1
00:00:00,000 --> 00:00:03,975
Alright, so this
is a project that,

2
00:00:03,975 --> 00:00:06,630
as you can see from the
authorship list here,

3
00:00:06,630 --> 00:00:10,260
has been a lot of
people actually,

4
00:00:10,260 --> 00:00:11,670
it's all undergraduate
students that have

5
00:00:11,670 --> 00:00:13,590
worked on this project
other than myself.

6
00:00:13,590 --> 00:00:14,550
So I've just listed them

7
00:00:14,550 --> 00:00:16,660
in alphabetical
order here.

8
00:00:16,760 --> 00:00:18,870
Some of them were,
of course staff,

9
00:00:18,870 --> 00:00:20,790
some of them were students

10
00:00:20,790 --> 00:00:21,990
interested in the topic.

11
00:00:21,990 --> 00:00:23,160
I do want to
particularly shout

12
00:00:23,160 --> 00:00:24,930
out gabriel, Wow,

13
00:00:24,930 --> 00:00:26,550
who came up with

14
00:00:26,550 --> 00:00:29,130
the initial design
for this thing.

15
00:00:29,130 --> 00:00:31,590
So I'm excited to

16
00:00:31,590 --> 00:00:32,520
present on behalf of

17
00:00:32,520 --> 00:00:34,735
the people who
actually did the work.

18
00:00:34,735 --> 00:00:38,615
So if you try
to investigate

19
00:00:38,615 --> 00:00:40,880
how people's
experiences go

20
00:00:40,880 --> 00:00:42,380
when they're learning
Rust borrow checker,

21
00:00:42,380 --> 00:00:43,460
you start to find

22
00:00:43,460 --> 00:00:45,830
really interesting
quotes or poems,

23
00:00:45,830 --> 00:00:47,910
almost this

24
00:00:48,670 --> 00:00:51,200
poem here from a
study conducted by

25
00:00:51,200 --> 00:00:52,670
Michael goblins
and colleagues as

26
00:00:52,670 --> 00:00:54,170
learning Russian
ownership is like

27
00:00:54,170 --> 00:00:55,760
navigating a maze
where the walls

28
00:00:55,760 --> 00:00:57,559
are made of asbestos
and frustration.

29
00:00:57,559 --> 00:00:59,075
And the maze has no exit.

30
00:00:59,075 --> 00:01:00,380
And every time you
hit a dead end

31
00:01:00,380 --> 00:01:02,300
you get an
aneurism and die.

32
00:01:02,300 --> 00:01:05,570
So I love this quote.

33
00:01:05,570 --> 00:01:07,370
It really expresses

34
00:01:07,370 --> 00:01:09,440
the frustration
that students

35
00:01:09,440 --> 00:01:11,180
feel in

36
00:01:11,180 --> 00:01:13,684
navigating this
so-called alien concept,

37
00:01:13,684 --> 00:01:16,460
this biggest struggle
that they face.

38
00:01:16,460 --> 00:01:18,770
The phrase fighting
the borrower checkers

39
00:01:18,770 --> 00:01:19,820
come up today and it comes

40
00:01:19,820 --> 00:01:20,900
up all over the place on

41
00:01:20,900 --> 00:01:23,200
the web when people
discuss these things.

42
00:01:23,200 --> 00:01:25,550
And we'll have an
interesting paper just

43
00:01:25,550 --> 00:01:26,690
kinda investigating
how people

44
00:01:26,690 --> 00:01:28,415
talk about those things.

45
00:01:28,415 --> 00:01:30,785
So why, why is,

46
00:01:30,785 --> 00:01:32,510
why is it rusts so tricky?

47
00:01:32,510 --> 00:01:32,750
Why is

48
00:01:32,750 --> 00:01:33,950
this ownership and
borrowing system

49
00:01:33,950 --> 00:01:34,805
so tricky?

50
00:01:34,805 --> 00:01:36,965
And you know, just,

51
00:01:36,965 --> 00:01:38,300
just from kind of thing.

52
00:01:38,300 --> 00:01:39,440
But the problem
looking at some of

53
00:01:39,440 --> 00:01:40,880
these these these
prior papers,

54
00:01:40,880 --> 00:01:43,070
you get the sense that
the real issue is

55
00:01:43,070 --> 00:01:44,210
that there's all
this invisible

56
00:01:44,210 --> 00:01:45,320
state that's happening.

57
00:01:45,320 --> 00:01:47,390
It's static state.

58
00:01:47,390 --> 00:01:50,480
Who owns the resource
and it's the life.

59
00:01:50,480 --> 00:01:52,790
How long has this
lifetime extend?

60
00:01:52,790 --> 00:01:54,800
What capabilities
do I have given

61
00:01:54,800 --> 00:01:56,240
the uniqueness
constraints that

62
00:01:56,240 --> 00:01:58,820
are imposed by the
reference system?

63
00:01:58,820 --> 00:02:00,200
And, um, all of these,

64
00:02:00,200 --> 00:02:01,910
all of these pieces

65
00:02:01,910 --> 00:02:02,870
of invisible state and

66
00:02:02,870 --> 00:02:03,620
the rules that
govern them,

67
00:02:03,620 --> 00:02:05,450
they all have to be
mentally stimulated

68
00:02:05,450 --> 00:02:07,280
by the programmer, right?

69
00:02:07,280 --> 00:02:08,420
So they have to
learn to mentally

70
00:02:08,420 --> 00:02:10,130
simulate these
rules and then

71
00:02:10,130 --> 00:02:11,570
keep a model in
their head as

72
00:02:11,570 --> 00:02:13,145
they comprehend,

73
00:02:13,145 --> 00:02:15,275
attempt to
comprehend code.

74
00:02:15,275 --> 00:02:17,660
That's a very
complicated thing.

75
00:02:17,660 --> 00:02:19,100
A lot of cognitive load

76
00:02:19,100 --> 00:02:20,540
is incurred when
you do that.

77
00:02:20,540 --> 00:02:22,940
And so if we want to build

78
00:02:22,940 --> 00:02:25,790
tools to lessen that
cognitive load,

79
00:02:25,790 --> 00:02:27,980
general principle that
we might follow is to

80
00:02:27,980 --> 00:02:30,800
make visible
the invisible.

81
00:02:30,800 --> 00:02:32,600
To help learners, we

82
00:02:32,600 --> 00:02:34,790
do this invisible work
for them instead of

83
00:02:34,790 --> 00:02:35,990
requiring them
to do it all

84
00:02:35,990 --> 00:02:37,190
in their head when
they still haven't

85
00:02:37,190 --> 00:02:40,370
really figured out
the rules themselves.

86
00:02:40,370 --> 00:02:41,900
And then you just show
it to them in some way

87
00:02:41,900 --> 00:02:43,415
that's comprehensible
in some way that

88
00:02:43,415 --> 00:02:45,230
they can refer to

89
00:02:45,230 --> 00:02:46,970
and engaging in a couple
of different things.

90
00:02:46,970 --> 00:02:49,490
One, as they're developing
this mental model,

91
00:02:49,490 --> 00:02:51,350
these visualizations can

92
00:02:51,350 --> 00:02:53,690
help provide scaffolding.

93
00:02:53,690 --> 00:02:56,900
So they provide
the vocabulary,

94
00:02:56,900 --> 00:02:57,980
the key structures,
the key

95
00:02:57,980 --> 00:02:59,030
concepts that they need to

96
00:02:59,030 --> 00:03:00,140
eventually be
able to come up

97
00:03:00,140 --> 00:03:01,430
with on their own
in their head.

98
00:03:01,430 --> 00:03:02,915
But until they do that,

99
00:03:02,915 --> 00:03:04,460
they can see that this is

100
00:03:04,460 --> 00:03:06,695
what I need to be able
to produce in my head.

101
00:03:06,695 --> 00:03:09,140
Then as they start
attempting to do that,

102
00:03:09,140 --> 00:03:10,805
as they start
to develop and

103
00:03:10,805 --> 00:03:13,490
try to apply their
mental model.

104
00:03:13,490 --> 00:03:15,350
It gives them
a way to take

105
00:03:15,350 --> 00:03:16,970
these, to make
them scientists,

106
00:03:16,970 --> 00:03:18,020
to take the predictions
that they're

107
00:03:18,020 --> 00:03:20,000
generating and
allow them to

108
00:03:20,000 --> 00:03:21,800
immediately verify that
what they predicted

109
00:03:21,800 --> 00:03:22,760
is actually
what's happening

110
00:03:22,760 --> 00:03:23,885
if it's not visible,

111
00:03:23,885 --> 00:03:26,165
there's no way to
verify your prediction.

112
00:03:26,165 --> 00:03:28,370
So that's a lot of words.

113
00:03:28,370 --> 00:03:29,720
I think this is a
general principle that I

114
00:03:29,720 --> 00:03:31,250
think is important
and told

115
00:03:31,250 --> 00:03:33,140
us I just want

116
00:03:33,140 --> 00:03:34,820
to jump into a
demo of the rest

117
00:03:34,820 --> 00:03:36,170
of this tool which
is focused on

118
00:03:36,170 --> 00:03:37,250
visualizing these
ownership and

119
00:03:37,250 --> 00:03:38,825
borrowing things in Rust.

120
00:03:38,825 --> 00:03:44,270
So let's switch
over to that tab.

121
00:03:44,270 --> 00:03:48,440
So what the system is,

122
00:03:48,440 --> 00:03:50,975
is it's a system for
including visualizations

123
00:03:50,975 --> 00:03:54,215
in narrative,
instructional narrative.

124
00:03:54,215 --> 00:03:56,870
So we are based on
the empty book system

125
00:03:56,870 --> 00:03:58,760
that a lot of rust
documentation

126
00:03:58,760 --> 00:04:00,575
and Russ books and
things already used.

127
00:04:00,575 --> 00:04:02,420
So you can see this
isn't empty book

128
00:04:02,420 --> 00:04:04,070
tutorial that
we've developed.

129
00:04:04,070 --> 00:04:05,540
And I'll talk
more about how we

130
00:04:05,540 --> 00:04:07,100
integrated into our
classroom and everything.

131
00:04:07,100 --> 00:04:09,140
But this is an
introduction to Ross.

132
00:04:09,140 --> 00:04:09,785
It's meant to be

133
00:04:09,785 --> 00:04:12,110
a 30-minute
introduction to rust,

134
00:04:12,110 --> 00:04:13,490
designed for people with

135
00:04:13,490 --> 00:04:14,600
some programming
background

136
00:04:14,600 --> 00:04:15,890
but no rest background.

137
00:04:15,890 --> 00:04:20,120
And we'll go through
some of the tutorial.

138
00:04:20,120 --> 00:04:20,600
You will see that

139
00:04:20,600 --> 00:04:22,190
all these code
examples we will have

140
00:04:22,190 --> 00:04:24,710
alongside of them,
this visualization.

141
00:04:24,710 --> 00:04:27,185
And it's an interactive
visualization.

142
00:04:27,185 --> 00:04:28,280
That word interactive is

143
00:04:28,280 --> 00:04:29,630
maybe doing a little
too much work.

144
00:04:29,630 --> 00:04:30,740
It's, what I mean by

145
00:04:30,740 --> 00:04:32,960
interactive is that
it allows you to

146
00:04:32,960 --> 00:04:35,465
hover over different
visual elements and get

147
00:04:35,465 --> 00:04:38,030
a verbal description of

148
00:04:38,030 --> 00:04:39,980
what that visual
element means.

149
00:04:39,980 --> 00:04:43,100
Okay, so let me, so
this initial examples

150
00:04:43,100 --> 00:04:44,420
meant to actually
showcase a lot of

151
00:04:44,420 --> 00:04:46,100
the different
visualization features.

152
00:04:46,100 --> 00:04:47,390
And then we'll go
through some of

153
00:04:47,390 --> 00:04:48,470
the more basic examples.

154
00:04:48,470 --> 00:04:49,610
But you can see as I hover

155
00:04:49,610 --> 00:04:53,225
over aspects of
this visualization,

156
00:04:53,225 --> 00:04:55,745
both the visualization,

157
00:04:55,745 --> 00:04:57,590
but both the hover
message shows up,

158
00:04:57,590 --> 00:04:59,570
showing you what
that element means.

159
00:04:59,570 --> 00:05:00,800
And then you see
some changes in

160
00:05:00,800 --> 00:05:02,270
the code itself
highlighting,

161
00:05:02,270 --> 00:05:02,990
in this case

162
00:05:02,990 --> 00:05:04,340
the corresponding
function string

163
00:05:04,340 --> 00:05:05,360
from and underlining

164
00:05:05,360 --> 00:05:06,710
the line just so you know.

165
00:05:06,710 --> 00:05:08,675
Which line you're
talking about,

166
00:05:08,675 --> 00:05:09,710
then you can see what's

167
00:05:09,710 --> 00:05:10,940
happening here is that

168
00:05:10,940 --> 00:05:14,375
it's visualizing
a move or cava,

169
00:05:14,375 --> 00:05:15,680
resource returned
from string

170
00:05:15,680 --> 00:05:17,660
from and moved to S.

171
00:05:17,660 --> 00:05:19,040
And then when you
hover over this

172
00:05:19,040 --> 00:05:20,540
kinda dot event on S,

173
00:05:20,540 --> 00:05:21,800
you see that S acquires

174
00:05:21,800 --> 00:05:23,225
ownership of a resource.

175
00:05:23,225 --> 00:05:24,440
And then from
there on out,

176
00:05:24,440 --> 00:05:25,700
if you hover
over this line,

177
00:05:25,700 --> 00:05:28,370
you see that S is the
owner of this resource.

178
00:05:28,370 --> 00:05:31,280
And it also communicates
something about

179
00:05:31,280 --> 00:05:33,980
the variable S being

180
00:05:33,980 --> 00:05:34,880
mutable since it was

181
00:05:34,880 --> 00:05:36,365
introduced with
let me mute.

182
00:05:36,365 --> 00:05:38,345
The binding can
be reassigned.

183
00:05:38,345 --> 00:05:38,660
It's

184
00:05:38,660 --> 00:05:41,015
kinda visualization of
these capabilities.

185
00:05:41,015 --> 00:05:42,920
So you can see
there's more of that

186
00:05:42,920 --> 00:05:44,030
for different things like

187
00:05:44,030 --> 00:05:46,295
borrows and things
here as well.

188
00:05:46,295 --> 00:05:48,845
So that's the, that's the,

189
00:05:48,845 --> 00:05:50,210
That's rough is okay,

190
00:05:50,210 --> 00:05:51,380
so it's a way of

191
00:05:51,380 --> 00:05:53,720
visualizing code
snippets like this.

192
00:05:53,720 --> 00:05:55,160
And if tutorial, what

193
00:05:55,160 --> 00:05:56,360
we expect people
to do with it

194
00:05:56,360 --> 00:05:57,740
is actually
just start with

195
00:05:57,740 --> 00:05:59,345
a very simple
visualizations.

196
00:05:59,345 --> 00:06:00,845
For example, just kind of

197
00:06:00,845 --> 00:06:02,330
initialization of a heap

198
00:06:02,330 --> 00:06:03,560
allocated string here.

199
00:06:03,560 --> 00:06:05,570
And you can see what

200
00:06:05,570 --> 00:06:08,165
happens is the same
thing we just saw.

201
00:06:08,165 --> 00:06:09,320
This is not letting me, so

202
00:06:09,320 --> 00:06:11,120
the binding cannot
be reassigned.

203
00:06:11,120 --> 00:06:13,280
And then you see something
happening with S.

204
00:06:13,280 --> 00:06:14,915
In this case it's
a print line which

205
00:06:14,915 --> 00:06:17,855
just reads from s, OK.

206
00:06:17,855 --> 00:06:19,955
And then
eventually it ends

207
00:06:19,955 --> 00:06:22,145
up at the end
of the scope of

208
00:06:22,145 --> 00:06:26,855
S and we know
enlarge the screen.

209
00:06:26,855 --> 00:06:31,415
So that way that works.

210
00:06:31,415 --> 00:06:33,485
Okay, so at the
end of dystopia

211
00:06:33,485 --> 00:06:35,375
of S, S goes out of scope.

212
00:06:35,375 --> 00:06:36,350
But we also talk not

213
00:06:36,350 --> 00:06:37,430
just about the
scoping here,

214
00:06:37,430 --> 00:06:38,510
which is an
important concept,

215
00:06:38,510 --> 00:06:40,685
but also about the,

216
00:06:40,685 --> 00:06:42,530
the resource being
dropped, right?

217
00:06:42,530 --> 00:06:44,210
So because as
owns resource at

218
00:06:44,210 --> 00:06:45,110
that point and
it goes out of

219
00:06:45,110 --> 00:06:46,790
scope, the
resources trunk.

220
00:06:46,790 --> 00:06:49,610
Then as we go through
the tutorial,

221
00:06:49,610 --> 00:06:50,365
you see that we start

222
00:06:50,365 --> 00:06:51,860
introducing more
concepts like

223
00:06:51,860 --> 00:06:54,560
moves via, I'll
let bindings.

224
00:06:54,560 --> 00:06:56,030
So you can see that x

225
00:06:56,030 --> 00:06:57,620
is resources moved here.

226
00:06:57,620 --> 00:06:59,495
There's a move and

227
00:06:59,495 --> 00:07:01,520
why acquires ownership
from that move.

228
00:07:01,520 --> 00:07:03,800
And notably, there's
also not a line here

229
00:07:03,800 --> 00:07:05,960
that makes it clear that

230
00:07:05,960 --> 00:07:07,100
x no longer owns

231
00:07:07,100 --> 00:07:08,540
a resource when
the move happens.

232
00:07:08,540 --> 00:07:11,450
So that's a key idea
with ownership moves

233
00:07:11,450 --> 00:07:13,070
is movement is really

234
00:07:13,070 --> 00:07:14,540
like handing off
to something.

235
00:07:14,540 --> 00:07:15,530
And you see that hands-off

236
00:07:15,530 --> 00:07:17,029
happening in the
visualization.

237
00:07:17,029 --> 00:07:19,070
And also at the end here
you see that x goes

238
00:07:19,070 --> 00:07:21,020
out of scope because

239
00:07:21,020 --> 00:07:23,120
it doesn't hurt
to resource at

240
00:07:23,120 --> 00:07:24,290
that point because
of the move,

241
00:07:24,290 --> 00:07:25,310
no resources dropped.

242
00:07:25,310 --> 00:07:26,600
Whereas when y
goes out of scope,

243
00:07:26,600 --> 00:07:28,430
it now it owns the
resource at that point,

244
00:07:28,430 --> 00:07:30,185
so the resource
is dropped.

245
00:07:30,185 --> 00:07:32,180
So it's giving you
the information

246
00:07:32,180 --> 00:07:34,085
that if I asked
you a question

247
00:07:34,085 --> 00:07:36,380
about when a resource
is dropped or what

248
00:07:36,380 --> 00:07:37,550
capabilities x has at

249
00:07:37,550 --> 00:07:39,230
a particular line
in this code.

250
00:07:39,230 --> 00:07:40,100
These are the kinds

251
00:07:40,100 --> 00:07:41,255
of thoughts that
you'd have to,

252
00:07:41,255 --> 00:07:41,960
you'd have to build

253
00:07:41,960 --> 00:07:42,770
this mental model in

254
00:07:42,770 --> 00:07:43,490
your head and we're trying

255
00:07:43,490 --> 00:07:44,600
to put that in front of

256
00:07:44,600 --> 00:07:46,760
you so that hopefully,

257
00:07:46,760 --> 00:07:49,490
as a student you can
learn to generate

258
00:07:49,490 --> 00:07:51,230
these things more
quickly on your

259
00:07:51,230 --> 00:07:54,005
own after some time.

260
00:07:54,005 --> 00:07:58,790
Okay, So the
tutorial goes,

261
00:07:58,790 --> 00:08:00,170
I'm not gonna go
through every example

262
00:08:00,170 --> 00:08:01,640
in the tutorial.
There is a link.

263
00:08:01,640 --> 00:08:03,860
I'll post that
maybe in the Zulu

264
00:08:03,860 --> 00:08:06,275
up as well if people
want to play with it.

265
00:08:06,275 --> 00:08:08,720
For borrows, there are

266
00:08:08,720 --> 00:08:10,415
some interesting things
that start happening.

267
00:08:10,415 --> 00:08:14,570
So you can see that
with a borrow,

268
00:08:14,570 --> 00:08:16,729
there's kinda
two variables

269
00:08:16,729 --> 00:08:18,590
are things of interests.

270
00:08:18,590 --> 00:08:20,435
There's the
reference itself,

271
00:08:20,435 --> 00:08:21,830
and then there's
the resource

272
00:08:21,830 --> 00:08:23,570
that the reference
refers to.

273
00:08:23,570 --> 00:08:26,225
And so we indicate that by

274
00:08:26,225 --> 00:08:28,220
showing both
this reference S

275
00:08:28,220 --> 00:08:30,425
in that function f,

276
00:08:30,425 --> 00:08:34,220
we mark the, we

277
00:08:34,220 --> 00:08:35,780
indicate that
the thing that's

278
00:08:35,780 --> 00:08:37,295
being referred to with
this de-reference,

279
00:08:37,295 --> 00:08:37,520
right?

280
00:08:37,520 --> 00:08:39,500
So star S, like
what capabilities

281
00:08:39,500 --> 00:08:41,270
do you have with star S?

282
00:08:41,270 --> 00:08:42,590
Then in the visualization,

283
00:08:42,590 --> 00:08:45,425
there's also a
correspondingly two lines.

284
00:08:45,425 --> 00:08:49,310
Like what can you do
through S2 star S,

285
00:08:49,310 --> 00:08:50,330
we can mutate it because

286
00:08:50,330 --> 00:08:51,785
it's not immutable
resource.

287
00:08:51,785 --> 00:08:54,680
Whereas here we're talking

288
00:08:54,680 --> 00:08:56,090
about the
reference itself,

289
00:08:56,090 --> 00:08:57,470
so it's a shared,

290
00:08:57,470 --> 00:08:58,970
we can only read
data through it.

291
00:08:58,970 --> 00:09:00,080
Okay, so we start

292
00:09:00,080 --> 00:09:01,100
getting into a
little bit more of

293
00:09:01,100 --> 00:09:04,020
the subtleties related
with references here.

294
00:09:04,570 --> 00:09:07,310
Here we're talking about

295
00:09:07,310 --> 00:09:09,035
lifetime is a little bit,

296
00:09:09,035 --> 00:09:11,330
and in a very
visual way as well,

297
00:09:11,330 --> 00:09:12,620
you can see what happens

298
00:09:12,620 --> 00:09:14,440
with There's these borrows

299
00:09:14,440 --> 00:09:16,495
that happened to y and z.

300
00:09:16,495 --> 00:09:20,230
And then you pass Y
and Z over into f.

301
00:09:20,230 --> 00:09:22,585
And because there's
no further use of

302
00:09:22,585 --> 00:09:25,675
the reference, these
two references,

303
00:09:25,675 --> 00:09:27,700
they're kind of their
lifetime ends and

304
00:09:27,700 --> 00:09:30,145
we indicate that
with this return,

305
00:09:30,145 --> 00:09:31,630
this notional return of

306
00:09:31,630 --> 00:09:33,070
the borrowed resource back

307
00:09:33,070 --> 00:09:35,095
to the original owner.

308
00:09:35,095 --> 00:09:38,050
We put a little bit
of work into making

309
00:09:38,050 --> 00:09:39,430
this visualization
scale and

310
00:09:39,430 --> 00:09:40,450
these kind of
circumstances.

311
00:09:40,450 --> 00:09:42,100
You can see the line
is a little pink tier,

312
00:09:42,100 --> 00:09:43,300
and I'll talk a
little bit more about

313
00:09:43,300 --> 00:09:44,875
the scaling problem a
little later though.

314
00:09:44,875 --> 00:09:47,170
It's not infinitely
scalable.

315
00:09:47,170 --> 00:09:48,565
Okay?

316
00:09:48,565 --> 00:09:51,640
So I'll stop there
with the example.

317
00:09:51,640 --> 00:09:55,400
So that's the idea there.

318
00:09:55,710 --> 00:09:57,895
We're back to the slides.

319
00:09:57,895 --> 00:10:01,480
So is this useful, right?

320
00:10:01,480 --> 00:10:03,175
That's, I think an
important thing,

321
00:10:03,175 --> 00:10:04,360
an important
question to ask

322
00:10:04,360 --> 00:10:05,500
whenever you
develop a tool,

323
00:10:05,500 --> 00:10:06,610
is this actually
useful and

324
00:10:06,610 --> 00:10:07,945
people getting
anything out of this.

325
00:10:07,945 --> 00:10:11,320
So the way we assess
that is in my horse,

326
00:10:11,320 --> 00:10:13,630
which is upper level

327
00:10:13,630 --> 00:10:14,710
of programming
languages course

328
00:10:14,710 --> 00:10:16,270
each for 90 at Michigan.

329
00:10:16,270 --> 00:10:19,104
I've taught that
for four semesters.

330
00:10:19,104 --> 00:10:20,050
Collectively,

331
00:10:20,050 --> 00:10:22,450
313 students have
taken this course.

332
00:10:22,450 --> 00:10:24,250
So good sample.

333
00:10:24,250 --> 00:10:25,990
These are typically almost

334
00:10:25,990 --> 00:10:27,355
entirely juniors
and seniors

335
00:10:27,355 --> 00:10:28,795
undergraduates
with two semesters

336
00:10:28,795 --> 00:10:31,480
of our introductory C
plus, plus sequence.

337
00:10:31,480 --> 00:10:34,900
And it's in the second
half of my POL course.

338
00:10:34,900 --> 00:10:36,355
So the first half
of the course

339
00:10:36,355 --> 00:10:38,325
is taught in OCaml.

340
00:10:38,325 --> 00:10:40,070
There's one assignment on

341
00:10:40,070 --> 00:10:41,750
just intro to functional
programming, OCaml.

342
00:10:41,750 --> 00:10:42,860
And then there's
a bunch of PL

343
00:10:42,860 --> 00:10:44,330
prototyping that
you would expect

344
00:10:44,330 --> 00:10:47,240
from a PL course.

345
00:10:47,240 --> 00:10:48,620
So people
understand concepts

346
00:10:48,620 --> 00:10:49,550
like scope and they have

347
00:10:49,550 --> 00:10:50,480
kind of a more

348
00:10:50,480 --> 00:10:52,100
sophisticated maybe
understanding

349
00:10:52,100 --> 00:10:53,555
of PL concepts.

350
00:10:53,555 --> 00:10:55,925
But Rostow hasn't
been used yet.

351
00:10:55,925 --> 00:10:59,465
And our introduction to
rust is by schedule,

352
00:10:59,465 --> 00:11:00,935
limited to two lectures,

353
00:11:00,935 --> 00:11:03,080
one discussion,
all in one week,

354
00:11:03,080 --> 00:11:05,810
and then a two-week
long assignment,

355
00:11:05,810 --> 00:11:07,505
roughly two-week
long assignment.

356
00:11:07,505 --> 00:11:09,950
One component of
which is to read and

357
00:11:09,950 --> 00:11:11,270
go through that tutorial

358
00:11:11,270 --> 00:11:12,050
that I just showed you,

359
00:11:12,050 --> 00:11:13,460
which we tell
students should

360
00:11:13,460 --> 00:11:15,920
take between 3045 minutes.

361
00:11:15,920 --> 00:11:17,660
And then there's
some assessment

362
00:11:17,660 --> 00:11:19,070
that we'll talk
about in a second.

363
00:11:19,070 --> 00:11:20,795
So this is

364
00:11:20,795 --> 00:11:22,610
the particular
educational context

365
00:11:22,610 --> 00:11:23,270
that we're working in.

366
00:11:23,270 --> 00:11:25,190
I think it's an
interesting one

367
00:11:25,190 --> 00:11:27,200
because it's students that

368
00:11:27,200 --> 00:11:29,150
have some programming
experience in

369
00:11:29,150 --> 00:11:31,745
imperative languages,
like in C plus plus.

370
00:11:31,745 --> 00:11:33,620
So this is, I think

371
00:11:33,620 --> 00:11:35,300
a decent model of
actually even if

372
00:11:35,300 --> 00:11:36,530
you were trying
to push rust

373
00:11:36,530 --> 00:11:38,525
in an industrial setting,

374
00:11:38,525 --> 00:11:39,890
you're gonna get people
with programming

375
00:11:39,890 --> 00:11:41,360
experience that are
interested in C plus

376
00:11:41,360 --> 00:11:42,800
plus it has some
attachment to it

377
00:11:42,800 --> 00:11:44,840
because it's
performance and so on.

378
00:11:44,840 --> 00:11:46,760
And so we think
this is, this was

379
00:11:46,760 --> 00:11:47,810
a good audience to test

380
00:11:47,810 --> 00:11:50,100
this tool on for
the first time.

381
00:11:50,230 --> 00:11:52,460
So what we're
learning goals,

382
00:11:52,460 --> 00:11:54,560
well, there's kind of
two parts to this.

383
00:11:54,560 --> 00:11:56,120
One, ownership and
when borrowing.

384
00:11:56,120 --> 00:11:57,620
So we want first students

385
00:11:57,620 --> 00:11:59,300
to understand
how ownership

386
00:11:59,300 --> 00:12:01,130
is the foundation of
memory management and

387
00:12:01,130 --> 00:12:03,035
Rostow automatic memory
management and rust,

388
00:12:03,035 --> 00:12:03,710
when you talk about

389
00:12:03,710 --> 00:12:04,910
garbage collection
is kind of

390
00:12:04,910 --> 00:12:07,700
an alternative before
we get into rust.

391
00:12:07,700 --> 00:12:08,630
So there's

392
00:12:08,630 --> 00:12:11,000
some specific learning
goals here in

393
00:12:11,000 --> 00:12:12,350
terms of things that
we want students

394
00:12:12,350 --> 00:12:14,120
to be able to do.

395
00:12:14,120 --> 00:12:16,115
So how do we assess
these learning goals?

396
00:12:16,115 --> 00:12:19,310
So this is actually an
important question,

397
00:12:19,310 --> 00:12:20,705
I think one that our
community should,

398
00:12:20,705 --> 00:12:23,660
should think through
it systematically.

399
00:12:23,660 --> 00:12:25,190
The challenge
is you want to

400
00:12:25,190 --> 00:12:26,900
ask questions that assess

401
00:12:26,900 --> 00:12:28,520
their understanding
of these

402
00:12:28,520 --> 00:12:30,214
invisible concepts.

403
00:12:30,214 --> 00:12:31,040
And you want to do it in

404
00:12:31,040 --> 00:12:31,700
a way where they can't

405
00:12:31,700 --> 00:12:33,110
just like load
up rust when

406
00:12:33,110 --> 00:12:35,120
they're doing an
assignment and type

407
00:12:35,120 --> 00:12:36,200
in the code that we give

408
00:12:36,200 --> 00:12:37,700
them and they
get the answer.

409
00:12:37,700 --> 00:12:41,150
So it's not just is
this well typed, right?

410
00:12:41,150 --> 00:12:43,070
So how do we

411
00:12:43,070 --> 00:12:44,915
assess their understanding
of ownership?

412
00:12:44,915 --> 00:12:46,400
So here's the kind
of question that

413
00:12:46,400 --> 00:12:47,900
we came up with.

414
00:12:47,900 --> 00:12:48,380
What I think is

415
00:12:48,380 --> 00:12:50,180
a really interesting
class of questions.

416
00:12:50,180 --> 00:12:51,530
So what we do is
we give them a

417
00:12:51,530 --> 00:12:53,285
little snippet
of Rust code.

418
00:12:53,285 --> 00:12:54,920
Nothing particularly

419
00:12:54,920 --> 00:12:56,330
interesting is going
on in this code,

420
00:12:56,330 --> 00:12:57,830
but there's a bunch
of moves and borrows

421
00:12:57,830 --> 00:12:58,580
and things like that that

422
00:12:58,580 --> 00:12:59,765
might be going on in it.

423
00:12:59,765 --> 00:13:01,520
Then we ask, and you

424
00:13:01,520 --> 00:13:02,120
can notice there's also

425
00:13:02,120 --> 00:13:03,005
a bunch of print
lines in here.

426
00:13:03,005 --> 00:13:03,860
So we asked between which

427
00:13:03,860 --> 00:13:04,790
two adjacent lines of

428
00:13:04,790 --> 00:13:06,530
this printed output is

429
00:13:06,530 --> 00:13:07,670
the string resource that a

430
00:13:07,670 --> 00:13:09,455
initially owns dropped.

431
00:13:09,455 --> 00:13:11,525
And then explain
why by describing

432
00:13:11,525 --> 00:13:13,700
what happens with
movies and borrows.

433
00:13:13,700 --> 00:13:15,260
This question
isn't one that you

434
00:13:15,260 --> 00:13:16,460
can easily
answered by just

435
00:13:16,460 --> 00:13:17,780
typing this code and

436
00:13:17,780 --> 00:13:19,310
getting feedback from
the Rust compiler.

437
00:13:19,310 --> 00:13:21,320
Rust compiler will
successfully execute this,

438
00:13:21,320 --> 00:13:22,790
but there are no
visible side effect

439
00:13:22,790 --> 00:13:23,810
for the drop, right?

440
00:13:23,810 --> 00:13:26,374
That's just when the
string is deallocated.

441
00:13:26,374 --> 00:13:28,010
So you do have to actually

442
00:13:28,010 --> 00:13:29,210
build that mental model

443
00:13:29,210 --> 00:13:30,260
that I'm talking
about, right?

444
00:13:30,260 --> 00:13:31,325
That we're trying to teach

445
00:13:31,325 --> 00:13:33,740
and model all
the moves and

446
00:13:33,740 --> 00:13:35,510
see when variable that

447
00:13:35,510 --> 00:13:37,385
owns that resource
goes out of scope.

448
00:13:37,385 --> 00:13:39,110
That's, that's the,
that's the goal that they

449
00:13:39,110 --> 00:13:40,100
understand that
and they can

450
00:13:40,100 --> 00:13:42,210
figure out that
the answer,

451
00:13:42,730 --> 00:13:45,960
in this case
is, what is it?

452
00:13:47,170 --> 00:13:49,160
After?

453
00:13:49,160 --> 00:13:51,440
Hello and

454
00:13:51,440 --> 00:13:53,030
goodbye, I forget
what the answer is.

455
00:13:53,030 --> 00:13:54,635
So I think that's it.

456
00:13:54,635 --> 00:13:58,100
We give them 678 questions

457
00:13:58,100 --> 00:14:00,725
like this, all
relatively small.

458
00:14:00,725 --> 00:14:03,260
And if people can do this,

459
00:14:03,260 --> 00:14:04,490
that means they're
starting to develop

460
00:14:04,490 --> 00:14:05,240
an understanding of

461
00:14:05,240 --> 00:14:06,410
this invisible
thing that's

462
00:14:06,410 --> 00:14:07,850
going on with
ownership and moves

463
00:14:07,850 --> 00:14:09,745
and we see some
success with it.

464
00:14:09,745 --> 00:14:10,820
So the median score on

465
00:14:10,820 --> 00:14:12,635
this kind of
question is 91%.

466
00:14:12,635 --> 00:14:13,580
We also have
questions like

467
00:14:13,580 --> 00:14:14,765
this on our final exam.

468
00:14:14,765 --> 00:14:17,720
Also, very people do

469
00:14:17,720 --> 00:14:19,670
very well on that
question as well.

470
00:14:19,670 --> 00:14:21,545
Then we also are able to,

471
00:14:21,545 --> 00:14:22,460
the way we have it setup,

472
00:14:22,460 --> 00:14:23,660
we are able to measure how

473
00:14:23,660 --> 00:14:25,040
long it takes
students to answer

474
00:14:25,040 --> 00:14:25,790
this question when it is

475
00:14:25,790 --> 00:14:26,810
from the time
it's presented

476
00:14:26,810 --> 00:14:29,000
to them to when they
submit an answer.

477
00:14:29,000 --> 00:14:30,905
And that's about
five to six minutes

478
00:14:30,905 --> 00:14:32,420
on average, right?

479
00:14:32,420 --> 00:14:35,495
So what that means
is that students are

480
00:14:35,495 --> 00:14:37,280
kinda like thinking

481
00:14:37,280 --> 00:14:38,990
through each
of the events.

482
00:14:38,990 --> 00:14:41,645
They're not just kinda
knee jerk answering.

483
00:14:41,645 --> 00:14:45,560
It's done. A quick
assessment of the code.

484
00:14:45,560 --> 00:14:47,900
We think this
is promising.

485
00:14:47,900 --> 00:14:50,750
After a week or two of
instruction in rust,

486
00:14:50,750 --> 00:14:52,130
Students are able
to understand

487
00:14:52,130 --> 00:14:53,450
the automatic
memory management

488
00:14:53,450 --> 00:14:55,265
where the resources
are being dropped.

489
00:14:55,265 --> 00:14:57,950
As a consequence of
these move events.

490
00:14:57,950 --> 00:15:00,380
I can set of
learning goals

491
00:15:00,380 --> 00:15:01,955
have to do with borrowing.

492
00:15:01,955 --> 00:15:04,040
So again, they
have to understand

493
00:15:04,040 --> 00:15:05,555
immutable and
mutable borrows.

494
00:15:05,555 --> 00:15:07,670
They understand
how lifetime

495
00:15:07,670 --> 00:15:09,920
and borrow relates to
the lifetime owner.

496
00:15:09,920 --> 00:15:11,720
Then also we want
them to understand

497
00:15:11,720 --> 00:15:13,280
the basics of non
lexical lifetimes.

498
00:15:13,280 --> 00:15:16,715
And it's actually
a challenge

499
00:15:16,715 --> 00:15:18,380
relatively to assess
understanding

500
00:15:18,380 --> 00:15:20,150
of borrowing relative
to ownership,

501
00:15:20,150 --> 00:15:21,680
because borrowing is
really more about

502
00:15:21,680 --> 00:15:23,450
restrictions on
what you can't do.

503
00:15:23,450 --> 00:15:25,250
And so we can ask,

504
00:15:25,250 --> 00:15:26,990
we do ask questions
of the form on

505
00:15:26,990 --> 00:15:28,730
which line of code does
the lifetime of x.

506
00:15:28,730 --> 00:15:32,150
And that's essentially
the return arrow.

507
00:15:32,150 --> 00:15:34,325
Like where should
the return arrow go?

508
00:15:34,325 --> 00:15:36,860
People do a little
bit worse on this.

509
00:15:36,860 --> 00:15:40,430
But it's about 80% or

510
00:15:40,430 --> 00:15:41,600
so is the medians

511
00:15:41,600 --> 00:15:44,040
colon questions
of this sort.

512
00:15:44,110 --> 00:15:46,820
Discussion topic that
I think I'd like to

513
00:15:46,820 --> 00:15:47,870
get more feedback from

514
00:15:47,870 --> 00:15:48,930
the community
on is that is,

515
00:15:48,930 --> 00:15:50,945
there, are there better
ways of assessing

516
00:15:50,945 --> 00:15:54,185
borrowing and understanding
of borrowing?

517
00:15:54,185 --> 00:15:55,955
Other than there's also,

518
00:15:55,955 --> 00:15:56,690
I mean, you could
just have them

519
00:15:56,690 --> 00:15:58,460
write code like the
borrower checker,

520
00:15:58,460 --> 00:16:01,430
but we want a more
targeted assessment.

521
00:16:01,430 --> 00:16:03,350
And then one additional

522
00:16:03,350 --> 00:16:04,100
thing that we
didn't want to be

523
00:16:04,100 --> 00:16:04,970
semesters where we did

524
00:16:04,970 --> 00:16:06,410
an extra credit
question where

525
00:16:06,410 --> 00:16:07,745
we ask students
to actually

526
00:16:07,745 --> 00:16:09,050
generate their own rust,

527
00:16:09,050 --> 00:16:10,670
this visualization,
this is also

528
00:16:10,670 --> 00:16:11,720
a nice segue
for me to talk

529
00:16:11,720 --> 00:16:13,070
about how that is done.

530
00:16:13,070 --> 00:16:14,630
It is not fully automatic.

531
00:16:14,630 --> 00:16:16,115
What we've done is
we've developed

532
00:16:16,115 --> 00:16:18,170
a domain-specific
language that's basically

533
00:16:18,170 --> 00:16:19,430
just a comment system

534
00:16:19,430 --> 00:16:21,980
in a formatted comments
system in Rust.

535
00:16:21,980 --> 00:16:22,520
So you just take

536
00:16:22,520 --> 00:16:23,750
your rescue that you
want to visualize.

537
00:16:23,750 --> 00:16:24,560
You annotate it with

538
00:16:24,560 --> 00:16:26,090
comments that
looked like this.

539
00:16:26,090 --> 00:16:27,080
You pass it through

540
00:16:27,080 --> 00:16:28,430
our recipes visualizer
and you get

541
00:16:28,430 --> 00:16:30,845
this kind of visualization
that you saw.

542
00:16:30,845 --> 00:16:31,400
So you can see

543
00:16:31,400 --> 00:16:33,319
there's a column
specification.

544
00:16:33,319 --> 00:16:34,580
What columns do you
want to visualize?

545
00:16:34,580 --> 00:16:35,750
You might not want
to visualize every

546
00:16:35,750 --> 00:16:37,340
variable and
every function.

547
00:16:37,340 --> 00:16:39,260
Then on each line you say

548
00:16:39,260 --> 00:16:41,165
what events happened
on that line.

549
00:16:41,165 --> 00:16:42,650
And this little
comment here.

550
00:16:42,650 --> 00:16:44,420
And we have this
kind of table saying

551
00:16:44,420 --> 00:16:46,475
what were all the
possibilities are.

552
00:16:46,475 --> 00:16:48,500
One thing I'll note
here is that this is

553
00:16:48,500 --> 00:16:50,750
a notional machine of
how Rust works, right?

554
00:16:50,750 --> 00:16:52,220
So it's not exactly

555
00:16:52,220 --> 00:16:54,575
a specification of the
Rus borrow checker.

556
00:16:54,575 --> 00:16:56,630
It's something that

557
00:16:56,630 --> 00:16:58,160
corresponds
roughly to what

558
00:16:58,160 --> 00:16:59,555
an introductory students

559
00:16:59,555 --> 00:17:00,935
understanding borrowing

560
00:17:00,935 --> 00:17:04,340
and moves would develop.

561
00:17:04,340 --> 00:17:06,800
And how do people

562
00:17:06,800 --> 00:17:08,959
do on this median
score is 100%.

563
00:17:08,959 --> 00:17:10,430
This was an extra
credit question.

564
00:17:10,430 --> 00:17:11,750
So people
essentially didn't

565
00:17:11,750 --> 00:17:13,190
submit it if they
didn't get it right.

566
00:17:13,190 --> 00:17:15,200
So that's tears were
at about the time.

567
00:17:15,200 --> 00:17:15,740
So we should wrap up.

568
00:17:15,740 --> 00:17:17,870
So if i'm I'm
wrapping up here.

569
00:17:17,870 --> 00:17:19,100
So I went to one
of the three,

570
00:17:19,100 --> 00:17:20,525
so it took about
two hours.

571
00:17:20,525 --> 00:17:21,740
Most of that was
actually set up

572
00:17:21,740 --> 00:17:22,880
a virus on their
own machine

573
00:17:22,880 --> 00:17:23,840
and things like that.

574
00:17:23,840 --> 00:17:25,760
But interestingly,
60% thought

575
00:17:25,760 --> 00:17:26,975
that this exercise
help them

576
00:17:26,975 --> 00:17:28,700
better understand Ross

577
00:17:28,700 --> 00:17:30,590
overall did
receive his help.

578
00:17:30,590 --> 00:17:31,880
Ninety-five percent
of students

579
00:17:31,880 --> 00:17:32,600
said it was either help,

580
00:17:32,600 --> 00:17:33,620
the visualization
was either

581
00:17:33,620 --> 00:17:34,940
helpful or very helpful.

582
00:17:34,940 --> 00:17:36,740
75% of students reported

583
00:17:36,740 --> 00:17:37,820
frequently interacting

584
00:17:37,820 --> 00:17:39,470
with the
visualization study.

585
00:17:39,470 --> 00:17:40,940
In our study,
we also did a

586
00:17:40,940 --> 00:17:43,145
small think-aloud
similar results.

587
00:17:43,145 --> 00:17:45,140
Then the feedback
was basically,

588
00:17:45,140 --> 00:17:46,730
I think these
things are helpful.

589
00:17:46,730 --> 00:17:48,830
I still have to think
through the problems.

590
00:17:48,830 --> 00:17:50,300
And then there were
some feedback about

591
00:17:50,300 --> 00:17:52,520
the actual mechanics of

592
00:17:52,520 --> 00:17:54,260
interaction that
I'll skip for now.

593
00:17:54,260 --> 00:17:56,045
So to conclude,

594
00:17:56,045 --> 00:17:58,115
we think this is a
promising approach.

595
00:17:58,115 --> 00:17:59,090
What are our next steps?

596
00:17:59,090 --> 00:18:01,340
One, automatic
visualization generation.

597
00:18:01,340 --> 00:18:02,630
There's some
challenges there with

598
00:18:02,630 --> 00:18:03,830
customizability and

599
00:18:03,830 --> 00:18:06,080
the scaling of
these things too,

600
00:18:06,080 --> 00:18:08,060
we want to explore

601
00:18:08,060 --> 00:18:09,500
alternative
visualization style

602
00:18:09,500 --> 00:18:10,790
for automatic
visualization.

603
00:18:10,790 --> 00:18:12,800
Can you do it without
a big sidebar?

604
00:18:12,800 --> 00:18:13,730
And then finally,

605
00:18:13,730 --> 00:18:15,455
integration into
more courses.

606
00:18:15,455 --> 00:18:18,020
Public release of
self-study tutorials

607
00:18:18,020 --> 00:18:19,040
and smaller learning

608
00:18:19,040 --> 00:18:20,555
modules for
different topics.

609
00:18:20,555 --> 00:18:22,280
If you are
interested in that,

610
00:18:22,280 --> 00:18:23,510
if you have a course
or you want to

611
00:18:23,510 --> 00:18:25,280
develop material
contact me.

612
00:18:25,280 --> 00:18:27,290
You can play
with Material at

613
00:18:27,290 --> 00:18:29,900
that URL and
I'll stop there.

614
00:18:29,900 --> 00:18:34,670
Thank you. Alright,
Thanks Cyrus.

615
00:18:34,670 --> 00:18:36,680
Fabulous talk. We'll
jump right into

616
00:18:36,680 --> 00:18:37,730
some questions
and then head

617
00:18:37,730 --> 00:18:39,290
into the big
group discussion.

618
00:18:39,290 --> 00:18:40,670
So following up on

619
00:18:40,670 --> 00:18:42,320
your point about
automatic generalization,

620
00:18:42,320 --> 00:18:44,510
Matthew did ask
how close to

621
00:18:44,510 --> 00:18:44,960
generating

622
00:18:44,960 --> 00:18:46,730
the visualizations
automatically RVs.

623
00:18:46,730 --> 00:18:47,900
And so it would be really

624
00:18:47,900 --> 00:18:48,980
critical for playing with

625
00:18:48,980 --> 00:18:50,090
code as we talked about in

626
00:18:50,090 --> 00:18:52,205
the last session or
in the last time.

627
00:18:52,205 --> 00:18:53,660
There's a bit
of automation.

628
00:18:53,660 --> 00:18:55,010
You can generate
the column headings

629
00:18:55,010 --> 00:18:56,660
automatically by
just parsing out

630
00:18:56,660 --> 00:18:58,430
the variables that are

631
00:18:58,430 --> 00:18:59,915
used as the initial state.

632
00:18:59,915 --> 00:19:00,860
And then you'll
see actually

633
00:19:00,860 --> 00:19:01,940
if you're looking at
this table down here,

634
00:19:01,940 --> 00:19:03,575
there's this go out
of scope of it,

635
00:19:03,575 --> 00:19:05,300
which can be visualized in

636
00:19:05,300 --> 00:19:07,100
three different ways
depending on if

637
00:19:07,100 --> 00:19:10,700
that variable x owns a
resource at that time.

638
00:19:10,700 --> 00:19:12,440
So we do track that
automatically.

639
00:19:12,440 --> 00:19:13,985
You don't have to
decide if it goes,

640
00:19:13,985 --> 00:19:15,260
if the resource is
dropped or not.

641
00:19:15,260 --> 00:19:16,715
The system does
that automatically.

642
00:19:16,715 --> 00:19:17,960
So we've taken baby steps

643
00:19:17,960 --> 00:19:19,535
towards automating it,

644
00:19:19,535 --> 00:19:22,040
but it's a bit
of a challenge

645
00:19:22,040 --> 00:19:24,650
to actually support
the entire system,

646
00:19:24,650 --> 00:19:26,870
the entire borrow
Checker of rust.

647
00:19:26,870 --> 00:19:28,370
We don't attempts
to do that.

648
00:19:28,370 --> 00:19:29,210
We want attempt to do

649
00:19:29,210 --> 00:19:30,530
a certain
advanced features

650
00:19:30,530 --> 00:19:32,330
that are described
in our paper.

651
00:19:32,330 --> 00:19:34,070
I think this is really
interesting question.

652
00:19:34,070 --> 00:19:35,180
If there's

653
00:19:35,180 --> 00:19:36,410
graduate students
out there that are

654
00:19:36,410 --> 00:19:37,460
interested in
this topic is

655
00:19:37,460 --> 00:19:39,215
definitely a topic
I want to explore.

656
00:19:39,215 --> 00:19:41,270
But I think there's

657
00:19:41,270 --> 00:19:42,740
some benefits to the
manual generation,

658
00:19:42,740 --> 00:19:45,080
both for students
doing a question like

659
00:19:45,080 --> 00:19:46,790
this and for instructors

660
00:19:46,790 --> 00:19:48,575
being able to
customize it.

661
00:19:48,575 --> 00:19:50,270
And we think it doesn't

662
00:19:50,270 --> 00:19:51,530
take that long
to generate it

663
00:19:51,530 --> 00:19:53,450
manually for any
particular example

664
00:19:53,450 --> 00:19:54,770
and we're working
on improving that.

665
00:19:54,770 --> 00:19:58,740
So that's where we're at.

666
00:19:59,710 --> 00:20:03,360
Any other questions
in the chat.

667
00:20:09,730 --> 00:20:13,115
One thing I'll
just toss in is,

668
00:20:13,115 --> 00:20:15,410
I think one metaphor
I find really

669
00:20:15,410 --> 00:20:17,060
interesting here
is thinking

670
00:20:17,060 --> 00:20:18,455
about this tool
kind of like dot,

671
00:20:18,455 --> 00:20:19,850
you know, the, the
sort of format you

672
00:20:19,850 --> 00:20:21,695
use to describe
graphs with this.

673
00:20:21,695 --> 00:20:23,495
This is sort of like, like

674
00:20:23,495 --> 00:20:26,000
a language of program
comprehension or

675
00:20:26,000 --> 00:20:27,980
visualizing static
semantics of programs is

676
00:20:27,980 --> 00:20:29,135
something that
there hasn't,

677
00:20:29,135 --> 00:20:30,740
doesn't really exist.

678
00:20:30,740 --> 00:20:32,885
But I think if you could
sort of like, like,

679
00:20:32,885 --> 00:20:34,430
I really like how
it's extracted

680
00:20:34,430 --> 00:20:36,140
out as this visual
language that you can

681
00:20:36,140 --> 00:20:37,280
generate however
you want and

682
00:20:37,280 --> 00:20:38,690
theoretically adopt
for use cases

683
00:20:38,690 --> 00:20:40,160
even beyond ownership

684
00:20:40,160 --> 00:20:42,035
potentially or
visualizing ownership.

685
00:20:42,035 --> 00:20:44,165
I think it'd be
very cool if

686
00:20:44,165 --> 00:20:46,715
it was applicable to
this sort of settings.

687
00:20:46,715 --> 00:20:48,950
Like if you had any
abstract interpretation

688
00:20:48,950 --> 00:20:49,640
of your program, if

689
00:20:49,640 --> 00:20:50,705
you could play
it that way.

690
00:20:50,705 --> 00:20:51,755
Awesome.

691
00:20:51,755 --> 00:20:52,640
We've been thinking

692
00:20:52,640 --> 00:20:53,960
about like certain
libraries,

693
00:20:53,960 --> 00:20:55,040
like higher-level
library is

694
00:20:55,040 --> 00:20:56,420
related to concurrency
and things like

695
00:20:56,420 --> 00:20:57,950
that that have
certain mental models

696
00:20:57,950 --> 00:20:59,495
like mutexes and
things like that,

697
00:20:59,495 --> 00:21:02,525
that you might want to ask

698
00:21:02,525 --> 00:21:04,370
people to
visualize it using

699
00:21:04,370 --> 00:21:06,125
some slightly different
visual language.

700
00:21:06,125 --> 00:21:08,390
And I think this
approach overall,

701
00:21:08,390 --> 00:21:10,820
it's kind of an
interesting direction

702
00:21:10,820 --> 00:21:12,450
to take with that.