and then on

107
00:06:24,732 --> 00:06:29,098
top of that, we also, come in
with a different vision. So

108
00:06:29,098 --> 00:06:34,000
Conda has one way of working and
we have another way of working.

109
00:06:34,000 --> 00:06:38,749
It's a bit like poetry and pip,
let's say. Where we are trying

110
00:06:38,749 --> 00:06:43,268
with Pixi, we're trying to have,
some of the modern package

111
00:06:43,268 --> 00:06:48,247
management, ideas like, having a
log file, having a single source

112
00:06:48,247 --> 00:06:52,230
of truth for all of your
dependencies in a Pixi toml

113
00:06:52,230 --> 00:06:56,826
file. And, integrating that very
well into one like nice CLI

114
00:06:56,826 --> 00:07:01,498
interface that the user can use.
And so lock files are really

115
00:07:01,498 --> 00:07:05,557
important to us, for example,
and we have taken great

116
00:07:05,557 --> 00:07:10,000
inspiration from the cargo world
to design our interfaces.

117
00:07:10,000 --> 00:07:14,463
Audrow Nash: Okay. Let's see.
okay, so you're building a very

118
00:07:14,463 --> 00:07:18,854
nice command line interface, the
CLI, for doing this. You're

119
00:07:18,854 --> 00:07:23,390
managing a lot of the, like the
parallelization and things you

120
00:07:23,390 --> 00:07:27,927
can do that they couldn't do in
Python, and you're doing it in

121
00:07:27,927 --> 00:07:32,317
rust, so it's super fast. and
then. One of the things that's

122
00:07:32,317 --> 00:07:36,268
interesting to me is the
different language support, I

123
00:07:36,268 --> 00:07:40,951
guess maybe cargo supports that
out of the box just with the, as

124
00:07:40,951 --> 00:07:45,122
long as you can make it into a
binary file or how is that

125
00:07:45,122 --> 00:07:46,000
included? I,

126
00:07:46,000 --> 00:07:49,980
Wolf Vollprecht: I think you can
say that, but Cargo doesn't

127
00:07:49,980 --> 00:07:53,959
really tackle the, really the
installation piece of software

128
00:07:53,959 --> 00:07:57,607
development. Like you can do
cargo install, but it will

129
00:07:57,607 --> 00:08:01,653
actually build everything from
source, which will take a long

130
00:08:01,653 --> 00:08:05,699
time and stuff like that. And in
our world with Pixi, like we

131
00:08:05,699 --> 00:08:09,944
deal with pre-built packages. So
installation is like very fast,

132
00:08:09,944 --> 00:08:12,000
as fast as you can download the

133
00:08:12,000 --> 00:08:13,000
Audrow Nash: them in the right

134
00:08:13,000 --> 00:08:15,320
Wolf Vollprecht: we just put
them in the right spot and we

135
00:08:15,320 --> 00:08:17,000
have like intelligent caching
and whatnot.

136
00:08:17,000 --> 00:08:21,410
Audrow Nash: Ah, okay. So as a
little more background, cargo is

137
00:08:21,410 --> 00:08:24,000
rusts package manager, right? Is
that

138
00:08:24,000 --> 00:08:26,657
Wolf Vollprecht: That is
correct. So cargo is very much

139
00:08:26,657 --> 00:08:29,507
rust specific. You can do some
like C stuff, but it doesn't

140
00:08:29,507 --> 00:08:32,454
install a C compiler for you.
You have to get your C compiler

141
00:08:32,454 --> 00:08:34,000
from somewhere else and whatnot.

142
00:08:34,000 --> 00:08:38,092
Audrow Nash: All the things.
Yeah. Okay. And so one thing I

143
00:08:38,092 --> 00:08:42,254
noticed when you were showing me
this before is, so cargo is

144
00:08:42,254 --> 00:08:46,347
really nicely set up in my
opinion. I love Rust and I love,

145
00:08:46,347 --> 00:08:50,439
like cargo is just fantastic.
And it does a lot of the lock

146
00:08:50,439 --> 00:08:54,879
files that you were mentioning.
So you basically constrain your,

147
00:08:54,879 --> 00:08:58,832
dependencies to be specific
versions. So you know it will

148
00:08:58,832 --> 00:09:03,133
work when you load it, which is
really nice, especially coming

149
00:09:03,133 --> 00:09:07,503
from like the Python world. but
as you said, cargo it, when you

150
00:09:07,503 --> 00:09:11,526
download a GitHub repository of,
rust Project, it installs

151
00:09:11,526 --> 00:09:15,827
everything by just compiling it
effectively. is that what it's

152
00:09:15,827 --> 00:09:19,642
doing? It's not, grabbing
artifacts, it's grabbing code

153
00:09:19,642 --> 00:09:22,000
and then turning it into
artifacts

154
00:09:22,000 --> 00:09:24,000
Wolf Vollprecht: is what it's
doing. Yeah.

155
00:09:24,000 --> 00:09:27,840
Audrow Nash: Okay, so you,
because robotics, we have a lot

156
00:09:27,840 --> 00:09:31,747
of software. It would take
forever to rebuild the whole ROS

157
00:09:31,747 --> 00:09:35,521
stack every time you're trying
to do anything. or even if

158
00:09:35,521 --> 00:09:39,758
you're just getting started. and
you need a lot of compute to do

159
00:09:39,758 --> 00:09:43,797
that, and it would take a long
time. It's much easier to just

160
00:09:43,797 --> 00:09:47,902
send these binary files for the
specific operating system that

161
00:09:47,902 --> 00:09:51,543
you're using. and so you had to
solve that a little bit

162
00:09:51,543 --> 00:09:53,000
differently with Pixi?

163
00:09:53,000 --> 00:09:57,212
Wolf Vollprecht: we're doing is
I, wanna go one step, back. so

164
00:09:57,212 --> 00:10:00,745
conda is like the package
format, but then there are

165
00:10:00,745 --> 00:10:05,025
different distributions of conda
packages. And the one, the two

166
00:10:05,025 --> 00:10:08,965
distributions that we are
related with is Conda Forge. And

167
00:10:08,965 --> 00:10:12,362
Robo Stack and these
distributions, they build the

168
00:10:12,362 --> 00:10:16,167
packages and we are trying to
build packages that are as

169
00:10:16,167 --> 00:10:20,379
compatible as possible with like
pretty old machines actually.

170
00:10:20,379 --> 00:10:24,387
and so you can install the
packages that we're building and

171
00:10:24,387 --> 00:10:28,667
conda forge and robo stack with
machines that are like 10 years

172
00:10:28,667 --> 00:10:32,539
old and have like old CPUs
because we're making sure that

173
00:10:32,539 --> 00:10:36,684
we set the right compiler flags
and stuff like that to build,

174
00:10:36,684 --> 00:10:40,624
executable binaries that are
like very compatible. And the

175
00:10:40,624 --> 00:10:44,836
only thing that we need from the
host system usually is like a

176
00:10:44,836 --> 00:10:48,844
version of let's say glibc on
Linux and some Mac Os, SDK on

177
00:10:48,844 --> 00:10:52,988
Mac os, et cetera. But that's
like technical details. But the

178
00:10:52,988 --> 00:10:56,997
thing is like you get a binary
package that you can install

179
00:10:56,997 --> 00:11:01,005
rapidly on your computer and
imagine you would have, like a

180
00:11:01,005 --> 00:11:04,945
room full of students and
everyone would have to build ROS

181
00:11:04,945 --> 00:11:09,361
from source that would. Would be
pretty annoying and would take a

182
00:11:09,361 --> 00:11:13,505
long time and everyone would
probably have some pretty, yeah.

183
00:11:13,505 --> 00:11:15,000
Annoying difficulties.

184
00:11:15,000 --> 00:11:18,499
Audrow Nash: Yeah, I was at,
ROSCon last year at a workshop

185
00:11:18,499 --> 00:11:21,820
and everyone was trying to
download the software for the

186
00:11:21,820 --> 00:11:25,378
workshop at the beginning of the
workshop, and it was on the

187
00:11:25,378 --> 00:11:28,759
hotel wifi or a conference venue
wifi. And it was a total

188
00:11:28,759 --> 00:11:32,436
nightmare. And it was supposed
to take a very, they budgeted a

189
00:11:32,436 --> 00:11:35,935
very small period of time for
everyone to be set up, and it

190
00:11:35,935 --> 00:11:39,434
took like hours, like it took
like half the workshop to get

191
00:11:39,434 --> 00:11:43,288
everyone set up, which was nuts.
Like the whole day workshop took

192
00:11:43,288 --> 00:11:44,000
like half of

193
00:11:44,000 --> 00:11:45,000
Wolf Vollprecht: yeah,

194
00:11:45,000 --> 00:11:47,000
Audrow Nash: Um, that's a big
pain.

195
00:11:47,000 --> 00:11:50,851
Wolf Vollprecht: that is exactly
what we had. we've had people

196
00:11:50,851 --> 00:11:54,578
use Pixi at workshops before,
not specifically for robotics,

197
00:11:54,578 --> 00:11:57,000
but it's usually a pretty good
success.

198
00:11:57,000 --> 00:12:01,176
Audrow Nash: Yeah. And so
jumping not exactly ahead. Wolf,

199
00:12:01,176 --> 00:12:05,639
you and I have talked before. at
length. And one of the things

200
00:12:05,639 --> 00:12:10,031
that I thought was really clever
about the thing you guys are

201
00:12:10,031 --> 00:12:14,494
doing is, so in the way we do
things now in ROS, it'll require

202
00:12:14,494 --> 00:12:18,814
a specific version of CMake or
some other software libraries

203
00:12:18,814 --> 00:12:22,486
and these ship with the
operating system. So we are

204
00:12:22,486 --> 00:12:26,805
very, we use Ubuntu 2204 and
that ships with this version of

205
00:12:26,805 --> 00:12:30,981
CMake and that version of
something else. And we use those

206
00:12:30,981 --> 00:12:35,373
system dependencies. And that's
nice because we just say, you

207
00:12:35,373 --> 00:12:39,836
use this operating system and it
will, you don't have to worry

208
00:12:39,836 --> 00:12:44,012
about installing all these
things. or you just install the

209
00:12:44,012 --> 00:12:48,403
default one that APT is pointing
you to, But one thing you're

210
00:12:48,403 --> 00:12:52,579
doing is you make it so that you
download a binary for the

211
00:12:52,579 --> 00:12:57,187
dependency that you have, which
might be c make when you need it

212
00:12:57,187 --> 00:13:01,650
in Pixi, and you're leveraging
what Conda does already because

213
00:13:01,650 --> 00:13:05,610
all these exist and are
maintained by other people than

214
00:13:05,610 --> 00:13:09,786
you. You're just grabbing the
appropriate ones. And to me,

215
00:13:09,786 --> 00:13:14,321
this was a very, clever approach
and made it so you didn't have

216
00:13:14,321 --> 00:13:18,569
to do a lot of things, but you
get a lot of crosss platform

217
00:13:18,569 --> 00:13:22,744
compatibility because you're
just downloading the resource

218
00:13:22,744 --> 00:13:26,704
you need. Is that correct? I'm
sure I didn't say it all

219
00:13:26,704 --> 00:13:28,000
correct. But, what

220
00:13:28,000 --> 00:13:31,710
Wolf Vollprecht: that, yeah, no,
I think that's a very good

221
00:13:31,710 --> 00:13:35,546
example. right now, if you want
to get CMake, let's say, in a

222
00:13:35,546 --> 00:13:39,318
crosss platform fashion, like I
wouldn't even know how to do

223
00:13:39,318 --> 00:13:42,903
that outside of the conda
ecosystem because on Mac Os you

224
00:13:42,903 --> 00:13:46,801
would maybe use Brew on, Linux,
you would use apt get, or DNFI

225
00:13:46,801 --> 00:13:50,700
don't know what, and on Windows
maybe Wing it. I don't know if

226
00:13:50,700 --> 00:13:54,661
it even has cmake, right. And,
so we are working on top of that

227
00:13:54,661 --> 00:13:58,434
ecosystem that's called Conda
Forge. And that is like 20,000

228
00:13:58,434 --> 00:14:01,830
git repositories that are
running, package bills in CI

229
00:14:01,830 --> 00:14:05,477
pipelines. And over 5,000
individuals that are maintaining

230
00:14:05,477 --> 00:14:09,375
all of these packages. and they
are maintaining cmake, they're

231
00:14:09,375 --> 00:14:12,959
maintaining meison, they're
maintaining Python and c plus

232
00:14:12,959 --> 00:14:16,669
compilers, et cetera. So all
these low level stuff, like we

233
00:14:16,669 --> 00:14:20,505
have a big community that is
already helping us with building

234
00:14:20,505 --> 00:14:24,026
it. And then what we are
setting, what we are putting on

235
00:14:24,026 --> 00:14:27,862
top of that is the robo stack,
which is like the ROS specific

236
00:14:27,862 --> 00:14:30,000
packages that we are building
out.

237
00:14:30,000 --> 00:14:34,016
Audrow Nash: Yes. So tell me
about Robo Stack with this. So

238
00:14:34,016 --> 00:14:38,168
what is it at a high level and,
how, does it work? how do you

239
00:14:38,168 --> 00:14:42,319
manage the artifacts you build
from it? And how does all the,

240
00:14:42,319 --> 00:14:43,000
that work?

241
00:14:43,000 --> 00:14:46,857
Wolf Vollprecht: Yeah, so the
idea of Robo Stack is to

242
00:14:46,857 --> 00:14:51,143
basically, take the two ideas
like the ROS distributions and

243
00:14:51,143 --> 00:14:55,571
the conduct packages and then
merge them together somehow. And

244
00:14:55,571 --> 00:14:59,500
what we have is, we have CI
pipelines that are building

245
00:14:59,500 --> 00:15:04,071
these thousands of packages that
exist in the ROS Universe, into

246
00:15:04,071 --> 00:15:08,214
conduct packages. And, we are
leveraging, one thing that's

247
00:15:08,214 --> 00:15:12,500
really nice about ROS is that
you have pretty good metadata.

248
00:15:12,500 --> 00:15:16,643
So every package comes with a
package XML, it comes with a

249
00:15:16,643 --> 00:15:20,786
cmake, usually, et cetera. And
like they have a very nice,

250
00:15:20,786 --> 00:15:24,857
infrastructure to build Ubuntu
packages. And we are doing

251
00:15:24,857 --> 00:15:29,286
something similar. We have our
own infrastructure that's a bit

252
00:15:29,286 --> 00:15:32,714
different, that is
automatically, converting the

253
00:15:32,714 --> 00:15:37,357
packages from ROS Distro, to the
Conda format and then running it

254
00:15:37,357 --> 00:15:41,857
through CI to build everything
using our tool chain. And so you

255
00:15:41,857 --> 00:15:45,857
have I wouldn't wanna say 80%
maybe on Linux of packages

256
00:15:45,857 --> 00:15:49,786
available in Robo stack
repositories. And it's also, we

257
00:15:49,786 --> 00:15:53,429
try to make it very easy to
contribute and, add new

258
00:15:53,429 --> 00:15:57,429
packages. Let's say if you're
missing something, we have

259
00:15:57,429 --> 00:16:01,214
every, we try to automate as
much as possible and ROS

260
00:16:01,214 --> 00:16:04,000
infrastructure is making that
possible.

261
00:16:04,000 --> 00:16:07,343
Audrow Nash: Yeah, it's very
clever. This was another thing

262
00:16:07,343 --> 00:16:10,799
when we talked last, how you
guys are leveraging the existing

263
00:16:10,799 --> 00:16:14,482
ROS infrastructure to make it so
you don't have to build the jobs

264
00:16:14,482 --> 00:16:18,164
again, and you also get a lot of
metadata or maybe you do have to

265
00:16:18,164 --> 00:16:21,507
build the jobs, but you get a
lot of metadata from the jobs

266
00:16:21,507 --> 00:16:24,000
that are set up on the ROS 2
build farm, for

267
00:16:24,000 --> 00:16:27,845
Wolf Vollprecht: we do have to
build all the packages ourselves

268
00:16:27,845 --> 00:16:31,446
with our compilers. And we also
have to do that for Mac and

269
00:16:31,446 --> 00:16:35,230
Windows, which is like there,
like I think Mac isn't supported

270
00:16:35,230 --> 00:16:37,000
really anymore and Windows is

271
00:16:37,000 --> 00:16:38,000
Audrow Nash: saying by ROS 2?

272
00:16:38,000 --> 00:16:39,000
Wolf Vollprecht: by ROS. Yeah,

273
00:16:39,000 --> 00:16:43,300
Audrow Nash: We, it's a tier
three, so it's from source. but

274
00:16:43,300 --> 00:16:47,528
no one guarantees it's working.
So from the official ROS, I

275
00:16:47,528 --> 00:16:51,612
don't know, like from open
robotics, open source robotics

276
00:16:51,612 --> 00:16:55,912
foundation, we don't guarantee
it's working on Mac. so yeah.

277
00:16:55,912 --> 00:17:00,212
You're saying bad, but it does
work with your, with Pixi, Or

278
00:17:00,212 --> 00:17:01,000
most things

279
00:17:01,000 --> 00:17:04,000
Wolf Vollprecht: Most things
work pretty nicely actually.

280
00:17:04,000 --> 00:17:07,798
Audrow Nash: I, hell yeah. And
so you were saying, I, just to

281
00:17:07,798 --> 00:17:11,471
make it clear what you were
saying earlier, you were saying

282
00:17:11,471 --> 00:17:15,019
about 80% of the packages that
are available on Linux are

283
00:17:15,019 --> 00:17:18,444
available on other operating
systems with Pixi. Is that

284
00:17:18,444 --> 00:17:20,000
correct? Did I understand

285
00:17:20,000 --> 00:17:21,000
Wolf Vollprecht: not quite.

286
00:17:21,000 --> 00:17:22,000
Audrow Nash: Okay.

287
00:17:22,000 --> 00:17:26,317
Wolf Vollprecht: I think a good
chunk of packages, from ROS are

288
00:17:26,317 --> 00:17:30,635
available on Linux, but then a
different subset is available on

289
00:17:30,635 --> 00:17:34,678
Macs and Windows. On Robo stack.
so it's not, equal because

290
00:17:34,678 --> 00:17:38,516
sometimes, sometimes you have
things that are very Linux

291
00:17:38,516 --> 00:17:42,559
specific. Sometimes you have
packages that just don't build

292
00:17:42,559 --> 00:17:46,808
on windows for reasons, and we
would have to go and patch them

293
00:17:46,808 --> 00:17:51,057
and things like that. we do that
for some of them. and we also

294
00:17:51,057 --> 00:17:55,238
try to upstream patches from
time to time. but a lot of, like

295
00:17:55,238 --> 00:17:59,144
one, like a lot of people don't
really care so much about

296
00:17:59,144 --> 00:18:03,256
Windows as we do, in the ROS
world, I wanna say. and testing

297
00:18:03,256 --> 00:18:07,299
is not as widely, done. And
sometimes the MSVC compiler has

298
00:18:07,299 --> 00:18:11,000
different bug than the GCC
compiler, et cetera. And so

299
00:18:11,000 --> 00:18:15,923
Audrow Nash: I know. Yeah. It's
so funny. For sure. Okay, so you

300
00:18:15,923 --> 00:18:20,538
have, I just wanna see if I
understand everything. this will

301
00:18:20,538 --> 00:18:24,923
be a real test to see. So we
have Pixi, Pixi is a package

302
00:18:24,923 --> 00:18:29,462
manager. Pixi makes it so you
can download pretty arbitrary

303
00:18:29,462 --> 00:18:34,000
code artifacts, but it's very
useful for robotics. you have

304
00:18:34,000 --> 00:18:38,846
conda, which is metadata, that
you have four packages that make

305
00:18:38,846 --> 00:18:43,385
it very easy to, have some
artifacts, but then have it very

306
00:18:43,385 --> 00:18:47,769
clear what it depends on for
this kind of thing. Then you

307
00:18:47,769 --> 00:18:51,846
have Conda Forge, which is a
place that a lot of like

308
00:18:51,846 --> 00:18:55,769
operating system level
dependencies, like CMake are

309
00:18:55,769 --> 00:19:00,000
stored and the community
maintains that and you can put

310
00:19:00,000 --> 00:19:04,923
more things there. And then you
have Robo Stack, which is a very

311
00:19:04,923 --> 00:19:09,769
similar to Condo Forge, but for
ROS packages. so it's basically

312
00:19:09,769 --> 00:19:14,308
a big list of all the artifacts
that you have access to and

313
00:19:14,308 --> 00:19:18,538
probably where they work. is
that correct or where am I

314
00:19:18,538 --> 00:19:19,000
wrong?

315
00:19:19,000 --> 00:19:22,000
Wolf Vollprecht: That's a very
good summary. I.

316
00:19:22,000 --> 00:19:23,000
Ruben Arts: That

317
00:19:22,000 --> 00:19:23,863
Audrow Nash: Okay. I just want
to make sure that I'm there.

318
00:19:23,863 --> 00:19:25,000
Ruben, did you have anything to
add?

319
00:19:25,000 --> 00:19:28,000
Ruben Arts: No, that, that
works. That.

320
00:19:28,000 --> 00:19:30,000
Audrow Nash: Okay. Hell yeah.

321
00:19:30,000 --> 00:19:34,244
Audrow Nash: and then so the
next thing that I wanna see

322
00:19:34,244 --> 00:19:39,020
before we get into our demos is
I just wanna talk about Prefix.

323
00:19:39,020 --> 00:19:43,416
so Pixi, the way of downloading
arbitrary artifacts that's

324
00:19:43,416 --> 00:19:48,342
really useful for robotics tasks
is maintained, built by you guys

325
00:19:48,342 --> 00:19:53,117
at Prefix. Let's see, Wolf, will
you tell me a bit about prefix

326
00:19:53,117 --> 00:19:57,817
and then I'd love to also get
into the business model and just

327
00:19:57,817 --> 00:20:01,000
all the details about you guys
real quick.

328
00:20:01,000 --> 00:20:05,657
Wolf Vollprecht: Yeah. So Prefix
is a small company. We're six

329
00:20:05,657 --> 00:20:10,315
right now. It's a startup. we
have some VC backing, let's say.

330
00:20:10,315 --> 00:20:14,972
and, we are working on package
management in general, not only

331
00:20:14,972 --> 00:20:19,704
for robotics, but also for like
data science or essentially any

332
00:20:19,704 --> 00:20:24,136
other applications where you
need packages and, yeah. we've

333
00:20:24,136 --> 00:20:28,493
been working very hard on
building out a platform and Pixi

334
00:20:28,493 --> 00:20:33,000
is our open source tooling. So
what we're doing currently is

335
00:20:33,000 --> 00:20:37,357
build the open source tooling
that everyone hopefully will

336
00:20:37,357 --> 00:20:41,789
love and use and it's supposed
to make dealing with conduct

337
00:20:41,789 --> 00:20:46,596
packages so much better than it
ever was. and that will be like,

338
00:20:46,596 --> 00:20:51,028
it is 100% free. We are like
total open source people. but,

339
00:20:51,028 --> 00:20:55,685
obviously what we're trying to
do is build an awesome platform

340
00:20:55,685 --> 00:21:00,268
as well where you can like, deal
with packages and build some

341
00:21:00,268 --> 00:21:04,850
enterprise products that are
interesting, like security stuff

342
00:21:04,850 --> 00:21:09,207
like CVEs package signing.
Getting, packages that are like

343
00:21:09,207 --> 00:21:14,014
signed by an entity like us, for
example, like a trusted sort of

344
00:21:14,014 --> 00:21:18,596
minimal distribution of your
core tooling that you wanna make

345
00:21:18,596 --> 00:21:23,103
sure that you, is secure and,
safe. That's something that we

346
00:21:23,103 --> 00:21:27,685
are like working towards, but we
are not there. And right now

347
00:21:27,685 --> 00:21:32,268
we're working a lot with the
open source community on getting

348
00:21:32,268 --> 00:21:37,000
better packages and yeah, just
improving the overall ecosystem.

349
00:21:37,000 --> 00:21:41,880
Audrow Nash: So are there, I'm
sure that there are. but what,

350
00:21:41,880 --> 00:21:46,440
companies are similar in
business model? Is it like a Red

351
00:21:46,440 --> 00:21:47,000
Hat or,

352
00:21:47,000 --> 00:21:48,000
Wolf Vollprecht: yeah.

353
00:21:48,000 --> 00:21:52,146
Audrow Nash: like what would be
a fair comparison to like the

354
00:21:52,146 --> 00:21:55,000
model that you guys are trying
to achieve?

355
00:21:55,000 --> 00:21:58,429
Wolf Vollprecht: I think like
there are definitely some,

356
00:21:58,429 --> 00:22:02,347
companies that are basically in
the Linux space that are. Making

357
00:22:02,347 --> 00:22:06,204
money out of software packages,
canonical Red Hat openSUSE. and

358
00:22:06,204 --> 00:22:09,939
then you have a new company, for
example, that's called Chain

359
00:22:09,939 --> 00:22:13,490
Guard. They just raised a
series, I don't know, B or c 120

360
00:22:13,490 --> 00:22:16,000
million. So we are not at that
stage yet.

361
00:22:16,000 --> 00:22:18,000
Audrow Nash: Wow.

362
00:22:18,000 --> 00:22:21,894
Wolf Vollprecht: but we are to
some degree generalizing ideas

363
00:22:21,894 --> 00:22:25,915
that these other companies have,
right? Like we don't only deal

364
00:22:25,915 --> 00:22:29,809
with Linux packages, but we also
deal with Mac Os and Windows

365
00:22:29,809 --> 00:22:33,447
packages. And so there is a
pretty big market for Windows

366
00:22:33,447 --> 00:22:37,468
people that is often overlooked.
And so I think there's, 80% of

367
00:22:37,468 --> 00:22:41,489
developers are using Windows and
then don't have a good package

368
00:22:41,489 --> 00:22:45,000
manager. So that's, something
that we're trying to fix.

369
00:22:45,000 --> 00:22:49,207
Audrow Nash: Yeah, very much.
Okay. And I love the idea, I am

370
00:22:49,207 --> 00:22:53,138
often programming in several
languages, rust, TypeScript,

371
00:22:53,138 --> 00:22:57,276
Python, c plus, and sometimes
projects will have multiple of

372
00:22:57,276 --> 00:23:01,483
these languages all in the same
project. And so the idea of a

373
00:23:01,483 --> 00:23:05,759
simple setup is so appealing and
just, I know it's been such a

374
00:23:05,759 --> 00:23:10,034
pain for different ROS, for just
general ROS development, so I

375
00:23:10,034 --> 00:23:14,379
really like what you're doing.
What's, for the, let's see, for,

376
00:23:14,379 --> 00:23:18,448
you mentioned that you are
venture backed. did you have, is

377
00:23:18,448 --> 00:23:22,379
it like a seed round or did you
have, like, how would you

378
00:23:22,379 --> 00:23:23,000
classify,

379
00:23:23,000 --> 00:23:28,069
Wolf Vollprecht: we are, we are
at the seed, stage, let's say.

380
00:23:28,069 --> 00:23:32,811
Yeah, that's correct. So we have
two VC investors that are

381
00:23:32,811 --> 00:23:36,000
believing in our ambition, and
so yeah,

382
00:23:36,000 --> 00:23:39,767
Audrow Nash: Hell yeah. And
then, I wonder, are you're

383
00:23:39,767 --> 00:23:43,535
currently not ren re revenue
generating because you're

384
00:23:43,535 --> 00:23:48,000
focusing on community buy-in at
the moment. Is that correct or I

385
00:23:48,000 --> 00:23:52,300
Wolf Vollprecht: focusing on
community right now. Yeah, we

386
00:23:52,300 --> 00:23:56,971
are. just, there's. So Conda at
its core is good, but there's a

387
00:23:56,971 --> 00:24:01,271
lot that we are currently
fixing. And so we are very happy

388
00:24:01,271 --> 00:24:05,942
that we have investors that are
also supporting this mission of

389
00:24:05,942 --> 00:24:10,465
really trying to really, do the
right things in the ecosystem

390
00:24:10,465 --> 00:24:14,468
and making the package
management experience nice. And

391
00:24:14,468 --> 00:24:18,472
also the package building
experience, which is another

392
00:24:18,472 --> 00:24:22,772
thing that we are like really
investing into right now. So

393
00:24:22,772 --> 00:24:27,443
that's current that, that's like
step one for us. And then step

394
00:24:27,443 --> 00:24:29,000
two is like the rest.

395
00:24:29,000 --> 00:24:33,219
Audrow Nash: Yeah, for sure.
what do you, where are you guys

396
00:24:33,219 --> 00:24:37,438
located? I think you're remote.
and where are your investors

397
00:24:37,438 --> 00:24:38,000
located?

398
00:24:38,000 --> 00:24:42,307
Wolf Vollprecht: Yeah. So I'm in
Berlin and Rub is in the

399
00:24:42,307 --> 00:24:46,916
Netherlands and we have a team
in the Netherlands, let's say,

400
00:24:46,916 --> 00:24:51,676
and two people in Germany right
now and one in Moldova. And our

401
00:24:51,676 --> 00:24:55,000
investors are like. American
German mixture,

402
00:24:55,000 --> 00:24:58,027
Audrow Nash: Oh wow. You said
two investors, or two? Two

403
00:24:58,027 --> 00:25:01,000
investors. So one, one American,
one German. Hell yeah.

404
00:25:01,000 --> 00:25:03,000
Wolf Vollprecht: yeah.

405
00:25:03,000 --> 00:25:06,399
Audrow Nash: Okay, that's
awesome. very happy to hear

406
00:25:06,399 --> 00:25:10,312
that. And happy you found some
good investors 'cause it seems

407
00:25:10,312 --> 00:25:14,289
like a really nice mission. So
with that, Ruben, I know you'll

408
00:25:14,289 --> 00:25:17,880
be talking a lot in the later
parts when we do demos and

409
00:25:17,880 --> 00:25:22,050
things, but do you have anything
to add or Wolf also, if you have

410
00:25:22,050 --> 00:25:25,000
anything to add before we jump
into, I. Demos.

411
00:25:25,000 --> 00:25:28,831
Ruben Arts: I think Wolf
explained it pretty well. I, I.

412
00:25:28,831 --> 00:25:33,140
Ex having so much fun with the
fact that we are open source and

413
00:25:33,140 --> 00:25:37,381
we're focusing on that part so
hard right now. So it's just an

414
00:25:37,381 --> 00:25:41,691
amazing experience, would advise
everyone to do that with their

415
00:25:41,691 --> 00:25:45,248
company. From a employee
standpoint, it's an amazing

416
00:25:45,248 --> 00:25:46,000
experience.

417
00:25:46,000 --> 00:25:50,000
Audrow Nash: yeah, It's nice
being open source. hell yeah.

418
00:25:50,000 --> 00:25:52,542
Audrow Nash: Okay, so with that
Wolf, you wanna jump into our

419
00:25:52,542 --> 00:25:53,000
first demo?

420
00:25:53,000 --> 00:25:58,728
Wolf Vollprecht: Yes. let's do
that. let me try to share my

421
00:25:58,728 --> 00:26:03,000
entire screen. Do we have the
infinity? Yes.

422
00:26:03,000 --> 00:26:04,000
Ruben Arts: We did.

423
00:26:04,000 --> 00:26:09,091
Wolf Vollprecht: Infinity,
achieve. so yeah, I am, we've

424
00:26:09,091 --> 00:26:14,636
been talking a bit about Pixi
and how to use conduct packages

425
00:26:14,636 --> 00:26:19,909
and I'm gonna show you guys, one
Pixi project now. so if I

426
00:26:19,909 --> 00:26:23,000
look@thePixi.il file, oh, I
don't,

427
00:26:23,000 --> 00:26:24,000
Audrow Nash: the font nice and
big?

428
00:26:24,000 --> 00:26:30,000
Wolf Vollprecht: is it not big
enough? is this big enough?

429
00:26:30,000 --> 00:26:33,664
Audrow Nash: If you could make
it a little larger? 'cause

430
00:26:33,664 --> 00:26:37,264
people may be watching on
smartphones, you never know. I

431
00:26:37,264 --> 00:26:39,000
think that's probably Okay.

432
00:26:39,000 --> 00:26:42,888
Wolf Vollprecht: Okay, so I was
trying to look at the Pixi to

433
00:26:42,888 --> 00:26:46,968
file and it didn't work. So I'm
gonna do one thing that Pixi can

434
00:26:46,968 --> 00:26:50,920
do, which is globally install me
some tools. So I want to have

435
00:26:50,920 --> 00:26:55,000
that globally installed. and now
I can look at the Pixi to file.

436
00:26:55,000 --> 00:26:58,853
Audrow Nash: the, so what's the
significance of that? Why is

437
00:26:58,853 --> 00:27:02,000
that impressive? Because I could
just apt get it,

438
00:27:02,000 --> 00:27:03,000
Wolf Vollprecht: yeah,

439
00:27:03,000 --> 00:27:06,354
Audrow Nash: Why is BAT
installing it? Or why, is, Pixi

440
00:27:06,354 --> 00:27:08,000
installing it a nice thing?

441
00:27:08,000 --> 00:27:15,000
Wolf Vollprecht: can you app get
it on, on Windows? Yeah.

442
00:27:15,000 --> 00:27:18,000
Audrow Nash: need to figure out
whatever Windows does for

443
00:27:18,000 --> 00:27:21,899
Wolf Vollprecht: yeah, so, with
Pixi you can just globally

444
00:27:21,899 --> 00:27:25,731
install some tools. And these
tools, they don't come from

445
00:27:25,731 --> 00:27:29,899
first of all, con Forge moves
faster usually than a PT get. So

446
00:27:29,899 --> 00:27:34,000
you'll get newer versions much
more quickly with Conda Forge.

447
00:27:34,000 --> 00:27:37,000
Audrow Nash: Oh, that's always
such a pain. Yeah.

448
00:27:37,000 --> 00:27:38,000
Ruben Arts: No more ripple
stories,

449
00:27:38,000 --> 00:27:42,000
Audrow Nash: then also there's a
crosss Go ahead. Sorry,

450
00:27:42,000 --> 00:27:45,916
Ruben Arts: okay, so no more
setting the repositories and

451
00:27:45,916 --> 00:27:50,244
figuring out which version you
need and all that stuff that you

452
00:27:50,244 --> 00:27:51,000
need to do.

453
00:27:51,000 --> 00:27:57,358
Audrow Nash: I as a, completely
selfish. Yeah, wondering, can I

454
00:27:57,358 --> 00:28:02,000
install Neo Vim to an arbitrary
version pretty

455
00:28:02,000 --> 00:28:05,000
Wolf Vollprecht: Yeah, you can
do Pixi Global Install En Vim.

456
00:28:05,000 --> 00:28:08,766
Audrow Nash: because Neo Vim is
such a pain to install on.

457
00:28:08,766 --> 00:28:12,468
'cause it, gets the wrong
version for, the plugins that I

458
00:28:12,468 --> 00:28:15,000
want to use when I try to use
Apt and I

459
00:28:15,000 --> 00:28:16,000
Wolf Vollprecht: There you go.

460
00:28:16,000 --> 00:28:19,758
Audrow Nash: Oh, that's awesome.
Yeah. Space Vim. Oh, so cool.

461
00:28:19,758 --> 00:28:20,000
Hell

462
00:28:20,000 --> 00:28:22,000
Wolf Vollprecht: I don't know
why.

463
00:28:22,000 --> 00:28:24,000
Audrow Nash: be using Lazy Vim,
but that's

464
00:28:24,000 --> 00:28:26,051
Wolf Vollprecht: but basically
here you see the version that

465
00:28:26,051 --> 00:28:28,000
was installed and you could do
something like Pixi Global

466
00:28:28,000 --> 00:28:31,000
Audrow Nash: Yeah. 10 is what I
want. Oh, that's so cool. Yeah,

467
00:28:31,000 --> 00:28:34,557
Wolf Vollprecht: and then
whatever, I don't know what

468
00:28:34,557 --> 00:28:38,651
versions are available, but you
can set it and then you get a

469
00:28:38,651 --> 00:28:41,000
different one. but back to our
ROS.

470
00:28:41,000 --> 00:28:42,000
Audrow Nash: me.

471
00:28:42,000 --> 00:28:46,098
Wolf Vollprecht: so what we're
doing here is we have a Pixi

472
00:28:46,098 --> 00:28:49,988
TOML file that, just defines the
project. And so this is

473
00:28:49,988 --> 00:28:54,364
something we copied from cargo,
for example. we are using these

474
00:28:54,364 --> 00:28:57,976
channels that I was talking
about, and, we have some

475
00:28:57,976 --> 00:29:02,074
dependencies and in this case we
depend on turtle bot three

476
00:29:02,074 --> 00:29:06,242
gazebo, and we also have a task,
which is, called start. And

477
00:29:06,242 --> 00:29:10,201
that's this entry point that I
can use. So I just do Pixi

478
00:29:10,201 --> 00:29:14,230
start, and that's everything,
right? that's everything you

479
00:29:14,230 --> 00:29:17,981
need to do or your users have to
do, to set up all the

480
00:29:17,981 --> 00:29:21,315
dependencies, to install
everything and download

481
00:29:21,315 --> 00:29:25,274
everything from the internet.
And it starts gazebo and it

482
00:29:25,274 --> 00:29:29,650
starts, rvu as you can see. So
getting to that point, usually I

483
00:29:29,650 --> 00:29:33,540
think, on Linux or other
operating systems usually takes

484
00:29:33,540 --> 00:29:37,708
quite the time, And, now this,
works by first setting a post

485
00:29:37,708 --> 00:29:40,000
estimate. So I'm just doing
that.

486
00:29:40,000 --> 00:29:41,000
Audrow Nash: ba

487
00:29:40,000 --> 00:29:41,000
Wolf Vollprecht: then I can do
the NAF goal.

488
00:29:41,000 --> 00:29:44,000
Audrow Nash: thing to highlight
this

489
00:29:44,000 --> 00:29:46,000
Wolf Vollprecht: This is on Mac.
Yeah.

490
00:29:46,000 --> 00:29:47,000
Audrow Nash: that's super,

491
00:29:47,000 --> 00:29:51,632
Wolf Vollprecht: me also get the
robot model activated. this is

492
00:29:51,632 --> 00:29:55,824
using all of these robo stack
packages that I was talking

493
00:29:55,824 --> 00:29:57,000
about and, yeah.

494
00:29:57,000 --> 00:29:59,000
Ruben Arts: Yeah, so maybe could
to elaborate,

495
00:29:59,000 --> 00:30:00,000
Audrow Nash: cool.

496
00:30:00,000 --> 00:30:04,279
Ruben Arts: So maybe could you
elaborate on why this specific

497
00:30:04,279 --> 00:30:08,487
example is not working on the
windows. So we can package any

498
00:30:08,487 --> 00:30:12,555
package in theory, but the
package itself needs to support

499
00:30:12,555 --> 00:30:16,834
the platform that we build it
for. So in this case, I believe

500
00:30:16,834 --> 00:30:20,481
some of the plugins of the
Turtle SIM didn't have an

501
00:30:20,481 --> 00:30:24,760
buildable project on Windows, so
we're not able to support it

502
00:30:24,760 --> 00:30:29,108
there yet. But if this community
gets more interest in getting

503
00:30:29,108 --> 00:30:33,176
projects, also working on
Windows, this tool allows you to

504
00:30:33,176 --> 00:30:35,000
easily test on it as well.

505
00:30:35,000 --> 00:30:39,662
Wolf Vollprecht: Yeah, and
especially also like having more

506
00:30:39,662 --> 00:30:44,323
projects using, for example,
robust stack packages in CI to

507
00:30:44,323 --> 00:30:49,301
test against Windows. That would
be one thing that would help a

508
00:30:49,301 --> 00:30:54,121
lot. but generally speaking,
I'm, you can run Gazebo and Rviz

509
00:30:54,121 --> 00:30:58,862
on Windows as well, and you can
connect Rviz to, your remote

510
00:30:58,862 --> 00:31:03,602
data source or your Linux on
Windows and then, get your data

511
00:31:03,602 --> 00:31:07,000
and visualize it and stuff like
that. yeah.

512
00:31:07,000 --> 00:31:10,427
Audrow Nash: Love it. That's so
super cool. and it looks like

513
00:31:10,427 --> 00:31:12,000
the performance is very good

514
00:31:12,000 --> 00:31:13,000
Wolf Vollprecht: Yeah,

515
00:31:13,000 --> 00:31:16,405
Audrow Nash: Like I'm impressed
by how things are moving around

516
00:31:16,405 --> 00:31:17,000
so quickly.

517
00:31:17,000 --> 00:31:18,000
Wolf Vollprecht: while doing a
video call.

518
00:31:18,000 --> 00:31:22,385
Audrow Nash: faster factor in,
realtime one. Wow. So your

519
00:31:22,385 --> 00:31:26,000
gazebo, real-time factor is one
That's amazing.

520
00:31:26,000 --> 00:31:30,337
Wolf Vollprecht: Yeah, this is
on an M two, MacBook Pro, and

521
00:31:30,337 --> 00:31:32,000
I'm apparently not able

522
00:31:32,000 --> 00:31:34,000
Audrow Nash: very performant,
but still.

523
00:31:34,000 --> 00:31:36,774
Ruben Arts: It's also
interesting about how we do

524
00:31:36,774 --> 00:31:40,396
things because we are just using
packages directly on your host.

525
00:31:40,396 --> 00:31:43,453
There's no docker layer in
between. There's no virtual

526
00:31:43,453 --> 00:31:46,736
machine running anywhere. It's
just. hard metal that we're

527
00:31:46,736 --> 00:31:50,245
running this on while it's also
isolated from the rest of your

528
00:31:50,245 --> 00:31:53,755
system. So if Wolf throws away
this folder, you don't have ROS

529
00:31:53,755 --> 00:31:57,038
on your system anymore. If you
move to another folder, you

530
00:31:57,038 --> 00:32:00,660
don't have ROS there. It's like
super contained while using bare

531
00:32:00,660 --> 00:32:01,000
metal.

532
00:32:01,000 --> 00:32:02,000
Audrow Nash: Yep.

533
00:32:02,000 --> 00:32:05,000
Wolf Vollprecht: and so I also
want to put, yeah.

534
00:32:05,000 --> 00:32:08,282
Audrow Nash: to NPM or something
like that. So you have your

535
00:32:08,282 --> 00:32:11,291
working directory and then
you're gonna install all the

536
00:32:11,291 --> 00:32:14,300
dependencies as a child
directory of that. So like your

537
00:32:14,300 --> 00:32:17,582
NPN projects has their node
modules folder, which is joke to

538
00:32:17,582 --> 00:32:20,755
be the heaviest thing in the
universe 'cause it's normally

539
00:32:20,755 --> 00:32:24,201
has a lot of stuff. but then, so
you're doing something similar

540
00:32:24,201 --> 00:32:27,155
where you put all the
dependencies in there and that's

541
00:32:27,155 --> 00:32:30,000
really nice because your
projects are very isolated,

542
00:32:30,000 --> 00:32:34,050
Wolf Vollprecht: So one of the
things that is, maybe mind

543
00:32:34,050 --> 00:32:38,100
blowing for some ROS people is
that you can have a noetic

544
00:32:38,100 --> 00:32:42,221
project and a humble project at
the same time, on the same

545
00:32:42,221 --> 00:32:46,768
machine without having to like,
use any docker without having to

546
00:32:46,768 --> 00:32:50,961
use any, like virtual machines
and whatnot. yeah, and that,

547
00:32:50,961 --> 00:32:54,655
that's the magic of, these,
environments that we are

548
00:32:54,655 --> 00:32:58,705
building here. And you can
totally, so you can basically,

549
00:32:58,705 --> 00:33:03,182
if you make that reference to,
to NPM, then the Pixi TOML would

550
00:33:03,182 --> 00:33:07,658
be a package, json essentially.
And then the package. json, you

551
00:33:07,658 --> 00:33:12,063
can also define some tasks like,
and then we also have the log

552
00:33:12,063 --> 00:33:16,184
files. So that's one of the
other key features that I also

553
00:33:16,184 --> 00:33:20,092
mentioned already. So you have
this whole list of exact

554
00:33:20,092 --> 00:33:24,000
versions of packages that you're
using. It's very long.

555
00:33:24,000 --> 00:33:27,133
Audrow Nash: And it's not really
meant to be read by humans, but

556
00:33:27,133 --> 00:33:30,119
it's something you can do to get
the exact version of all the

557
00:33:30,119 --> 00:33:31,000
packages you need.

558
00:33:31,000 --> 00:33:34,629
Wolf Vollprecht: And while the
whole lock file may be not be

559
00:33:34,629 --> 00:33:38,560
read by a human, but the diff is
something that's pretty nice, to

560
00:33:38,560 --> 00:33:42,492
have, like someone makes a PR or
changes the version of something

561
00:33:42,492 --> 00:33:46,000
and you can see exactly what
changed and you can also roll

562
00:33:46,000 --> 00:33:48,000
Audrow Nash: You're like, why
are you changing this? Yeah,

563
00:33:48,000 --> 00:33:51,818
Wolf Vollprecht: Yeah. And, like
the ability to be able to roll

564
00:33:51,818 --> 00:33:55,636
back is something that you would
maybe otherwise only have with

565
00:33:55,636 --> 00:33:58,000
docker containers. but we also
give you

566
00:33:58,000 --> 00:34:01,876
Audrow Nash: is a giant pain
though, to do pinning all your

567
00:34:01,876 --> 00:34:06,080
versions in a Docker container,
in your docker file. Oh, it's so

568
00:34:06,080 --> 00:34:07,000
painful. Okay,

569
00:34:07,000 --> 00:34:08,000
Wolf Vollprecht: So yeah,

570
00:34:07,000 --> 00:34:11,326
Audrow Nash: love this, I love
all the different approaches or

571
00:34:11,326 --> 00:34:15,581
I, and then one thing to make
very clear, so when you do your

572
00:34:15,581 --> 00:34:20,047
tasks dot start online 10, what
you can do, I assume, is you can

573
00:34:20,047 --> 00:34:23,744
write Pixi task start or
something to that effect and

574
00:34:23,744 --> 00:34:25,000
it'll use that run

575
00:34:25,000 --> 00:34:27,000
Wolf Vollprecht: run starts.
Yeah.

576
00:34:27,000 --> 00:34:30,232
Audrow Nash: And so then what
that does is that calls that

577
00:34:30,232 --> 00:34:33,520
command and you're setting your
environment variables to be

578
00:34:33,520 --> 00:34:36,362
these things that you're
defining there. And that's

579
00:34:36,362 --> 00:34:39,706
really nice because a lot of
times you'll have very long ROS

580
00:34:39,706 --> 00:34:43,272
2 commands. especially if you're
doing like specific things with

581
00:34:43,272 --> 00:34:46,672
specific packages and I don't
know, these kinds of things. So

582
00:34:46,672 --> 00:34:50,294
this is a very nice way of doing
it and we can often do this kind

583
00:34:50,294 --> 00:34:53,693
of thing in a make file, but
your representation is very nice

584
00:34:53,693 --> 00:34:56,981
and clean and make files. You
have to do your dot phony and

585
00:34:56,981 --> 00:35:00,381
stuff at the top and it gets
weird and it's not totally clear

586
00:35:00,381 --> 00:35:03,000
to new developers. So I like
this a lot better.

587
00:35:03,000 --> 00:35:06,389
Wolf Vollprecht: maybe another
motivation is to, you know how

588
00:35:06,389 --> 00:35:09,722
you often have readmes that's
say pseudo apt gets installed,

589
00:35:09,722 --> 00:35:13,056
yada, yada, and then run this
and then run compile and cmake

590
00:35:13,056 --> 00:35:16,111
and this. And like you can
encapsulate all of this into

591
00:35:16,111 --> 00:35:19,444
this one file. It installs all
your dependencies and it runs

592
00:35:19,444 --> 00:35:22,500
all your tasks. And so
everything, the user, everything

593
00:35:22,500 --> 00:35:26,000
you need to put in your README
is just pixi run start, ideally.

594
00:35:26,000 --> 00:35:29,000
Audrow Nash: Love it. Yeah,
that's super, super

595
00:35:29,000 --> 00:35:32,919
Wolf Vollprecht: this, we
haven't really dove deeply into

596
00:35:32,919 --> 00:35:37,113
these task system, but you can
also have tasks that depend on

597
00:35:37,113 --> 00:35:41,100
other tasks and you can also
have some caching so that you

598
00:35:41,100 --> 00:35:44,813
can skip over, the task if it
has already executed and

599
00:35:44,813 --> 00:35:49,144
fulfills the conditions. So the
whole system is pretty powerful

600
00:35:49,144 --> 00:35:51,000
if you, really get into it.

601
00:35:51,000 --> 00:35:54,342
Audrow Nash: I bet hell yeah.
And this looks very cargo like,

602
00:35:54,342 --> 00:35:57,685
which is, it's really nice to
see you took a very good design

603
00:35:57,685 --> 00:36:00,753
and just applied it more
broadly. So I really, I like, I

604
00:36:00,753 --> 00:36:04,205
really like the design decisions
of cargo. It's a very nice way

605
00:36:04,205 --> 00:36:07,767
to develop and I really like how
you guys are doing this, so hell

606
00:36:07,767 --> 00:36:11,000
yeah. anything else to add? I
think this is the first demo.

607
00:36:11,000 --> 00:36:16,015
Wolf Vollprecht: Yeah. That was
the first demo. And I think we

608
00:36:16,015 --> 00:36:20,221
can go over to Ruben to give
more of a ROS developer

609
00:36:20,221 --> 00:36:22,000
experience, like dive.

610
00:36:22,000 --> 00:36:25,000
Ruben Arts: Yeah, of course.

611
00:36:25,000 --> 00:36:26,000
Audrow Nash: All right.

612
00:36:26,000 --> 00:36:29,987
Ruben Arts: If there's no more
questions on this part, I can go

613
00:36:29,987 --> 00:36:31,000
a bit deeper, so

614
00:36:31,000 --> 00:36:33,000
Audrow Nash: Yeah, definitely.

615
00:36:33,000 --> 00:36:36,738
Ruben Arts: say A few First
things, I'm gonna share my

616
00:36:36,738 --> 00:36:41,031
screen, which is only gonna be a
VS. Code instance. I hope you

617
00:36:41,031 --> 00:36:42,000
can see it. I,

618
00:36:42,000 --> 00:36:43,000
Wolf Vollprecht: We do,

619
00:36:42,000 --> 00:36:45,333
Audrow Nash: So one, one note
with this is, we were having

620
00:36:45,333 --> 00:36:48,897
trouble sharing the full screen
Ruben. And so what we're doing

621
00:36:48,897 --> 00:36:52,460
is we're gonna share the window
and it's gonna be a little bit

622
00:36:52,460 --> 00:36:55,908
clunky, but we shouldn't, there
shouldn't, you'll be able to

623
00:36:55,908 --> 00:36:57,000
share a turtle sim,

624
00:36:57,000 --> 00:36:58,000
Ruben Arts: Exactly.

625
00:36:58,000 --> 00:36:59,000
Audrow Nash: when it's up

626
00:36:59,000 --> 00:37:03,089
Ruben Arts: Yeah. So what I want
to do is, share the experience

627
00:37:03,089 --> 00:37:07,114
of a world workspace. you guys
listening to this will probably

628
00:37:07,114 --> 00:37:11,073
all know what they do on a daily
base, but we are gonna go in

629
00:37:11,073 --> 00:37:15,162
depth on how you could use pixi
and some other tools to improve

630
00:37:15,162 --> 00:37:18,992
that workflow. And I'm gonna do
it on Linux because my main

631
00:37:18,992 --> 00:37:22,822
machine is Linux, but this
should work on also Mac Os, and,

632
00:37:22,822 --> 00:37:26,781
windows. I'm just gonna do it
because I prepared the demo and

633
00:37:26,781 --> 00:37:30,611
that's why Wolf's not running it
on Mac. and I'm gonna go a

634
00:37:30,611 --> 00:37:31,000
little

635
00:37:31,000 --> 00:37:34,351
Audrow Nash: some weird Linux,
right? what? What were you, it's

636
00:37:34,351 --> 00:37:36,000
Debian based, but what were you

637
00:37:36,000 --> 00:37:39,784
Ruben Arts: I'm on Pope OS like
it because of the Nvidia

638
00:37:39,784 --> 00:37:41,000
support, but it's,

639
00:37:41,000 --> 00:37:42,000
Audrow Nash: Ah,

640
00:37:42,000 --> 00:37:45,248
Ruben Arts: it's an improved
version of Ubuntu if you would

641
00:37:45,248 --> 00:37:48,000
ask me. But let's not get into
that, conversation.

642
00:37:48,000 --> 00:37:51,000
Audrow Nash: Ah, okay.

643
00:37:51,000 --> 00:37:54,997
Ruben Arts: so I started with
this, folder, which, I already

644
00:37:54,997 --> 00:37:58,727
prepared, but it's an empty
folder and I'm gonna start a

645
00:37:58,727 --> 00:38:02,790
picture project in there and I'm
just gonna copy in some, oh,

646
00:38:02,790 --> 00:38:06,787
this is already going wrong, but
I'm gonna copy in, command,

647
00:38:06,787 --> 00:38:11,050
that's called Pixi in it, which
Initializes project, which gives

648
00:38:11,050 --> 00:38:15,247
you these three files, just some
gi information and the Pixi to

649
00:38:15,247 --> 00:38:19,244
itself. I will, it'll add the
name and the platform that I'm

650
00:38:19,244 --> 00:38:23,507
running on to add a version and
an empty tasks and dependencies,

651
00:38:23,507 --> 00:38:27,237
list what Wolf just showed. And
the first step you would

652
00:38:27,237 --> 00:38:31,234
normally do is start with adding
some more dependencies. You

653
00:38:31,234 --> 00:38:35,031
could add them here yourself,
but you could also add them

654
00:38:35,031 --> 00:38:38,761
through the CLI as well. I'm
constantly doing the Ctrl C

655
00:38:38,761 --> 00:38:42,825
without the Shift Classic. Which
is gonna add a list of, some

656
00:38:42,825 --> 00:38:46,888
ROS, dependencies. It's gonna,
add some compiler information.

657
00:38:46,888 --> 00:38:51,084
And that's basically an initial
setup of a source, workspace in

658
00:38:51,084 --> 00:38:55,081
ROS. What you normally do with
APT get I just did with Pixi.

659
00:38:55,081 --> 00:38:59,278
And it shows you what versions
it added. It adds it to the fall

660
00:38:59,278 --> 00:39:03,408
for you. we really like to have
upper and lower bounds to make

661
00:39:03,408 --> 00:39:07,138
sure that your package isn't
suddenly gonna break if you

662
00:39:07,138 --> 00:39:11,002
update. but you can yeah, define
any way of, of defining a

663
00:39:11,002 --> 00:39:13,000
package in there. I also added

664
00:39:13,000 --> 00:39:15,000
Audrow Nash: you know if you
wanna, so

665
00:39:15,000 --> 00:39:16,000
Ruben Arts: go ahead.

666
00:39:16,000 --> 00:39:19,459
Audrow Nash: I see that, you
have you have ones where it's

667
00:39:19,459 --> 00:39:23,156
saying don't go into the next
major version. Version, and then

668
00:39:23,156 --> 00:39:26,853
you have other ones where it's
saying, don't go into the next,

669
00:39:26,853 --> 00:39:29,000
I. Minor version. how does it
decide

670
00:39:29,000 --> 00:39:30,000
Ruben Arts: Yeah, that's a good
question.

671
00:39:30,000 --> 00:39:33,586
Audrow Nash: So like your
package config is 29.2 and

672
00:39:33,586 --> 00:39:37,862
anything less than 30 and your
for minor version. And then you

673
00:39:37,862 --> 00:39:42,138
have other ones that I see the
major version like, ROS, humble

674
00:39:42,138 --> 00:39:46,138
Turtle Sim 1.4, and anything
less than two should be fine.

675
00:39:46,138 --> 00:39:50,000
How, does it decide that? 'cause
that seems smart to me.

676
00:39:50,000 --> 00:39:54,165
Ruben Arts: Yeah, so good that
you notice that it's actually

677
00:39:54,165 --> 00:39:58,677
relatively simple. We, adhere to
the sim fair, way of versioning,

678
00:39:58,677 --> 00:40:03,120
which means, if you break your,
application or API, you bump the

679
00:40:03,120 --> 00:40:07,146
first, number, which is the
major one. this counts for all

680
00:40:07,146 --> 00:40:11,242
the numbers except the zero
versions, which means these are

681
00:40:11,242 --> 00:40:15,615
in development mode, which means
that if you are in development

682
00:40:15,615 --> 00:40:19,642
mode, you don't have a major
version yet, so the minor can

683
00:40:19,642 --> 00:40:23,598
also break. And that's the
reason we decide if you have a

684
00:40:23,598 --> 00:40:27,000
minor, you only depend on the
minor specifically.

685
00:40:27,000 --> 00:40:30,150
Audrow Nash: Hell yeah. That's a
smart heuristic in my opinion.

686
00:40:30,150 --> 00:40:31,000
That seems great.

687
00:40:31,000 --> 00:40:35,361
Ruben Arts: Yeah. So that
basically gives you a start and.

688
00:40:35,361 --> 00:40:39,346
I am going to run Turtle Sim
already, because it just

689
00:40:39,346 --> 00:40:44,158
installed Turtle Sim for me. Let
me do this and you will not see

690
00:40:44,158 --> 00:40:48,895
the Turtle Sim running, but for
this example, I will, I'm going

691
00:40:48,895 --> 00:40:53,481
to make sure that you trust me
on that the little turtle sim,

692
00:40:53,481 --> 00:40:57,767
window opened and we already
have a full ROS installation

693
00:40:57,767 --> 00:41:01,000
here. you might see this as
familiar, logs,

694
00:41:01,000 --> 00:41:03,000
Audrow Nash: Yeah,

695
00:41:03,000 --> 00:41:06,144
Ruben Arts: because I can share
that right now. I'll continue

696
00:41:06,144 --> 00:41:08,000
with making an actual package,
so I,

697
00:41:08,000 --> 00:41:12,350
Audrow Nash: And so the one
thing, so you did Pixi run and

698
00:41:12,350 --> 00:41:17,225
Pixi run was how we called tasks
in our, in our. Pixi instance or

699
00:41:17,225 --> 00:41:21,425
Pixi workspace and, but then
what you're doing is you're

700
00:41:21,425 --> 00:41:25,925
giving it an arbitrary ROS 2
command, which is the ROS 2 run

701
00:41:25,925 --> 00:41:30,725
turtle sim, which is the package
and then turtle sim node, which

702
00:41:30,725 --> 00:41:32,000
is the executable

703
00:41:32,000 --> 00:41:34,000
Ruben Arts: Yeah, exactly. So
Pixi has,

704
00:41:34,000 --> 00:41:37,013
Audrow Nash: that's cool that
it's scopes your command for

705
00:41:37,013 --> 00:41:38,000
this kind of thing.

706
00:41:38,000 --> 00:41:42,113
Ruben Arts: Pixi has multiple
commands that can run, run your

707
00:41:42,113 --> 00:41:46,092
environment. So because the
environment is in, this folder,

708
00:41:46,092 --> 00:41:50,475
we need to activate it before we
can run anything in it. And Pixi

709
00:41:50,475 --> 00:41:53,576
Run will activate your
environment and run any

710
00:41:53,576 --> 00:41:57,555
arbitrary task in it. it'll just
spawn it in its own little

711
00:41:57,555 --> 00:42:01,128
shell, which is also
cross-platform. So all the tasks

712
00:42:01,128 --> 00:42:04,972
you define, in Pixi, are
cross-platform. if Wolf runs the

713
00:42:04,972 --> 00:42:09,355
same command, it should work the
same way, as it did for me. But,

714
00:42:09,355 --> 00:42:13,536
there's also another thing that
you can do, which is more used

715
00:42:13,536 --> 00:42:17,581
if you're developing hard on a
project, this Pixi shell. And

716
00:42:17,581 --> 00:42:21,762
that starts an actual shell, a
new shell, which now has ROS to

717
00:42:21,762 --> 00:42:25,943
available in, in that specific
shell. And, I don't have ROS to

718
00:42:25,943 --> 00:42:30,191
in the system apart from that.
So if I exit that again, I don't

719
00:42:30,191 --> 00:42:31,000
have ROS to,

720
00:42:31,000 --> 00:42:32,000
Audrow Nash: That's really

721
00:42:32,000 --> 00:42:35,537
Ruben Arts: I do, oh, sorry. I
have a VS. Code plugin that

722
00:42:35,537 --> 00:42:39,500
actually activates it in a shell
for me already. So if I would do

723
00:42:39,500 --> 00:42:42,000
another terminal that wasn't in
VS. Code.

724
00:42:42,000 --> 00:42:43,000
Audrow Nash: I think it's still
true.

725
00:42:43,000 --> 00:42:44,000
Ruben Arts: Yeah,

726
00:42:44,000 --> 00:42:47,949
Audrow Nash: It's fine. So one
thing with this, when you were

727
00:42:47,949 --> 00:42:51,768
in the shell, so when you did
Pixi shell, it looks like you

728
00:42:51,768 --> 00:42:55,587
get auto complete. Do you get
auto complete when you do the

729
00:42:55,587 --> 00:42:59,277
Pixi run command? I think Wolf
and I were looking at this

730
00:42:59,277 --> 00:43:03,290
before and I think that because
he was using the ZSH shell, it

731
00:43:03,290 --> 00:43:07,109
was difficult to set up. and he
didn't do it before we were

732
00:43:07,109 --> 00:43:11,252
talking last time, but, I don't
know what shell you're using. It

733
00:43:11,252 --> 00:43:13,000
looks like it could be ZSH,

734
00:43:13,000 --> 00:43:15,000
Ruben Arts: I am.

735
00:43:15,000 --> 00:43:17,667
Audrow Nash: do you get auto
complete with your Pixi run

736
00:43:17,667 --> 00:43:18,000
command

737
00:43:18,000 --> 00:43:19,000
Ruben Arts: Yeah. So if

738
00:43:19,000 --> 00:43:21,000
Audrow Nash: you do Pixi run
ROS? Two

739
00:43:21,000 --> 00:43:25,789
Ruben Arts: Oh, so that, that's
a, there's two different things

740
00:43:25,789 --> 00:43:30,123
here. So Pixi run, ROS 2 will
currently not auto complete

741
00:43:30,123 --> 00:43:34,152
because to turn on auto
completion, you would have to

742
00:43:34,152 --> 00:43:38,789
forward that to an activated
environment, which is, too slow.

743
00:43:38,789 --> 00:43:43,199
So we cannot activate the
environment while you're tapping

744
00:43:43,199 --> 00:43:47,684
away on your command line. but
what you can do if you have,

745
00:43:47,684 --> 00:43:52,550
let's quickly add a simple task.
So Pixi task at, let's call it,

746
00:43:52,550 --> 00:43:57,339
hello? and we just echo, hello.
So this now added a task called

747
00:43:57,339 --> 00:44:00,000
Hello, which will just echo.
Hello?

748
00:44:00,000 --> 00:44:01,875
Audrow Nash: Oh, and you will
get auto complete with this. I

749
00:44:01,875 --> 00:44:02,000
see.

750
00:44:02,000 --> 00:44:04,940
Ruben Arts: Yeah. So now I can
do h and it will autocomplete.

751
00:44:04,940 --> 00:44:06,000
Hello for me. So then,

752
00:44:06,000 --> 00:44:09,701
Audrow Nash: it's in your task.
I see. that's nice that you do

753
00:44:09,701 --> 00:44:10,000
that.

754
00:44:10,000 --> 00:44:13,554
Ruben Arts: So yeah, so that's
what, what does autocomplete for

755
00:44:13,554 --> 00:44:17,051
me? and auto completion. Turning
it on is just one script that

756
00:44:17,051 --> 00:44:20,267
you have to add to your bashrc
or zshrc or whatever shell

757
00:44:20,267 --> 00:44:21,000
you're using.

758
00:44:21,000 --> 00:44:25,000
Audrow Nash: it is. Yeah. Okay,

759
00:44:25,000 --> 00:44:28,930
Ruben Arts: so yeah, just fire
more questions, but I'm gonna

760
00:44:28,930 --> 00:44:32,859
continue with actually building
a project. So I will do this

761
00:44:32,859 --> 00:44:36,658
command, which is a ROS 2
package, create command that you

762
00:44:36,658 --> 00:44:40,457
might know. We're gonna do a
Python version of it, and the

763
00:44:40,457 --> 00:44:44,518
destination is gonna be in an
src folder, and we're gonna name

764
00:44:44,518 --> 00:44:48,447
it talker for those listening
in. And this is the basic roll

765
00:44:48,447 --> 00:44:52,377
two creation of a package. So we
have a, yeah, the, just the

766
00:44:52,377 --> 00:44:56,307
normal workflow of, of a ROS
packages includes the package m

767
00:44:56,307 --> 00:45:00,367
and the Python package. So also
the setup of the pie, which is

768
00:45:00,367 --> 00:45:03,970
simple enough. Let's, let's see
if we can build it. I'm

769
00:45:03,970 --> 00:45:06,000
constantly doing the control C.

770
00:45:06,000 --> 00:45:08,000
Audrow Nash: Oh, whatever. It
doesn't matter.

771
00:45:08,000 --> 00:45:12,397
Ruben Arts: So we are using
Colcon. We previously installed

772
00:45:12,397 --> 00:45:17,017
it. so that already works and
now we should be able to, should

773
00:45:17,017 --> 00:45:21,414
be able to run it. There's one
caveat in the ROS world. You

774
00:45:21,414 --> 00:45:26,109
need to add this little, setup
dos h to your bashrc or zshrc to

775
00:45:26,109 --> 00:45:30,282
make sure that, that the
environment knows where to find

776
00:45:30,282 --> 00:45:34,828
your, ROS project. So I'm gonna
quickly run this echo command

777
00:45:34,828 --> 00:45:39,449
into my, pixi toml with which
adds an activation script and it

778
00:45:39,449 --> 00:45:43,846
make sure when the, when Pixi
activates an environment that

779
00:45:43,846 --> 00:45:48,466
also runs this script, which is
now, in the install because of

780
00:45:48,466 --> 00:45:51,000
colcon. And if I then do Pixi
run,

781
00:45:51,000 --> 00:45:53,431
Audrow Nash: I like that you did
that from the command line as

782
00:45:53,431 --> 00:45:55,000
opposed to just putting it in
your file.

783
00:45:55,000 --> 00:45:59,000
Ruben Arts: yeah, I prepared it
to be a command line example.

784
00:45:59,000 --> 00:46:03,286
Audrow Nash: love it. So with
activation, is that a special,

785
00:46:03,286 --> 00:46:07,714
that's a special thing that it
knows as like a, like, that's a

786
00:46:07,714 --> 00:46:12,000
key word in your, Pixi tomal
file that says, run this so you

787
00:46:12,000 --> 00:46:16,357
can set up your environment when
you're in the Correct, when,

788
00:46:16,357 --> 00:46:21,000
you're doing Pixi run and stuff,
it, does this all first. Is that

789
00:46:21,000 --> 00:46:24,581
Ruben Arts: yeah. No, that's
exactly, the way it works. you

790
00:46:24,581 --> 00:46:28,344
can also define an end. Oh, it's
actually autocomplete. That's

791
00:46:28,344 --> 00:46:31,803
great. which you can define, for
instance, Python Path or

792
00:46:31,803 --> 00:46:35,202
whatever, which is just an
environment variable. But the

793
00:46:35,202 --> 00:46:38,662
scripts is for, ROS very
important because you, it's more

794
00:46:38,662 --> 00:46:42,000
than just an environment
variable that you have to set.

795
00:46:42,000 --> 00:46:43,000
Audrow Nash: Mm-Hmm

796
00:46:43,000 --> 00:46:46,543
Ruben Arts: we just built, the
package and we now can already,

797
00:46:46,543 --> 00:46:49,000
run the talker, which is hello
from talker.

798
00:46:49,000 --> 00:46:52,422
Ruben Arts: This is the basic
code from, the ROS examples.

799
00:46:52,422 --> 00:46:56,022
but, so I'm not sure how long
this took me, but it's probably

800
00:46:56,022 --> 00:46:59,621
didn't take me much longer than
five minutes while explaining

801
00:46:59,621 --> 00:47:03,043
it. So this is a good show of
how quick you can get set up

802
00:47:03,043 --> 00:47:06,702
with the Pixi. I do have to say
that this is of course, with a

803
00:47:06,702 --> 00:47:08,000
cached download speed.

804
00:47:08,000 --> 00:47:12,013
Ruben Arts: So I have done this
before, which means Pixi caches

805
00:47:12,013 --> 00:47:15,771
all the packages that I've
installed, and so it only has to

806
00:47:15,771 --> 00:47:19,592
verify if there's new stuff
available and otherwise it pulls

807
00:47:19,592 --> 00:47:23,414
it from my cache. So my machine
has probably, a few hundred,

808
00:47:23,414 --> 00:47:26,662
installations, different
environments, but it's all

809
00:47:26,662 --> 00:47:30,675
pulled from the cache and hard
linked into your environment. So

810
00:47:30,675 --> 00:47:34,561
this folder might look big, but
it's all hard linked files to

811
00:47:34,561 --> 00:47:38,000
the cache folder, which makes
this much easier to use.

812
00:47:38,000 --> 00:47:41,706
Audrow Nash: So you're, this is
a bit like yarn or something, if

813
00:47:41,706 --> 00:47:45,007
I understand, which is, an
alternative to NPM. So you are

814
00:47:45,007 --> 00:47:48,423
in, like your home directory,
you're caching your packages,

815
00:47:48,423 --> 00:47:51,955
and then when you have a project
that uses those, it goes and

816
00:47:51,955 --> 00:47:55,372
looks for them in your home
directory and downloads them if

817
00:47:55,372 --> 00:47:58,904
you don't have them. But if you
do use them, you just link to

818
00:47:58,904 --> 00:48:02,552
them and it puts that into your.
Your workspace effectively, is

819
00:48:02,552 --> 00:48:04,000
that exactly what's going

820
00:48:04,000 --> 00:48:05,000
Ruben Arts: Yeah, a hundred
percent.

821
00:48:05,000 --> 00:48:06,000
Audrow Nash: Very clever.

822
00:48:06,000 --> 00:48:09,338
Ruben Arts: Yeah. And depending
on your machine, it de it

823
00:48:09,338 --> 00:48:13,144
decides which version of linking
is the fastest for you. there's,

824
00:48:13,144 --> 00:48:16,599
I don't, I'm not gonna go in
depth, but there, there's some

825
00:48:16,599 --> 00:48:19,000
smart things going on there.
Yeah. Sorry.

826
00:48:19,000 --> 00:48:22,234
Audrow Nash: Yeah, I don't know
anything about the different

827
00:48:22,234 --> 00:48:25,305
types of linking. I just like
symbolic links. I got that,

828
00:48:25,305 --> 00:48:28,000
exists, but I don't know if
there's more variants,

829
00:48:28,000 --> 00:48:29,000
Wolf Vollprecht: now,

830
00:48:28,000 --> 00:48:29,000
Audrow Nash: don't have to talk

831
00:48:29,000 --> 00:48:31,977
Wolf Vollprecht: now that we're
talking about it, so there's one

832
00:48:31,977 --> 00:48:34,767
cool link that's called copy on
write, which is nice because

833
00:48:34,767 --> 00:48:35,000
then.

834
00:48:35,000 --> 00:48:36,000
Audrow Nash: Oh,

835
00:48:36,000 --> 00:48:39,736
Wolf Vollprecht: you don't like,
you link it, but when you modify

836
00:48:39,736 --> 00:48:41,000
it, it becomes a copy.

837
00:48:41,000 --> 00:48:44,950
Ruben Arts: Yeah, so I could
modify this environment and then

838
00:48:44,950 --> 00:48:48,446
I don't break all the other
environments that are also

839
00:48:48,446 --> 00:48:50,000
linked to the same file,

840
00:48:50,000 --> 00:48:51,000
Wolf Vollprecht: yeah.

841
00:48:51,000 --> 00:48:52,000
Ruben Arts: is sometimes a
problem.

842
00:48:52,000 --> 00:48:55,908
Audrow Nash: Oh, that's clever.
Okay. What an interesting thing.

843
00:48:55,908 --> 00:48:59,695
Then you may go outta sync, but
that's a very neat feature for

844
00:48:59,695 --> 00:49:00,000
this.

845
00:49:00,000 --> 00:49:05,087
Audrow Nash: Okay, so now you've
activated, so now you've run, so

846
00:49:05,087 --> 00:49:09,000
we're running a ROS 2, node on
this. That's great.

847
00:49:09,000 --> 00:49:13,820
Ruben Arts: This just this, this
simple code from the example,

848
00:49:13,820 --> 00:49:18,329
that it's created. I'm gonna add
another thing. the c plus

849
00:49:18,329 --> 00:49:23,071
package. So this is doing the
same thing again, just creating

850
00:49:23,071 --> 00:49:27,890
a package now based on CMake. it
has some dependencies and the

851
00:49:27,890 --> 00:49:32,788
note name is gonna be navigator.
I will get into it that later,

852
00:49:32,788 --> 00:49:37,763
why that is. And now it created
another package called Navigator

853
00:49:37,763 --> 00:49:42,428
with some source. And also here
is just a simple hello world

854
00:49:42,428 --> 00:49:47,403
thing. what we need to do now is
build this package, with a cool

855
00:49:47,403 --> 00:49:52,300
command. I'm gonna run it really
quickly. Shouldn't be too hard

856
00:49:52,300 --> 00:49:53,000
to build.

857
00:49:53,000 --> 00:49:54,000
Audrow Nash: the ninja?

858
00:49:54,000 --> 00:49:57,545
Ruben Arts: Oh, that's a
different, I have to say this

859
00:49:57,545 --> 00:50:01,419
correctly. Different tool to
configure CMake. Believe I say

860
00:50:01,419 --> 00:50:05,424
that correctly. Otherwise, Wolf,
please, please tell me how I

861
00:50:05,424 --> 00:50:07,000
should say it otherwise,

862
00:50:07,000 --> 00:50:11,247
Wolf Vollprecht: cmake generates
ninja file or cmake generates

863
00:50:11,247 --> 00:50:15,493
the make file, right? so it's an
alternative to make. That was

864
00:50:15,493 --> 00:50:17,000
written at some point.

865
00:50:17,000 --> 00:50:21,000
Ruben Arts: yeah, I just like to
use that one better.

866
00:50:21,000 --> 00:50:23,000
Audrow Nash: Huh? Why?

867
00:50:23,000 --> 00:50:27,326
Ruben Arts: because it seems to
be faster and tests some more.

868
00:50:27,326 --> 00:50:29,000
Yeah. How do we get into

869
00:50:29,000 --> 00:50:32,012
Wolf Vollprecht: line output is
a bit nicer. I would say like

870
00:50:32,012 --> 00:50:33,000
the console logs are

871
00:50:33,000 --> 00:50:34,000
Ruben Arts: that's definitely
already my

872
00:50:34,000 --> 00:50:37,303
Audrow Nash: are hard to. I
understand. Okay. Hell yeah. Oh,

873
00:50:37,303 --> 00:50:40,000
the command line output is nice.
Okay. Hell yeah.

874
00:50:40,000 --> 00:50:43,949
Ruben Arts: so I will Oh, yeah.
So I just built that one, that

875
00:50:43,949 --> 00:50:47,643
was super plus instead of
Python, and I should now be able

876
00:50:47,643 --> 00:50:50,000
to also run that package, so we
just,

877
00:50:50,000 --> 00:50:52,727
Audrow Nash: Oh, and I saw the
activating. That was cool. So

878
00:50:52,727 --> 00:50:55,091
that was, sourcing your
environment for this kind of

879
00:50:55,091 --> 00:50:56,000
thing. That's great.

880
00:50:56,000 --> 00:50:58,887
Ruben Arts: so it was like
starting a new shell, running

881
00:50:58,887 --> 00:51:01,000
the command and then getting
back to you.

882
00:51:01,000 --> 00:51:04,994
Audrow Nash: Now, one, one
question with this is, I think

883
00:51:04,994 --> 00:51:09,199
sometimes when you build, you
don't want to have your script

884
00:51:09,199 --> 00:51:13,473
sourced, or you don't wanna have
your workspace sourced. some

885
00:51:13,473 --> 00:51:17,328
people say that this reduces
different classes of bugs,

886
00:51:17,328 --> 00:51:21,462
especially if you're doing
symbolic linking and these kinds

887
00:51:21,462 --> 00:51:25,456
of things. is there a way of
making activation only occur

888
00:51:25,456 --> 00:51:29,731
when I'm running a certain. Type
of commands or have you guys

889
00:51:29,731 --> 00:51:34,006
handled this kind of thing or
you don't see these bugs 'cause

890
00:51:34,006 --> 00:51:38,000
there's no reason to symbolic
link or what? any thoughts?

891
00:51:38,000 --> 00:51:43,407
Ruben Arts: Yeah, so I, I might
get, a bit quick on the features

892
00:51:43,407 --> 00:51:48,645
of, Pixi. But Pixi has a thing
called a feature, which is like

893
00:51:48,645 --> 00:51:53,460
how you could think of it as an
optional dependency or an

894
00:51:53,460 --> 00:51:58,614
optional thing, in, in Pixi. So
I could do feature dot, let's

895
00:51:58,614 --> 00:52:03,514
call it act of activation. And
this now lives in a feature

896
00:52:03,514 --> 00:52:08,583
which is not used, but if I
would want this feature, I could

897
00:52:08,583 --> 00:52:12,976
build an environment that
specifically only uses the

898
00:52:12,976 --> 00:52:18,129
activated, stuff. And then I
could do act what GitHub copilot

899
00:52:18,129 --> 00:52:23,452
is helping me with. And now if I
run the activated environment,

900
00:52:23,452 --> 00:52:27,000
only then is my ROS environment
activated?

901
00:52:27,000 --> 00:52:30,000
Audrow Nash: How do you run the
activated environment?

902
00:52:30,000 --> 00:52:34,909
Ruben Arts: so if I now do Pixi
run the e act, I can do ROS.

903
00:52:34,909 --> 00:52:39,000
Yeah. So the rest of the
examples would run there.

904
00:52:39,000 --> 00:52:41,766
Audrow Nash: okay. I love it.
That's great. I love that you

905
00:52:41,766 --> 00:52:44,156
guys have a very simple
workaround for this. That's

906
00:52:44,156 --> 00:52:45,000
super, super cool.

907
00:52:45,000 --> 00:52:47,990
Ruben Arts: Yeah. I'm gonna
remove it for now because I do

908
00:52:47,990 --> 00:52:50,000
need the activation in my
example, but.

909
00:52:50,000 --> 00:52:53,161
Audrow Nash: Yeah, and rather
than doing the Dashie Act

910
00:52:53,161 --> 00:52:55,000
everywhere, so that makes sense.

911
00:52:55,000 --> 00:52:58,982
Ruben Arts: so what I just
showed you was two different

912
00:52:58,982 --> 00:53:03,327
types of packages, a C package,
C plus package, and a Python

913
00:53:03,327 --> 00:53:07,888
package and a ROS workspace. but
you already saw me with pretty

914
00:53:07,888 --> 00:53:12,378
long, commands that I need to
run. you've already seen it. But

915
00:53:12,378 --> 00:53:17,084
I am gonna add the tasks through
the CLI. So I'm gonna add a task

916
00:53:17,084 --> 00:53:21,646
for just running the sim and I'm
gonna add three more tasks for

917
00:53:21,646 --> 00:53:26,207
build and for actually running
the nodes that you've seen me do

918
00:53:26,207 --> 00:53:30,479
already. So now I have, can I
make this, should I make this

919
00:53:30,479 --> 00:53:32,000
smaller or bigger or,

920
00:53:32,000 --> 00:53:33,000
Audrow Nash: I like it how it

921
00:53:33,000 --> 00:53:34,000
Ruben Arts: cool. Then I have,

922
00:53:34,000 --> 00:53:37,000
Audrow Nash: it's a little
cumbersome, but still I.

923
00:53:37,000 --> 00:53:41,574
Ruben Arts: That's good. The
tasks are now added. So now Pixi

924
00:53:41,574 --> 00:53:45,923
Run has these five tasks,
including the hello one, which I

925
00:53:45,923 --> 00:53:49,972
can run. So I can run the
simulation of the turtle sim

926
00:53:49,972 --> 00:53:54,096
node. I can build this,
workspace. I can run the talker

927
00:53:54,096 --> 00:53:58,745
navigator separately. And these
now also how to complete on my

928
00:53:58,745 --> 00:54:02,794
command line. So Pixi run,
navigator, just the end tap

929
00:54:02,794 --> 00:54:07,293
completion and the navigator
starts, running. And that makes

930
00:54:07,293 --> 00:54:11,867
it just easier for someone that
coming in and is oh, how does

931
00:54:11,867 --> 00:54:16,217
this p project work? You can
just do Pixi run, tap, and he

932
00:54:16,217 --> 00:54:20,566
sees all the available tasks
that he has, or task list, of

933
00:54:20,566 --> 00:54:25,140
course. And then those will work
as well. I'm gonna copy some

934
00:54:25,140 --> 00:54:29,564
code that I prepared into the
packages. is just to make the

935
00:54:29,564 --> 00:54:33,838
example a little bit more
interesting, which is gonna be.

936
00:54:33,838 --> 00:54:37,812
dock is gonna accept the CLI
command for an x and a y

937
00:54:37,812 --> 00:54:42,611
position for the turtle. And the
navigator is gonna, listen on a

938
00:54:42,611 --> 00:54:46,810
topic which the talker is
talking to, and then interpret

939
00:54:46,810 --> 00:54:51,609
that information and putting it
to the, CMD valve of the turtle.

940
00:54:51,609 --> 00:54:56,033
This is all prepared, so you
don't really have to look into

941
00:54:56,033 --> 00:55:00,832
it, but it's just the most basic
simple, ROS example of a, topic

942
00:55:00,832 --> 00:55:05,556
and sending it to another server
that's listening to that topic

943
00:55:05,556 --> 00:55:09,905
and then sending it forward to
the topic of the turtle sim

944
00:55:09,905 --> 00:55:14,554
that's running. I can now also
still build this package, so if

945
00:55:14,554 --> 00:55:19,278
I do Pixi run built, that will
start building the package. Both

946
00:55:19,278 --> 00:55:23,852
of them actually, 'cause it's in
a workspace of two packages.

947
00:55:23,852 --> 00:55:28,576
It's gonna be a little bit more
time because we have to rebuild

948
00:55:28,576 --> 00:55:33,000
with actual dependencies. Es but
it should just take a. Few

949
00:55:33,000 --> 00:55:36,000
Audrow Nash: This is super cool
though.

950
00:55:36,000 --> 00:55:37,000
Ruben Arts: So now it did

951
00:55:37,000 --> 00:55:38,000
Audrow Nash: I'm very impressed.

952
00:55:38,000 --> 00:55:42,194
Ruben Arts: And I'm thinking now
because I cannot easily share my

953
00:55:42,194 --> 00:55:44,000
screen, what I will do, I'll

954
00:55:44,000 --> 00:55:47,117
Audrow Nash: Yeah. Don't worry
about it. we'll see the, we'll

955
00:55:47,117 --> 00:55:50,336
see the positions, right? Aren't
you echoing the positions with

956
00:55:50,336 --> 00:55:51,000
the talker or

957
00:55:51,000 --> 00:55:52,000
Ruben Arts: I

958
00:55:51,000 --> 00:55:55,028
Wolf Vollprecht: I actually
like, while, Ruben was demoing

959
00:55:55,028 --> 00:55:59,194
it, I, made sure that it works
on, the M1 chip as well, so I

960
00:55:59,194 --> 00:56:01,000
could show it if you want.

961
00:56:01,000 --> 00:56:04,956
Ruben Arts: if you wanna do
that, then we could, definitely,

962
00:56:04,956 --> 00:56:08,451
share it on your screen and,
come back for the actual

963
00:56:08,451 --> 00:56:12,604
building of, package for on my
screen. I can prepare that to go

964
00:56:12,604 --> 00:56:13,000
ahead.

965
00:56:13,000 --> 00:56:17,957
Wolf Vollprecht: that's a good
idea. then let me share my

966
00:56:17,957 --> 00:56:23,000
screen again. same thing as
Ruben had, but what I did was,

967
00:56:23,000 --> 00:56:25,000
Audrow Nash: make it nice and
big too.

968
00:56:25,000 --> 00:56:30,096
Wolf Vollprecht: I just added,
OSX, arm 64 here as a

969
00:56:30,096 --> 00:56:35,388
additionally, supported
platform. And, everything else

970
00:56:35,388 --> 00:56:41,365
is pretty much the same. And,
what room was about to show? So

971
00:56:41,365 --> 00:56:47,539
I ran Pixi Run Build, which is
using Code Con on s to build it.

972
00:56:47,539 --> 00:56:53,125
And now I can do Pixi run sim.
This is going to start the

973
00:56:53,125 --> 00:56:59,396
turtle sim. And then, I'll bring
it to the front in a second. So

974
00:56:59,396 --> 00:57:05,668
we have the navigator, which is
controller and the talker. So I,

975
00:57:05,668 --> 00:57:09,000
let me try to get the window
here.

976
00:57:09,000 --> 00:57:12,316
Audrow Nash: You can pin it. I
think if you can you right click

977
00:57:12,316 --> 00:57:13,000
or something.

978
00:57:13,000 --> 00:57:15,000
Wolf Vollprecht: no, this is not
Linux.

979
00:57:15,000 --> 00:57:16,000
Audrow Nash: or maybe that's,
oh,

980
00:57:16,000 --> 00:57:17,000
Ruben Arts: a Linux thing?

981
00:57:17,000 --> 00:57:19,000
Wolf Vollprecht: make is not so
good.

982
00:57:19,000 --> 00:57:21,000
Audrow Nash: I guess so. Yeah.
That's funny.

983
00:57:21,000 --> 00:57:26,041
Wolf Vollprecht: And, but now I,
we have to talk a note that room

984
00:57:26,041 --> 00:57:30,461
was showing so I can give some
coordinates, I don't know,

985
00:57:30,461 --> 00:57:35,114
three, five. and as you would
expect, it sends a message and

986
00:57:35,114 --> 00:57:40,000
the controller does its job, et
cetera. So let's try some more.

987
00:57:40,000 --> 00:57:44,360
Ruben Arts: Yeah, so the,
example itself is super basic

988
00:57:44,360 --> 00:57:48,640
robotics, which you're gonna
experience your first day

989
00:57:48,640 --> 00:57:53,317
learning ROS, I assume these
days. the great part from this

990
00:57:53,317 --> 00:57:58,232
standpoint is Wolf runs it in
his terminal, which doesn't have

991
00:57:58,232 --> 00:58:03,146
ROS. but if he uses Pixi, this
all gets activated and run just

992
00:58:03,146 --> 00:58:06,000
like a normal ROS experience on
Mac.

993
00:58:06,000 --> 00:58:07,000
Audrow Nash: amazing.

994
00:58:07,000 --> 00:58:08,000
Wolf Vollprecht: and this is on
Mac,

995
00:58:08,000 --> 00:58:09,000
Audrow Nash: 'cause I've,

996
00:58:09,000 --> 00:58:10,000
Wolf Vollprecht: is not,

997
00:58:10,000 --> 00:58:11,000
Audrow Nash: yep.

998
00:58:11,000 --> 00:58:12,000
Ruben Arts: No.

999
00:58:12,000 --> 00:58:13,000
Audrow Nash: Oh, I've done so

1000
00:58:13,000 --> 00:58:14,000
Wolf Vollprecht: it hit the war.

1001
00:58:14,000 --> 00:58:18,259
Audrow Nash: with virtual
machines and all this adding to

1002
00:58:18,259 --> 00:58:22,741
this just. oh, it's so much
easier now. It's so crazy. Okay.

1003
00:58:22,741 --> 00:58:27,000
and then, yeah, then you have
your tele op at the bottom,

1004
00:58:27,000 --> 00:58:28,000
Wolf Vollprecht: yeah. This is,

1005
00:58:28,000 --> 00:58:31,000
Ruben Arts: teop could be run,
but that's another command.

1006
00:58:31,000 --> 00:58:32,000
Wolf Vollprecht: yeah.

1007
00:58:32,000 --> 00:58:33,000
Audrow Nash: Oh,

1008
00:58:32,000 --> 00:58:35,000
Ruben Arts: These are just two
simple, nodes that we run.

1009
00:58:35,000 --> 00:58:38,869
Audrow Nash: Oh, I see. The
navigator's just, publishing it

1010
00:58:38,869 --> 00:58:39,000
to

1011
00:58:39,000 --> 00:58:41,594
Wolf Vollprecht: the talker, is
publishing and then the

1012
00:58:41,594 --> 00:58:44,000
navigator is like a PID or
whatever control thingy.

1013
00:58:44,000 --> 00:58:49,500
Audrow Nash: and then showing
you, yes. Okay. Very, cool.

1014
00:58:49,500 --> 00:58:55,000
Okay. I love it. That's super,
super cool. Thank you for,

1015
00:58:55,000 --> 00:58:56,000
Wolf Vollprecht: back to Ruben,

1016
00:58:56,000 --> 00:58:58,000
Ruben Arts: Yep.

1017
00:58:58,000 --> 00:59:00,000
Audrow Nash: yep.

1018
00:59:00,000 --> 00:59:03,375
Ruben Arts: I can share my
window again. so Wolf just

1019
00:59:03,375 --> 00:59:07,133
showed you. My little example
works, which is great also on

1020
00:59:07,133 --> 00:59:10,508
Mac s next to Linux. but now
you're in a situation. I

1021
00:59:10,508 --> 00:59:14,456
developed this project. I can
run it locally. I can connect to

1022
00:59:14,456 --> 00:59:18,404
another robot. I can do all the
things that the developer does

1023
00:59:18,404 --> 00:59:22,034
in this machine. Now I need to
start deploying it, and my

1024
00:59:22,034 --> 00:59:25,855
colleagues need to also start
using this package. And I want

1025
00:59:25,855 --> 00:59:29,995
to be able to easily fetch it in
my 20 different CI jobs that I'm

1026
00:59:29,995 --> 00:59:33,880
running. you don't want to do
the build everywhere. You don't

1027
00:59:33,880 --> 00:59:37,000
want to, rebuild for every
workspace, every time.

1028
00:59:37,000 --> 00:59:41,276
Ruben Arts: So what we can do is
we can start, building a conduct

1029
00:59:41,276 --> 00:59:45,224
package. And for that we built
our own tool, which is called

1030
00:59:45,224 --> 00:59:48,908
rattler build, which is a
version of Con Build. and this

1031
00:59:48,908 --> 00:59:53,053
command will install it and add
it to the feature build just as

1032
00:59:53,053 --> 00:59:57,000
I showed you with activation
earlier. It's gonna add another

1033
00:59:57,000 --> 01:00:00,000
Audrow Nash: Oh, that's what the
def Dutch F does.

1034
01:00:00,000 --> 01:00:01,000
Ruben Arts: I could write it

1035
01:00:01,000 --> 01:00:03,976
Audrow Nash: very cargo like too
for this kind of thing. But I

1036
01:00:03,976 --> 01:00:06,712
think you're using features a
little bit differently than

1037
01:00:06,712 --> 01:00:07,000
cargo,

1038
01:00:07,000 --> 01:00:08,000
Ruben Arts: A little bit.

1039
01:00:08,000 --> 01:00:09,000
Audrow Nash: in this case.

1040
01:00:09,000 --> 01:00:10,000
Ruben Arts: Yeah, no, that's
true.

1041
01:00:10,000 --> 01:00:13,694
Audrow Nash: features will often
give you different access to,

1042
01:00:13,694 --> 01:00:17,506
you can get the derive module in
something that you want to then

1043
01:00:17,506 --> 01:00:20,962
inherit from in some part of
your code. But here it's like

1044
01:00:20,962 --> 01:00:24,000
you're adding it to your project
in a sense. Is it?

1045
01:00:24,000 --> 01:00:27,135
Ruben Arts: Yeah, no, that's
correct. although the feature

1046
01:00:27,135 --> 01:00:30,541
itself now doesn't do anything,
so it's just in existence in my

1047
01:00:30,541 --> 01:00:32,000
configuration file. But it,

1048
01:00:32,000 --> 01:00:33,000
Audrow Nash: a star.

1049
01:00:33,000 --> 01:00:36,585
Ruben Arts: it just doesn't, it
hasn't downloaded anything right

1050
01:00:36,585 --> 01:00:40,003
now. I did that thing again. I.
So now what I'm gonna do, I'm

1051
01:00:40,003 --> 01:00:43,420
gonna add a project, the project
built, and it has no default

1052
01:00:43,420 --> 01:00:46,894
features in there because, and
what that means is I don't need

1053
01:00:46,894 --> 01:00:50,479
any of the features that I just
defined now. So no dependencies,

1054
01:00:50,479 --> 01:00:54,008
no tasks, all not necessary in
this environment. I just want to

1055
01:00:54,008 --> 01:00:57,258
build environment. So just the
build feature in there. You

1056
01:00:57,258 --> 01:01:00,731
could also have a test feature
or a documentation feature or a

1057
01:01:00,731 --> 01:01:04,036
linting feature. And this way
you can separate the concerns

1058
01:01:04,036 --> 01:01:07,398
through different environments,
which makes it super easy to

1059
01:01:07,398 --> 01:01:10,479
install in ci. So if you just
wanna do formatting, just

1060
01:01:10,479 --> 01:01:13,000
install formatting and you can
split that up.

1061
01:01:13,000 --> 01:01:16,449
Audrow Nash: Oh, that's so cool.
Yeah, because I, it's always so

1062
01:01:16,449 --> 01:01:19,683
silly when you run CI and you
have to set up the whole thing

1063
01:01:19,683 --> 01:01:22,000
just to run clang, tidy, or
whatever it is.

1064
01:01:22,000 --> 01:01:26,479
Ruben Arts: Yeah. Yeah, exactly.
So if I now do Pixi install,

1065
01:01:26,479 --> 01:01:30,739
write it out, it'll start,
updating my log file with a new

1066
01:01:30,739 --> 01:01:35,365
in installed package and it will
start, making that environment

1067
01:01:35,365 --> 01:01:39,698
for me. Now, I'm going to teach
you another thing, which is

1068
01:01:39,698 --> 01:01:44,030
called recipes. This is pretty
conduct specific, and we are

1069
01:01:44,030 --> 01:01:48,510
working on making it easier to
do it through Pixi itself, but

1070
01:01:48,510 --> 01:01:52,989
currently we, don't have that
feature in yet. It'll be coming

1071
01:01:52,989 --> 01:01:56,000
soon, but I'm gonna show you the
old way.

1072
01:01:56,000 --> 01:02:00,150
Audrow Nash: I, I want to just,
ask one question about where we

1073
01:02:00,150 --> 01:02:04,168
are right now. So whenever you
do Pixi install, it's going to

1074
01:02:04,168 --> 01:02:07,000
set up all the environments. Is
it Correct?

1075
01:02:07,000 --> 01:02:08,000
Ruben Arts: Actually,

1076
01:02:08,000 --> 01:02:09,000
Audrow Nash: so you set up,

1077
01:02:09,000 --> 01:02:13,833
Ruben Arts: I actually misspoke
and it only did the default

1078
01:02:13,833 --> 01:02:18,911
environment again. So it didn't
actually install Rattler build

1079
01:02:18,911 --> 01:02:23,498
it only locked it, into my log
file. If I wanna install,

1080
01:02:23,498 --> 01:02:28,577
either, sorry, install either
all or, I can do also dash e and

1081
01:02:28,577 --> 01:02:33,000
then build. So now the built
environment is installed.

1082
01:02:33,000 --> 01:02:36,409
Audrow Nash: So that makes
sense. So you can select which

1083
01:02:36,409 --> 01:02:40,236
environment, or you can do dash,
all. this is really nice 'cause

1084
01:02:40,236 --> 01:02:43,944
you can only install what you
want if you wanna install only a

1085
01:02:43,944 --> 01:02:47,113
subset of it. Okay. But
otherwise it just updates the

1086
01:02:47,113 --> 01:02:50,701
lock file with these, you don't
actually do the downloading.

1087
01:02:50,701 --> 01:02:51,000
Okay.

1088
01:02:51,000 --> 01:02:52,835
Wolf Vollprecht: This is also
how we deal with like

1089
01:02:52,835 --> 01:02:54,849
development dependencies and
production dependencies and

1090
01:02:54,849 --> 01:02:56,000
things like this. Like you don't

1091
01:02:56,000 --> 01:02:59,741
Audrow Nash: Oh, I love that.
That's so clever. Yes. That's

1092
01:02:59,741 --> 01:03:03,737
really nice. 'cause it's such a
pain. a lot of projects you end

1093
01:03:03,737 --> 01:03:07,415
up shipping your development
dependencies for these things

1094
01:03:07,415 --> 01:03:09,000
and there's no reason to.

1095
01:03:09,000 --> 01:03:10,000
Ruben Arts: yeah,

1096
01:03:09,000 --> 01:03:10,000
Audrow Nash: A lot of times

1097
01:03:10,000 --> 01:03:14,387
Ruben Arts: Yeah. The show there
even one more. Now we're just

1098
01:03:14,387 --> 01:03:18,350
talking about it. One more
amazing feature that I wish I

1099
01:03:18,350 --> 01:03:22,524
had before I was starting on
Pixi. You can also make a prod

1100
01:03:22,524 --> 01:03:26,912
environment called and then do
basically the same thing as you

1101
01:03:26,912 --> 01:03:31,369
did before. So this is already
completed by gt and if you don't

1102
01:03:31,369 --> 01:03:35,898
do the no default features. You
get all the dependencies in this

1103
01:03:35,898 --> 01:03:40,002
one environment. and so this
would be your dev environment

1104
01:03:40,002 --> 01:03:44,318
actually. So dev this depends on
all the features of your own

1105
01:03:44,318 --> 01:03:48,706
environment, but if you then do,
prod environment, which I was

1106
01:03:48,706 --> 01:03:52,810
actually gonna explain and you
say, I want this to be in a

1107
01:03:52,810 --> 01:03:56,772
solve group of food, doesn't
matter which solve group a,

1108
01:03:56,772 --> 01:04:01,018
solve group, you can define
other environments that are also

1109
01:04:01,018 --> 01:04:04,909
in the solve group, for
instance, the test environment.

1110
01:04:04,909 --> 01:04:09,226
And then you could, you, you
make sure that while solving all

1111
01:04:09,226 --> 01:04:13,754
the dependencies are exactly the
same. And then when installing,

1112
01:04:13,754 --> 01:04:18,000
you only install what you need,
which makes it super easy to

1113
01:04:18,000 --> 01:04:19,000
Audrow Nash: so

1114
01:04:18,000 --> 01:04:20,000
Ruben Arts: it up.

1115
01:04:20,000 --> 01:04:23,076
Audrow Nash: And when you say
solve, you mean in terms of how

1116
01:04:23,076 --> 01:04:26,000
it solves for dependencies or
what's, what are we solving?

1117
01:04:26,000 --> 01:04:29,548
Ruben Arts: No, it's just the
dependency puzzle that it

1118
01:04:29,548 --> 01:04:30,000
solves.

1119
01:04:30,000 --> 01:04:33,703
Audrow Nash: That's super, super
cool. I really like that because

1120
01:04:33,703 --> 01:04:37,006
it's often, I, I've worked with
some things where you have

1121
01:04:37,006 --> 01:04:40,367
different environments and it's
very tedious to set them up

1122
01:04:40,367 --> 01:04:43,614
generally. 'cause I end up
copying all of the things from

1123
01:04:43,614 --> 01:04:47,089
production that I want to, all
the things in test that I want

1124
01:04:47,089 --> 01:04:50,506
to, all the thing. And then,
grabbing a, like a, all of that

1125
01:04:50,506 --> 01:04:53,411
is a subset of the dev
environment for this kind of

1126
01:04:53,411 --> 01:04:57,000
thing. And it's so nice that you
have a nice way of doing that.

1127
01:04:57,000 --> 01:05:00,739
Ruben Arts: Yeah. Yeah. I was
really excited about this myself

1128
01:05:00,739 --> 01:05:03,935
as well. I didn't come up with
it. We had some great,

1129
01:05:03,935 --> 01:05:07,492
contributors that thought of
this. I'll delete this for now

1130
01:05:07,492 --> 01:05:09,000
because we don't need it.

1131
01:05:09,000 --> 01:05:11,816
Wolf Vollprecht: one of the
other pain points is also like

1132
01:05:11,816 --> 01:05:14,340
sometimes you have a dev
dependency that may be only

1133
01:05:14,340 --> 01:05:17,252
available for, I don't know,
Python three point 10, and then

1134
01:05:17,252 --> 01:05:20,311
your pro environment might be on
three point 12, but that's not

1135
01:05:20,311 --> 01:05:23,466
what you're testing against. And
so that's also one of the things

1136
01:05:23,466 --> 01:05:26,282
that the Solve group makes sure
of. you're testing exactly

1137
01:05:26,282 --> 01:05:29,000
against the same stuff that
you're using for deployment.

1138
01:05:29,000 --> 01:05:33,075
Audrow Nash: That's really cool.
Okay. Yeah, you guys have made a

1139
01:05:33,075 --> 01:05:36,710
lot of this is a very user
friendly approach. This is very

1140
01:05:36,710 --> 01:05:40,221
nice. I could solve so many
problems that I've had while

1141
01:05:40,221 --> 01:05:44,045
coding and different workspace
environments like this is just

1142
01:05:44,045 --> 01:05:47,931
wonderful. Okay. Hell yeah. So
you have your build environment

1143
01:05:47,931 --> 01:05:50,000
and let's see, back to you,
Ruben

1144
01:05:50,000 --> 01:05:54,634
Ruben Arts: Yep. Yeah. So the
next step would be to make a

1145
01:05:54,634 --> 01:05:59,347
recipe for a project. So I'm
gonna add a simple file that's

1146
01:05:59,347 --> 01:06:04,380
gonna be called Remo, and this
is a com lab package recipe. And

1147
01:06:04,380 --> 01:06:09,413
I'm gonna copy it from what we
prepared, because otherwise it's

1148
01:06:09,413 --> 01:06:14,287
gonna be too much of an typing
problem. but the navigator has

1149
01:06:14,287 --> 01:06:19,000
this recipe and it's a bit
bigger than you want, but sorry.

1150
01:06:19,000 --> 01:06:22,490
Audrow Nash: okay, so you have
40 lines. I don't know anything

1151
01:06:22,490 --> 01:06:25,586
about recipes for this. I'm
imagining it's kinda like a

1152
01:06:25,586 --> 01:06:29,132
Docker file, docker, script. Is
that fair or are they something

1153
01:06:29,132 --> 01:06:32,679
because you're saying all your,
just different, dependencies or

1154
01:06:32,679 --> 01:06:36,000
what, I guess what is the, is
that what it's doing or what,

1155
01:06:36,000 --> 01:06:37,000
Ruben Arts: maybe Wolf should
explain

1156
01:06:37,000 --> 01:06:42,000
Audrow Nash: do I understand
this? Okay, Wolf, go ahead.

1157
01:06:42,000 --> 01:06:46,389
Wolf Vollprecht: Sure. Yeah. So
it's a little bit like a Docker

1158
01:06:46,389 --> 01:06:50,848
file, but. most package managers
have a concept like this. Also,

1159
01:06:50,848 --> 01:06:54,679
like you have for Debian, you
have some sort of I don't

1160
01:06:54,679 --> 01:06:58,790
actually know what they're
called, but files like this that

1161
01:06:58,790 --> 01:07:02,970
define the dependencies of the
things that you need at build

1162
01:07:02,970 --> 01:07:07,358
time at link time. And then you
have some script that does some

1163
01:07:07,358 --> 01:07:11,678
arbitrary commands like CMake
install, blah blah, blah, or PIP

1164
01:07:11,678 --> 01:07:15,509
install. And that installs
things into what we call the

1165
01:07:15,509 --> 01:07:19,620
prefix. And that's also where
the name of the company comes

1166
01:07:19,620 --> 01:07:24,078
from. and it installs that into
the prefix. And then we take the

1167
01:07:24,078 --> 01:07:28,467
diff and the Diff is basically
your new package. And that's how

1168
01:07:28,467 --> 01:07:30,000
the packages are made.

1169
01:07:30,000 --> 01:07:33,691
Ruben Arts: Yep. So this defines
all the required metadata that

1170
01:07:33,691 --> 01:07:37,030
you're gonna put into your
project and also how you built

1171
01:07:37,030 --> 01:07:40,604
your project. So in this case,
we define a script. The script

1172
01:07:40,604 --> 01:07:44,236
doesn't really matter for you
right now, but it just runs some

1173
01:07:44,236 --> 01:07:47,810
either C make or commands or
whatever to build your. Project,

1174
01:07:47,810 --> 01:07:51,560
and then it's gonna move it into
the output. It's gonna add some

1175
01:07:51,560 --> 01:07:54,723
other metadata, like in
description, in this case, the

1176
01:07:54,723 --> 01:07:58,238
version. And, yeah, the conda
tooling, ecosystem understands

1177
01:07:58,238 --> 01:07:59,000
this file. So

1178
01:07:59,000 --> 01:08:02,243
Wolf Vollprecht: Yeah, maybe you
can compare it. Also, it's the

1179
01:08:02,243 --> 01:08:05,537
package XML, actually, like the
package XML is more or less what

1180
01:08:05,537 --> 01:08:06,000
this does

1181
01:08:06,000 --> 01:08:07,000
Ruben Arts: yeah,

1182
01:08:07,000 --> 01:08:09,281
Audrow Nash: One, one thing just
to point out. Yeah. So it's very

1183
01:08:09,281 --> 01:08:11,000
similar to the package at the
mail. That's great.

1184
01:08:11,000 --> 01:08:15,718
Ruben Arts: I have one of these
recipes as well for the talker,

1185
01:08:15,718 --> 01:08:19,913
which I'm quickly gonna add.
this is something as I said

1186
01:08:19,913 --> 01:08:24,256
before, that we want to fix
within Pixi itself, so that we

1187
01:08:24,256 --> 01:08:28,825
don't need an extra tool, but
for now we still require it. So

1188
01:08:28,825 --> 01:08:33,319
it's gonna be exactly the same.
Only that this one does have

1189
01:08:33,319 --> 01:08:37,887
Python and no CMake stuff in
there. this is all we require to

1190
01:08:37,887 --> 01:08:42,231
build the actual packages. And
for that I'm gonna add some

1191
01:08:42,231 --> 01:08:46,650
commands, tasks again, which are
just a little bit long and

1192
01:08:46,650 --> 01:08:51,144
that's why I'm actually making
em a task in Pixi. So you see

1193
01:08:51,144 --> 01:08:55,712
here I added the LER build and
the recipe, yamo. In there you

1194
01:08:55,712 --> 01:09:00,356
define some of the robo stack
and conda forge things, and that

1195
01:09:00,356 --> 01:09:04,999
extra build project is coming
from our own, channel. Which is,

1196
01:09:04,999 --> 01:09:09,643
open and public to everyone to
use. but it's just to make this

1197
01:09:09,643 --> 01:09:13,987
a little bit, oh, sorry, that
was not the correct place. I

1198
01:09:13,987 --> 01:09:18,031
also have the one for the
navigator. So now I have two

1199
01:09:18,031 --> 01:09:22,600
more tasks which can build the
actual coma packages. So let's

1200
01:09:22,600 --> 01:09:26,944
do that now. Xi run, build
Docker, and this, oh, this is a

1201
01:09:26,944 --> 01:09:31,512
bit small for what it usually as
an output framework can have

1202
01:09:31,512 --> 01:09:36,006
really nice output, but your
terminal should be a little bit

1203
01:09:36,006 --> 01:09:40,649
larger. And that was the build
for package. what you'll see in

1204
01:09:40,649 --> 01:09:45,293
your folder, actually it's maybe
the best way to look at it is

1205
01:09:45,293 --> 01:09:49,862
you have an output folder. It
had a build, directory where it

1206
01:09:49,862 --> 01:09:54,280
did the actual build. then you
get a general directory with

1207
01:09:54,280 --> 01:09:58,999
your Linux 64 in my case. And it
has this artifact, which is a.

1208
01:09:58,999 --> 01:10:03,418
Total package for the talker.
this contains everything that

1209
01:10:03,418 --> 01:10:07,762
package requires to, to be
installed and run. So I'm gonna

1210
01:10:07,762 --> 01:10:12,480
also do that for the navigator.
This one is gonna take a little

1211
01:10:12,480 --> 01:10:16,000
bit longer because it has to do
some compiling.

1212
01:10:16,000 --> 01:10:17,000
Audrow Nash: It is bigger

1213
01:10:17,000 --> 01:10:19,000
Ruben Arts: Yep, exactly. but.

1214
01:10:19,000 --> 01:10:23,557
Audrow Nash: and Wolf added the
dependency of, or, said that

1215
01:10:23,557 --> 01:10:28,190
this could be built on OSX. So
you are currently on Linux and

1216
01:10:28,190 --> 01:10:32,823
so it, you didn't add that OSX
one, but if we did this, would

1217
01:10:32,823 --> 01:10:37,000
it also build it for all the
other supported platforms?

1218
01:10:37,000 --> 01:10:40,514
Wolf Vollprecht: the best way to
do that is actually on, let's

1219
01:10:40,514 --> 01:10:43,972
say GitHub actions, for example.
And then you would just have

1220
01:10:43,972 --> 01:10:47,543
your ci and we have very nice CI
integration as well with a red

1221
01:10:47,543 --> 01:10:50,887
label build action and the Pixi
action and all that kind of

1222
01:10:50,887 --> 01:10:54,401
stuff. So that would just go and
build it on all the platforms

1223
01:10:54,401 --> 01:10:58,085
that you're defining this on and
then upload it to a channel. And

1224
01:10:58,085 --> 01:11:01,656
from a channel, you can just use
it in your Pixi TOML again. So

1225
01:11:01,656 --> 01:11:05,000
once you have the package, it's
just, very easy to install.

1226
01:11:05,000 --> 01:11:07,000
Audrow Nash: Nice.

1227
01:11:07,000 --> 01:11:12,991
Ruben Arts: I am quickly gonna
do a little hike on the side of

1228
01:11:12,991 --> 01:11:18,595
my screen. I just made another
folder called ROS Workspace

1229
01:11:18,595 --> 01:11:24,586
Production because now we have
two packages that we built and,

1230
01:11:24,586 --> 01:11:30,770
I can open our workspace that is
closed. You can see the rest of

1231
01:11:30,770 --> 01:11:36,664
my screen, But, now that we have
built those packages, we can

1232
01:11:36,664 --> 01:11:42,558
actually use them. We somewhere
else. do you see this screen?

1233
01:11:42,558 --> 01:11:48,066
Yes. So this is a new Pixi
project. It's completely empty

1234
01:11:48,066 --> 01:11:53,767
again. let's simplify it even
more. we, Don't add any other

1235
01:11:53,767 --> 01:11:59,661
dependencies, but I should add
the ROS. Sorry, roll stack st.

1236
01:11:59,661 --> 01:12:05,845
Staging here as well. And we can
also add the dot slash and then

1237
01:12:05,845 --> 01:12:10,000
the other ROS, workspace, where
we had that

1238
01:12:10,000 --> 01:12:12,000
Audrow Nash: so your path on
your local system,

1239
01:12:12,000 --> 01:12:16,519
Ruben Arts: Yeah, so now I have
it on my local system. We can

1240
01:12:16,519 --> 01:12:21,333
also push it online. I also have
the example prepared online, but

1241
01:12:21,333 --> 01:12:25,852
in this case, that shows exactly
what we're running, from the

1242
01:12:25,852 --> 01:12:30,000
other, folder. And I already
added these things another.

1243
01:12:30,000 --> 01:12:32,500
Audrow Nash: Yeah, the online
one is probably just downloading

1244
01:12:32,500 --> 01:12:34,839
it, downloading the artifact,
and putting it somewhere for

1245
01:12:34,839 --> 01:12:35,000
you.

1246
01:12:35,000 --> 01:12:40,922
Ruben Arts: Yes, exactly. I'm
not sure what I just copied why

1247
01:12:40,922 --> 01:12:45,000
this is coming in. This
shouldn't come in.

1248
01:12:45,000 --> 01:12:48,000
Wolf Vollprecht: It looks like
somewhere there's the.

1249
01:12:48,000 --> 01:12:49,000
Ruben Arts: Sarah.

1250
01:12:49,000 --> 01:12:50,000
Audrow Nash: You can just undo
stuff,

1251
01:12:50,000 --> 01:12:55,285
Ruben Arts: Yeah, I actually
copied it from Markdown, so I've

1252
01:12:55,285 --> 01:13:00,830
expected it to copy. Let me try
again. Yes. So pretty simple. We

1253
01:13:00,830 --> 01:13:05,769
have Humble Desktop, humble
Turtle Sim, and the Navigator

1254
01:13:05,769 --> 01:13:10,621
and the Docker. And we again
have this sim navigator and

1255
01:13:10,621 --> 01:13:14,000
talker command. So I could start
a new,

1256
01:13:14,000 --> 01:13:17,304
Audrow Nash: star, you're saying
whatever version, because so

1257
01:13:17,304 --> 01:13:20,608
navigator equals star and talk
or equals star because it's on

1258
01:13:20,608 --> 01:13:23,967
your system. You're just saying
whatever version it happens to

1259
01:13:23,967 --> 01:13:27,000
be is totally fine. And that's
what that star indicates.

1260
01:13:27,000 --> 01:13:31,929
Ruben Arts: So the solve that we
do is a satisfiability solve

1261
01:13:31,929 --> 01:13:36,778
with the inclusion of get the
highest version we can get. So

1262
01:13:36,778 --> 01:13:41,869
it takes these two packages, it
says, okay, I need specifically

1263
01:13:41,869 --> 01:13:46,556
between these two versions. And
then for the navigator and

1264
01:13:46,556 --> 01:13:51,485
Docker, get whatever you. Can
get, and from those, the latest

1265
01:13:51,485 --> 01:13:56,737
one. and if we then run the Pixi
run sim, you already start that,

1266
01:13:56,737 --> 01:13:59,000
thing. What did I do now? I,

1267
01:13:59,000 --> 01:14:05,000
Wolf Vollprecht: think maybe
just try the dot and

1268
01:14:05,000 --> 01:14:09,678
Ruben Arts: oh, sorry. No, the
example went wrong. Actually.

1269
01:14:09,678 --> 01:14:14,278
Think I mis missed it. This I
will quickly use the, example

1270
01:14:14,278 --> 01:14:19,268
that we prepared online. So what
you can do with these packages,

1271
01:14:19,268 --> 01:14:23,868
they're just artifacts, just
single files. You can drag and

1272
01:14:23,868 --> 01:14:28,234
drop them into our website, and
then you would have them

1273
01:14:28,234 --> 01:14:32,444
available under your own
channel, which is exactly the

1274
01:14:32,444 --> 01:14:37,122
same workflow that you would
have, except that you move them

1275
01:14:37,122 --> 01:14:42,034
to a, channel online. And if you
then install that, this should

1276
01:14:42,034 --> 01:14:46,478
actually. Solve it and, build
the environment and run the

1277
01:14:46,478 --> 01:14:51,000
node. So you will not see my ROS
the turtle sim, but it is

1278
01:14:51,000 --> 01:14:52,000
Audrow Nash: Same reason as
before.

1279
01:14:52,000 --> 01:14:58,120
Ruben Arts: exactly, it is on my
system. It's already running. I

1280
01:14:58,120 --> 01:15:04,143
can show the navigator as well.
sorry, I'm mixing the commands,

1281
01:15:04,143 --> 01:15:09,402
navigator. so this runs
navigator and I can also start,

1282
01:15:09,402 --> 01:15:15,235
the talker pixi run talker. And
that also, that also runs the

1283
01:15:15,235 --> 01:15:16,000
command.

1284
01:15:16,000 --> 01:15:17,000
Audrow Nash: Yep.

1285
01:15:16,000 --> 01:15:17,000
Ruben Arts: I added,

1286
01:15:17,000 --> 01:15:18,000
Audrow Nash: Very cool.

1287
01:15:18,000 --> 01:15:22,000
Ruben Arts: oh, I need to add
some numbers.

1288
01:15:22,000 --> 01:15:23,000
Audrow Nash: Yep.

1289
01:15:23,000 --> 01:15:26,986
Ruben Arts: Yeah. So you see
that the navigator got it. And

1290
01:15:26,986 --> 01:15:31,243
my little turtle is moving, as
we all know. I just, I can maybe

1291
01:15:31,243 --> 01:15:33,000
share that screen quickly.

1292
01:15:33,000 --> 01:15:36,933
Audrow Nash: Yeah, for sure. I
trust you with it, we saw it

1293
01:15:36,933 --> 01:15:38,000
working earlier.

1294
01:15:38,000 --> 01:15:39,000
Wolf Vollprecht: Yay.

1295
01:15:39,000 --> 01:15:44,361
Ruben Arts: But, Please go ahead
and ask him questions while this

1296
01:15:44,361 --> 01:15:47,000
little turtle is doing its thing

1297
01:15:47,000 --> 01:15:51,927
Audrow Nash: Let's see, Wolf, so
these demos are awesome and it

1298
01:15:51,927 --> 01:15:56,542
really shows me how to get
started with it. I think for all

1299
01:15:56,542 --> 01:16:01,000
of my new ROS, two personal
projects, I'll be using this.

1300
01:16:01,000 --> 01:16:05,067
Audrow Nash: what, I guess for
both of you, what are, some of

1301
01:16:05,067 --> 01:16:09,333
the big takeaways from all this?
if some, Wolf, I'll have you go

1302
01:16:09,333 --> 01:16:13,067
first, but what do you hope
someone takes away from this

1303
01:16:13,067 --> 01:16:16,000
demo and this, like what we've
shared today?

1304
01:16:16,000 --> 01:16:20,642
Wolf Vollprecht: so I think, I
hope we were able to demonstrate

1305
01:16:20,642 --> 01:16:25,063
a different sort of approach to.
Installing ROS, which in my

1306
01:16:25,063 --> 01:16:28,821
opinion is one of the bigger
pain points in the ROS

1307
01:16:28,821 --> 01:16:33,316
ecosystem. And while like this
works, and we are really happy

1308
01:16:33,316 --> 01:16:37,958
about it, there's still a lot to
do, in the Robo Stack project.

1309
01:16:37,958 --> 01:16:42,453
For example, we would want to
support the latest, like we are

1310
01:16:42,453 --> 01:16:47,168
currently supporting ROS, humble
very well, but then, we want to

1311
01:16:47,168 --> 01:16:51,663
get ROS Jazzy done and like all
the new things and maybe even

1312
01:16:51,663 --> 01:16:55,937
rolling. And so if we would
have, like we want to grow the

1313
01:16:55,937 --> 01:17:00,284
community, help people to get
started much more easily with

1314
01:17:00,284 --> 01:17:04,632
ROS and get all the latest and
greatest packages on to this

1315
01:17:04,632 --> 01:17:08,832
form of distribution and get
more maintainers to also buy

1316
01:17:08,832 --> 01:17:13,179
into this technology so that we
can just make it easier for

1317
01:17:13,179 --> 01:17:17,305
everyone. I think that's our
goals and that's one of the

1318
01:17:17,305 --> 01:17:19,000
reasons why we're here.

1319
01:17:19,000 --> 01:17:21,583
Audrow Nash: Hell yeah. Ruben,
do you have anything to add for

1320
01:17:21,583 --> 01:17:22,000
takeaways?

1321
01:17:22,000 --> 01:17:25,601
Ruben Arts: Yeah, the, hardest
thing for us now is we are

1322
01:17:25,601 --> 01:17:29,517
currently, we, were all robotics
engineers. Almost everyone in

1323
01:17:29,517 --> 01:17:33,560
the company has been a robotics
engineer. And we are now all not

1324
01:17:33,560 --> 01:17:37,476
running robotics from anymore on
a daily basis. So it's really

1325
01:17:37,476 --> 01:17:41,267
hard to test these things and
it's really hard to make these

1326
01:17:41,267 --> 01:17:44,994
examples. We would love to work
together with more of these

1327
01:17:44,994 --> 01:17:48,910
companies that saw this as an
issue and now fix it, can fix it

1328
01:17:48,910 --> 01:17:52,511
through Pixi, but might need
some help where we can, take

1329
01:17:52,511 --> 01:17:56,617
them as examples and improve the
product as it is right now. just

1330
01:17:56,617 --> 01:18:00,154
one simple example where I
previously worked, we did the

1331
01:18:00,154 --> 01:18:04,197
whole integration with Conda and
we went from an hour and a half

1332
01:18:04,197 --> 01:18:07,861
of just op pip and installing it
from a readme. it was one

1333
01:18:07,861 --> 01:18:11,082
command and it would is
equivalent to Pixi install,

1334
01:18:11,082 --> 01:18:15,062
which to. 10 minutes to download
all the packages and you would

1335
01:18:15,062 --> 01:18:19,042
have an isolated environment. So
that's, my goal, to share that

1336
01:18:19,042 --> 01:18:21,000
with everyone, that experience.

1337
01:18:21,000 --> 01:18:24,586
Audrow Nash: Hell yeah. Okay. I
love it. Alright, thank you

1338
01:18:24,586 --> 01:18:28,536
both. This was so much fun. Like
what a cool thing that I got to,

1339
01:18:28,536 --> 01:18:32,365
see how all this works. And it's
a wonderful tool. Like I, I'll

1340
01:18:32,365 --> 01:18:36,133
be using it and it's exciting
also, it's it seems wonderful in

1341
01:18:36,133 --> 01:18:39,536
the robotics space. It seems
like there's a lot of other

1342
01:18:39,536 --> 01:18:43,000
potential too for using this.
just thank you for sharing.

1343
01:18:43,000 --> 01:18:46,554
Wolf Vollprecht: Yeah. I think a
few things that we haven't

1344
01:18:46,554 --> 01:18:50,349
highlighted yet is that we have
very nice Cuda support as well,

1345
01:18:50,349 --> 01:18:53,000
so Oh, wow. It's the apple
thing, so if you,

1346
01:18:53,000 --> 01:18:56,848
Audrow Nash: that's a really
good thing to highlight. Yeah.

1347
01:18:56,848 --> 01:18:59,000
Yeah. anything we missed that
in?

1348
01:18:59,000 --> 01:19:03,548
Wolf Vollprecht: yeah, like Cuda
support or optimized binaries is

1349
01:19:03,548 --> 01:19:07,747
also something that's coming.
there's already a lot like big

1350
01:19:07,747 --> 01:19:11,455
ecosystem of scientific
libraries on this Conda forge

1351
01:19:11,455 --> 01:19:15,584
distribution because we have a
lot of like data scientists,

1352
01:19:15,584 --> 01:19:20,062
machine learning, AI physicists,
whatever, People that are doing

1353
01:19:20,062 --> 01:19:24,121
number crunching. very similar
fields to robotics and that

1354
01:19:24,121 --> 01:19:28,319
also, yeah, it's just a nice
community to be active in. And,

1355
01:19:28,319 --> 01:19:32,658
speaking of community, one thing
I also wanted to highlight is

1356
01:19:32,658 --> 01:19:36,226
that, we have some, like
Robostack is not a project

1357
01:19:36,226 --> 01:19:40,425
that's only done by me or Ruben,
it's a group of people and,

1358
01:19:40,425 --> 01:19:44,623
especially Tobias Fisher is
doing a lot of work and Sio Zaro

1359
01:19:44,623 --> 01:19:48,892
is helping a lot with making
sure that gazebo and other tools

1360
01:19:48,892 --> 01:19:52,810
are running smoothly on Windows.
And so that's really a,

1361
01:19:52,810 --> 01:19:54,000
community effort.

1362
01:19:54,000 --> 01:19:58,276
Audrow Nash: Okay. That's great.
one thing that I feel like it

1363
01:19:58,276 --> 01:20:02,552
would be good to cover too, what
kind of limitations, like why

1364
01:20:02,552 --> 01:20:06,966
would someone not choose to use
Pixi? like what, are some of the

1365
01:20:06,966 --> 01:20:11,379
things you would run into or, I,
yeah. I guess why would you not

1366
01:20:11,379 --> 01:20:14,000
choose to use it for different
things?

1367
01:20:14,000 --> 01:20:16,000
Wolf Vollprecht: That is a tough
question.

1368
01:20:16,000 --> 01:20:17,000
Ruben Arts: Yeah, I have

1369
01:20:17,000 --> 01:20:22,763
Wolf Vollprecht: like myself,
it's very hard for me to imagine

1370
01:20:22,763 --> 01:20:28,434
a life without Pixi right now. I
think if you, like sometimes

1371
01:20:28,434 --> 01:20:34,197
maybe Ubuntu or something like
this, they have maybe a smaller

1372
01:20:34,197 --> 01:20:39,961
set of packages that might be a
bit more sort of battle tested

1373
01:20:39,961 --> 01:20:45,631
by, more engineers or something
like this because they have a

1374
01:20:45,631 --> 01:20:50,000
larger like paid staff to do
things, let's say.

1375
01:20:50,000 --> 01:20:52,000
Audrow Nash: optimize only for,
say, an Ubuntu target,

1376
01:20:52,000 --> 01:20:56,044
Wolf Vollprecht: right? Yeah. So
they have one distribution and

1377
01:20:56,044 --> 01:20:59,768
one set of like versions that
they might test that they're

1378
01:20:59,768 --> 01:21:03,491
working together while we
support, like going back in time

1379
01:21:03,491 --> 01:21:07,279
and stuff like this and mixing
versions and everything. but

1380
01:21:07,279 --> 01:21:11,259
obviously I don't want to say
that things cannot be stable. So

1381
01:21:11,259 --> 01:21:15,240
they are stable usually, but,
Yeah. You live a bit more on the

1382
01:21:15,240 --> 01:21:18,000
edge, I would say, with, with
our solution.

1383
01:21:18,000 --> 01:21:19,000
Ruben Arts: I am not,

1384
01:21:18,000 --> 01:21:21,397
Audrow Nash: And that might not
be as true with lock files and

1385
01:21:21,397 --> 01:21:22,000
things like

1386
01:21:22,000 --> 01:21:27,000
Wolf Vollprecht: So lock files
also save you from that problem.

1387
01:21:27,000 --> 01:21:29,000
Audrow Nash: Okay. Ruben, you
were gonna say something.

1388
01:21:29,000 --> 01:21:33,516
Ruben Arts: oh yeah. The living
on the edge might not be the

1389
01:21:33,516 --> 01:21:37,656
best way of saying it, but you
live on the edge of what

1390
01:21:37,656 --> 01:21:42,398
computer stuff is doing, as in
the latest, greatest, might be a

1391
01:21:42,398 --> 01:21:46,839
better way of saying it. which
sometimes brings some, some,

1392
01:21:46,839 --> 01:21:50,000
pain of being the latest,
greatest worker.

1393
01:21:50,000 --> 01:21:53,567
Audrow Nash: But, so I guess
with the question about

1394
01:21:53,567 --> 01:21:57,888
limitations, is there any reason
that a company wouldn't choose

1395
01:21:57,888 --> 01:22:01,798
to use Robo Stack for like their
developer use? if, I'm a

1396
01:22:01,798 --> 01:22:06,119
developer at a robotics company,
is there a reason that I would

1397
01:22:06,119 --> 01:22:09,000
choose not to use Robo? not Robo
pre Pixi.

1398
01:22:09,000 --> 01:22:13,019
Wolf Vollprecht: So I, think if
you are really at the robotics

1399
01:22:13,019 --> 01:22:16,715
company, like one of your
problems might be that we don't

1400
01:22:16,715 --> 01:22:20,605
cover all the packages right
now. intrinsic is only, or Open

1401
01:22:20,605 --> 01:22:24,365
Robotics is only building the
packages for Debian. I'm not

1402
01:22:24,365 --> 01:22:28,319
sure if they're doing for Fedora
these days or not. but yeah,

1403
01:22:28,319 --> 01:22:32,339
basically like Upstream is only
building it for Ubuntu, right?

1404
01:22:32,339 --> 01:22:36,358
and so that's, we would love to
get there. So we would love to

1405
01:22:36,358 --> 01:22:39,794
work with, those guys and make
sure that this is like

1406
01:22:39,794 --> 01:22:43,749
officially supported and maybe
the build farm is building the

1407
01:22:43,749 --> 01:22:47,639
packages instead of our own CI
that runs some GitHub actions

1408
01:22:47,639 --> 01:22:49,000
and things like that.

1409
01:22:49,000 --> 01:22:50,000
Audrow Nash: Ah

1410
01:22:50,000 --> 01:22:54,968
Wolf Vollprecht: so having, I
think that would be one of the

1411
01:22:54,968 --> 01:23:00,019
main reasons why you might want
to hold off is that Yeah, the

1412
01:23:00,019 --> 01:23:03,000
offic, it's not official yet. I
hope

1413
01:23:03,000 --> 01:23:07,138
Audrow Nash: Yeah. I also think
this conversation, like I'll

1414
01:23:07,138 --> 01:23:11,276
bring this to some of the open
robotics folks and I hope for

1415
01:23:11,276 --> 01:23:15,345
more official support too.
'cause it seems like a wonderful

1416
01:23:15,345 --> 01:23:19,000
tool. okay. Any, anything else
to add, Ruben or Wolf?

1417
01:23:19,000 --> 01:23:23,558
Ruben Arts: I think there's one
thing that I walk into sometimes

1418
01:23:23,558 --> 01:23:27,902
myself, which is we use this
cached sim linked version of the

1419
01:23:27,902 --> 01:23:32,460
packages, but we don't force the
cache on environments. So if my

1420
01:23:32,460 --> 01:23:36,662
environment has Python three
point 12 and Wolfs environment

1421
01:23:36,662 --> 01:23:41,077
he updated has three point 11,
and I install them both, I will

1422
01:23:41,077 --> 01:23:45,635
get both versions of python, at
least cached on my system. So if

1423
01:23:45,635 --> 01:23:49,623
you do need to think about
sometimes, do I need all this

1424
01:23:49,623 --> 01:23:54,039
cache? then you can just remove
it. Basically clean cache and,

1425
01:23:54,039 --> 01:23:58,312
it'll download everything you
need again. But it at least is

1426
01:23:58,312 --> 01:24:02,585
something to think about. It's,
not a why not to use it, but

1427
01:24:02,585 --> 01:24:07,000
it's more of a. Make sure that
you understand what it's doing.

1428
01:24:07,000 --> 01:24:09,000
Audrow Nash: I feel like you
could make that nicer

1429
01:24:09,000 --> 01:24:10,000
Ruben Arts: Oh, yeah.

1430
01:24:10,000 --> 01:24:15,223
Audrow Nash: through, so like
additional info in the lock file

1431
01:24:15,223 --> 01:24:19,857
or something like that,
potentially, or something else.

1432
01:24:19,857 --> 01:24:25,249
But I. All right. So that seems
great. what do you guys, so Wolf

1433
01:24:25,249 --> 01:24:30,051
thinking about the next year,
and then maybe three years,

1434
01:24:30,051 --> 01:24:33,000
where do you hope you guys will
be?

1435
01:24:33,000 --> 01:24:37,046
Wolf Vollprecht: that is a good
question. so first of all, we

1436
01:24:37,046 --> 01:24:40,959
have our immediate roadmap,
right? For Pixi. One of the big

1437
01:24:40,959 --> 01:24:45,071
items that Rub already talked
about is Pixi build. So we wanna

1438
01:24:45,071 --> 01:24:48,785
make sure that you can build
packages right from Pixi to

1439
01:24:48,785 --> 01:24:52,632
files. And we have our first
prototype of that ready since

1440
01:24:52,632 --> 01:24:56,280
this week. then we have other
items that are more about

1441
01:24:56,280 --> 01:25:00,127
software security and supply
chain securities. so we wanna

1442
01:25:00,127 --> 01:25:04,239
integrate something called Six
Store. I don't know if you know

1443
01:25:04,239 --> 01:25:07,356
it, but it's basically
cryptographic signing of

1444
01:25:07,356 --> 01:25:11,269
packages. So if Open Robotics
would decide to publish their

1445
01:25:11,269 --> 01:25:15,315
own, Conda packages, then they
could attach a little, secret,

1446
01:25:15,315 --> 01:25:19,361
cryptographic thing and, that
would then make it, add a check

1447
01:25:19,361 --> 01:25:23,406
mark to the, package that would
say, okay, this is officially

1448
01:25:23,406 --> 01:25:27,319
built by open robotics. Or we,
as prefix could say, this is

1449
01:25:27,319 --> 01:25:31,299
officially built by a prefix.
And users could make sure that

1450
01:25:31,299 --> 01:25:34,814
yeah, they only get authentic
stuff. so that's on the

1451
01:25:34,814 --> 01:25:38,727
immediate roadmap I would say
for Pixi and our projects. We

1452
01:25:38,727 --> 01:25:42,375
are also, Ruben showed you
Rattler Build. And so we are

1453
01:25:42,375 --> 01:25:46,222
currently working extremely hard
on getting that into this

1454
01:25:46,222 --> 01:25:49,803
distribution so that they can
make use of all of these

1455
01:25:49,803 --> 01:25:53,716
improvements because Rattler
Build is also re-engineered on

1456
01:25:53,716 --> 01:25:57,696
top of our low level library
that's called Rattler. And it's

1457
01:25:57,696 --> 01:26:01,476
much better to build Conda
packages that way. And we have

1458
01:26:01,476 --> 01:26:05,058
an amazing community that's
already trying it out. For

1459
01:26:05,058 --> 01:26:09,170
example, folks from Nvidia or QU
Co or different companies are

1460
01:26:09,170 --> 01:26:13,282
like starting to use it. And so
that's also really exciting to

1461
01:26:13,282 --> 01:26:17,195
see. and then lastly, we would
just hope that, we have much

1462
01:26:17,195 --> 01:26:20,777
more distribution of Pixi
itself, because I wanna see,

1463
01:26:20,777 --> 01:26:24,823
research papers, for example, if
you have a robotics research

1464
01:26:24,823 --> 01:26:29,067
paper, right? Like it's usually
maybe you have a link to GitHub,

1465
01:26:29,067 --> 01:26:32,848
but then can you actually
execute it like. What, like you

1466
01:26:32,848 --> 01:26:37,092
have to pip install this or APT
get installed that and then it's

1467
01:26:37,092 --> 01:26:40,807
pretty tough usually. And we
would love to see more like

1468
01:26:40,807 --> 01:26:44,919
research being done with Pixi
and make it more reproducible as

1469
01:26:44,919 --> 01:26:49,031
well. if, you have a paper that
says PIP install this and will

1470
01:26:49,031 --> 01:26:53,077
it work in 10 years from now? I
don't know. But like with the

1471
01:26:53,077 --> 01:26:55,000
lock file you at least. Yeah.

1472
01:26:55,000 --> 01:26:59,067
Audrow Nash: Like many research
things, it's six months later

1473
01:26:59,067 --> 01:27:01,000
they're completely impossible

1474
01:27:01,000 --> 01:27:02,000
Wolf Vollprecht: And so with the

1475
01:27:02,000 --> 01:27:03,000
Audrow Nash: get running

1476
01:27:03,000 --> 01:27:07,310
Wolf Vollprecht: yeah, and with
the lock file, we're trying to

1477
01:27:07,310 --> 01:27:11,411
solve that problem as well. And
we're currently a anchoring

1478
01:27:11,411 --> 01:27:15,720
ourselves a little bit in that
scientific community. there and

1479
01:27:15,720 --> 01:27:19,752
then, three years out, instead
of people like doing GitHub

1480
01:27:19,752 --> 01:27:23,992
releases, we would love to see
them push conduct packages and

1481
01:27:23,992 --> 01:27:28,093
make that the universal sort of
format. ideally it would be

1482
01:27:28,093 --> 01:27:32,403
standardized, like screws or I
don't know, things like that so

1483
01:27:32,403 --> 01:27:36,921
that everyone knows what they're
getting and it's one format. You

1484
01:27:36,921 --> 01:27:40,883
don't need to use brew, you
don't need to use APT get, at

1485
01:27:40,883 --> 01:27:45,054
least not for your like, user
level software. yeah, we would

1486
01:27:45,054 --> 01:27:47,000
love to replace all of them.

1487
01:27:47,000 --> 01:27:51,049
Audrow Nash: That would be
amazing. Hell yeah. Big plans. I

1488
01:27:51,049 --> 01:27:54,000
love it. Let's see. Ruben,
anything to add?

1489
01:27:54,000 --> 01:27:59,000
Ruben Arts: No, I follow that
vision completely.

1490
01:27:59,000 --> 01:28:03,200
Audrow Nash: Hell yeah. said. so
I. That's all for today. Thank

1491
01:28:03,200 --> 01:28:05,000
you guys. This was a blast.

1492
01:28:05,000 --> 01:28:06,000
Ruben Arts: Thank you very

1493
01:28:06,000 --> 01:28:07,000
Wolf Vollprecht: you.

1494
01:28:07,000 --> 01:28:12,000
Ruben Arts: It's great to have
you, or great having us.

1495
01:28:12,000 --> 01:28:15,000
Audrow Nash: Yes, for sure. All
right. Bye everyone. ​

1496
01:28:15,000 --> 01:28:18,155
Audrow Nash: That's all for
today's episode. I hope you

1497
01:28:18,155 --> 01:28:21,826
enjoyed this format and learned
something new about Pixi and how

1498
01:28:21,826 --> 01:28:25,039
it can help your robotics
projects. If you're interested

1499
01:28:25,039 --> 01:28:28,767
in trying out Pixi or just wanna
learn more, be sure to check out

1500
01:28:28,767 --> 01:28:32,324
the links in our description. As
we wrap up, I wanna leave you

1501
01:28:32,324 --> 01:28:35,307
with a question. How do you
currently handle package

1502
01:28:35,307 --> 01:28:38,749
management in your robotics
projects? Are there any specific

1503
01:28:38,749 --> 01:28:42,133
challenges you're facing and do
you think Pixi could be the

1504
01:28:42,133 --> 01:28:45,518
solution you're looking for?
Thanks again to Wolf and Ruben

1505
01:28:45,518 --> 01:28:48,960
for joining me today. Don't
forget to like and subscribe for

1506
01:28:48,960 --> 01:28:52,000
more content and we'll catch you
in the next episode.
