language-creators.vtt

WEBVTT

NOTE
# Named participants, in order of appearance:
# RUTHE: Ruthe Farmer (MC)
# CAROL: Carol Willing (panel host)
# JAMES: James Gosling, creator of Java
# GUIDO: Guido van Rossum, creator of Python
# LARRY: Larry Wall, creator of Perl
# ANDERS: Anders Hejlsberg, creator of Turbo Pascal, Delphi, C#, and TypeScript


1
00:00:00.530 --> 00:00:01.729
You guys ready some language

2
00:00:01.730 --> 00:00:02.730
creator.

3
00:00:07.700 --> 00:00:09.669
I have further to introduce

4
00:00:09.940 --> 00:00:10.940
Carol Willey.

5
00:00:11.420 --> 00:00:13.089
She is a python during counseling

6
00:00:13.090 --> 00:00:14.589
Project Jupiter Steering Council

7
00:00:14.590 --> 00:00:15.590
member.

8
00:00:15.610 --> 00:00:17.129
She blends the strings of pythonic.

9
00:00:17.140 --> 00:00:18.759
Jupiter notebook's to improve access

10
00:00:18.760 --> 00:00:19.929
to learning and education

11
00:00:20.730 --> 00:00:22.179
due to the worldwide foundation of

12
00:00:22.180 --> 00:00:22.989
the Pyfrom language.

13
00:00:22.990 --> 00:00:24.339
The hard work of the Jupiter team

14
00:00:24.410 --> 00:00:25.410
users worldwide.

15
00:00:25.570 --> 00:00:26.679
Carroll was awarded the Twenty

16
00:00:26.680 --> 00:00:28.659
Seventeen ACM Software System

17
00:00:28.660 --> 00:00:30.130
Award, recognizing the lasting

18
00:00:30.850 --> 00:00:32.228
influence of Jupiter, a broad

19
00:00:32.229 --> 00:00:34.149
collaboration to develop its open

20
00:00:34.150 --> 00:00:35.522
source tools for interactive

21
00:00:35.560 --> 00:00:37.226
computing with a language agnostic

22
00:00:37.840 --> 00:00:38.619
design.

23
00:00:38.620 --> 00:00:39.620
Thank you.

24
00:00:47.800 --> 00:00:49.630
Welcome, everybody, for being here.

25
00:00:49.720 --> 00:00:51.639
It is my absolute

26
00:00:51.850 --> 00:00:54.009
pleasure to share this evening.

27
00:00:54.040 --> 00:00:55.040
You and some

28
00:00:55.870 --> 00:00:57.249
of the people that actually

29
00:00:57.250 --> 00:00:59.289
influenced my career in software

30
00:00:59.290 --> 00:01:00.290
development.

31
00:01:01.180 --> 00:01:03.010
So let's get through this.

32
00:01:03.220 --> 00:01:04.641
And for those of you that are

33
00:01:04.810 --> 00:01:06.099
watching us on the lifespan.

34
00:01:06.520 --> 00:01:08.590
Hello and welcome and

35
00:01:08.770 --> 00:01:10.419
hope you have a wonderful evening

36
00:01:10.420 --> 00:01:11.420
with us.

37
00:01:12.160 --> 00:01:13.659
We haven't really distinguished

38
00:01:13.660 --> 00:01:15.969
panel of language creators.

39
00:01:16.060 --> 00:01:17.529
And I'd like to invite them to the

40
00:01:17.530 --> 00:01:19.359
stage as we read

41
00:01:19.360 --> 00:01:20.409
their bios. I'm going to keep the

42
00:01:20.410 --> 00:01:22.689
bios really short because

43
00:01:22.810 --> 00:01:24.519
I think most of them need no

44
00:01:24.520 --> 00:01:25.969
introduction whatsoever.

45
00:01:26.710 --> 00:01:27.886
And that'll save us some

46
00:01:28.540 --> 00:01:29.540
time more.

47
00:01:29.820 --> 00:01:31.094
Do you see language design

48
00:01:31.840 --> 00:01:32.840
stuff? OK.

49
00:01:33.190 --> 00:01:35.159
So first we have been

50
00:01:35.190 --> 00:01:36.190
Rassam

51
00:01:37.030 --> 00:01:39.009
to develop the Python language that

52
00:01:39.010 --> 00:01:40.235
touches pretty much every

53
00:01:40.870 --> 00:01:42.389
corner of society today in some

54
00:01:43.060 --> 00:01:43.949
way.

55
00:01:43.950 --> 00:01:45.819
And it's.

56
00:01:51.540 --> 00:01:53.510
He's the creator of Java,

57
00:01:53.570 --> 00:01:54.991
which is UK-EU, used across a

58
00:01:55.400 --> 00:01:57.349
broad variety of devices and

59
00:01:57.350 --> 00:01:58.700
deployments at scale.

60
00:01:59.590 --> 00:02:00.239
Well done.

61
00:02:00.240 --> 00:02:01.540
And please get a.

62
00:02:07.740 --> 00:02:09.719
Childcare has architected

63
00:02:09.720 --> 00:02:11.435
a number of languages over the past

64
00:02:11.520 --> 00:02:13.650
several decades, including

65
00:02:13.710 --> 00:02:16.079
Turbo Pascal, Delphi,

66
00:02:16.170 --> 00:02:17.885
C-sharp, and so you might know this

67
00:02:18.120 --> 00:02:19.469
little language typescript.

68
00:02:26.580 --> 00:02:27.580
Definitely not these.

69
00:02:27.700 --> 00:02:29.790
Larry Wall to combined

70
00:02:29.800 --> 00:02:31.025
his unique perspective on

71
00:02:31.750 --> 00:02:33.699
text and computing with

72
00:02:33.700 --> 00:02:35.589
his strong background in linguistics

73
00:02:35.620 --> 00:02:37.449
to Craig Pearl and its

74
00:02:37.450 --> 00:02:39.469
sister sibling Pearl 6.

75
00:02:39.820 --> 00:02:40.820
Welcome.

76
00:02:51.420 --> 00:02:52.420
So just a couple of

77
00:02:53.280 --> 00:02:55.139
housekeeping things about this

78
00:02:55.140 --> 00:02:56.140
evening's format.

79
00:02:57.000 --> 00:02:58.740
You're in for an engaging

80
00:02:58.950 --> 00:03:00.539
and hopefully very lively

81
00:03:00.540 --> 00:03:02.279
discussion. More likely than we were

82
00:03:02.280 --> 00:03:03.280
at dinner.

83
00:03:03.300 --> 00:03:04.440
I think that's possible, right?

84
00:03:04.910 --> 00:03:05.910
OK.

85
00:03:05.940 --> 00:03:06.769
The format.

86
00:03:06.770 --> 00:03:07.770
More beer.

87
00:03:08.870 --> 00:03:10.379
OK. We can do that.

88
00:03:11.280 --> 00:03:12.280
Keep it coming.

89
00:03:13.830 --> 00:03:15.055
OK. So the format will be

90
00:03:15.750 --> 00:03:17.729
two sessions with a fifteen minute

91
00:03:18.300 --> 00:03:19.300
intermission.

92
00:03:19.740 --> 00:03:21.539
What I'll do is I will start by

93
00:03:21.540 --> 00:03:23.550
asking one of the panelists a

94
00:03:23.580 --> 00:03:25.429
question, and hopefully

95
00:03:25.430 --> 00:03:27.089
a response will be like one to two

96
00:03:27.090 --> 00:03:29.009
minutes and then kind of throw

97
00:03:29.010 --> 00:03:30.480
it open for other panelists to

98
00:03:30.660 --> 00:03:31.885
respond and hopefully get

99
00:03:32.610 --> 00:03:34.169
a really good discussion going.

100
00:03:34.500 --> 00:03:35.921
And the more you discuss, the

101
00:03:36.360 --> 00:03:37.928
less I have to ask questions and

102
00:03:38.220 --> 00:03:40.349
the more in-depth we can go

103
00:03:40.410 --> 00:03:41.410
on the really cool stuff

104
00:03:43.620 --> 00:03:45.710
after the 15 minute intermission.

105
00:03:46.290 --> 00:03:47.290
We're going to take

106
00:03:48.120 --> 00:03:50.459
questions from the audience over

107
00:03:52.140 --> 00:03:53.789
a winner

108
00:03:54.210 --> 00:03:54.839
Twitter.

109
00:03:54.840 --> 00:03:56.789
OK. So tweet out your questions.

110
00:03:57.090 --> 00:03:59.390
Hashtag is happy bbf

111
00:03:59.400 --> 00:04:00.400
out.

112
00:04:00.570 --> 00:04:01.746
And if you can get those

113
00:04:02.580 --> 00:04:04.349
in by 8:00, we'll kind of collect

114
00:04:04.350 --> 00:04:05.399
them and

115
00:04:06.870 --> 00:04:08.339
have some really great audience

116
00:04:08.340 --> 00:04:09.779
questions about language creation

117
00:04:09.780 --> 00:04:10.780
and language design.

118
00:04:11.280 --> 00:04:13.439
So I'm going to

119
00:04:13.500 --> 00:04:15.629
invite you to sit back,

120
00:04:16.260 --> 00:04:18.299
enjoy a cup of strong Java

121
00:04:19.070 --> 00:04:20.910
Sea, start offering responses and

122
00:04:20.940 --> 00:04:23.100
insights, learned pearls of wisdom

123
00:04:23.670 --> 00:04:25.389
and embrace. The Python program

124
00:04:25.400 --> 00:04:26.400
makes it.

125
00:04:33.190 --> 00:04:34.856
We'll start the celebration with a

126
00:04:34.900 --> 00:04:36.566
question for Kito about principles

127
00:04:36.990 --> 00:04:38.019
of language design.

128
00:04:39.200 --> 00:04:40.621
You mentioned before, perhaps

129
00:04:41.170 --> 00:04:43.089
somewhat jokingly, the Harry

130
00:04:43.090 --> 00:04:44.829
Potter theory of language design.

131
00:04:45.490 --> 00:04:47.205
What is this theory and what do you

132
00:04:47.380 --> 00:04:48.729
see as the key principles of

133
00:04:48.730 --> 00:04:49.730
language design?

134
00:04:57.520 --> 00:04:58.520
I think you're good to me.

135
00:04:58.610 --> 00:04:59.610
Am I good?

136
00:04:59.620 --> 00:05:00.620
OK.

137
00:05:01.800 --> 00:05:03.515
The Harry Potter theory of language

138
00:05:03.720 --> 00:05:05.149
design was close to.

139
00:05:08.400 --> 00:05:10.709
I do know that I cannot believe

140
00:05:10.710 --> 00:05:11.710
that every detail

141
00:05:12.740 --> 00:05:14.455
that J.K. Rowling puts on the first

142
00:05:14.790 --> 00:05:16.559
Harry Potter book was

143
00:05:16.620 --> 00:05:18.480
intended as.

144
00:05:20.530 --> 00:05:22.539
Some important

145
00:05:22.860 --> 00:05:24.100
Wolf point game

146
00:05:25.020 --> 00:05:26.550
part 6 or 7.

147
00:05:29.350 --> 00:05:31.239
And even if J.K.

148
00:05:31.240 --> 00:05:33.129
Rowling was such a genius that

149
00:05:33.130 --> 00:05:34.999
he hadn't actually owned land, that

150
00:05:35.180 --> 00:05:36.879
and that was

151
00:05:37.410 --> 00:05:39.439
who I think we found out the

152
00:05:39.670 --> 00:05:40.670
FUTHI was.

153
00:05:44.460 --> 00:05:45.460
But who won?

154
00:05:45.610 --> 00:05:46.982
It was just like a temporary

155
00:05:47.470 --> 00:05:49.149
diversion during the train ride.

156
00:05:49.200 --> 00:05:50.679
And otherwise, you didn't hear much

157
00:05:50.680 --> 00:05:52.839
about the rap with fingerprints

158
00:05:54.330 --> 00:05:55.099
in languages.

159
00:05:55.100 --> 00:05:56.989
I often.

160
00:05:57.250 --> 00:05:59.079
That's exactly how things go.

161
00:05:59.110 --> 00:06:00.110
You

162
00:06:01.260 --> 00:06:03.129
you choose some detail

163
00:06:03.760 --> 00:06:05.439
because you have to commit.

164
00:06:05.510 --> 00:06:06.510
You sort of

165
00:06:07.720 --> 00:06:09.509
you have to pick keywords.

166
00:06:09.970 --> 00:06:11.910
You have to choose a style of

167
00:06:12.640 --> 00:06:14.559
coding like you maybe

168
00:06:14.560 --> 00:06:16.749
you choose indentation

169
00:06:16.780 --> 00:06:18.340
instead of curly braces

170
00:06:18.720 --> 00:06:19.720
or maybe you don't.

171
00:06:22.220 --> 00:06:23.339
You're a special case,

172
00:06:25.330 --> 00:06:26.330
but

173
00:06:27.570 --> 00:06:28.620
whatever you do

174
00:06:29.670 --> 00:06:31.110
in some way, you're stuck with them

175
00:06:32.550 --> 00:06:34.499
and you find new

176
00:06:34.500 --> 00:06:36.070
uses of that.

177
00:06:36.950 --> 00:06:38.770
That detail that you picked

178
00:06:38.790 --> 00:06:40.809
before you knew how important

179
00:06:40.810 --> 00:06:42.829
that would be and

180
00:06:43.100 --> 00:06:45.509
the sort of the craft of designing

181
00:06:45.560 --> 00:06:47.620
the language is on one hand pick

182
00:06:47.640 --> 00:06:49.379
your initial set of

183
00:06:50.250 --> 00:06:51.690
choices so that

184
00:06:52.350 --> 00:06:54.269
there are a lot

185
00:06:54.330 --> 00:06:55.555
of possible continuations

186
00:06:56.520 --> 00:06:57.520
of the story.

187
00:06:57.990 --> 00:06:59.490
The other half of the

188
00:07:00.600 --> 00:07:02.399
of the art of language design

189
00:07:02.430 --> 00:07:04.229
is going back to your story

190
00:07:04.260 --> 00:07:05.260
and

191
00:07:06.420 --> 00:07:08.579
inventing creative ways

192
00:07:08.670 --> 00:07:09.670
of continuing

193
00:07:10.590 --> 00:07:12.600
it in a way that you had not.

194
00:07:15.420 --> 00:07:16.743
So I think each of you have

195
00:07:16.800 --> 00:07:18.959
languages that span multiple

196
00:07:18.960 --> 00:07:19.709
decades.

197
00:07:19.710 --> 00:07:21.479
So what are some of the principles

198
00:07:21.480 --> 00:07:22.480
that made,

199
00:07:24.090 --> 00:07:25.949
you know, so so

200
00:07:26.670 --> 00:07:28.649
Javal would sort of on in the.

201
00:07:30.000 --> 00:07:31.279
It didn't come out of like a

202
00:07:31.290 --> 00:07:32.613
personal passion project or

203
00:07:33.060 --> 00:07:34.079
something. It was.

204
00:07:36.870 --> 00:07:39.029
From trying to build a prototype,

205
00:07:39.810 --> 00:07:41.279
you know, I mean, we were trying to

206
00:07:41.280 --> 00:07:42.829
understand it because the.

207
00:07:48.840 --> 00:07:50.359
A lot of time talking to people

208
00:07:50.670 --> 00:07:52.139
who build software or embedded

209
00:07:52.140 --> 00:07:53.140
systems.

210
00:07:59.800 --> 00:08:01.679
And you know, what was

211
00:08:01.680 --> 00:08:03.599
it in the whole process?

212
00:08:10.920 --> 00:08:12.870
How did their systems

213
00:08:12.890 --> 00:08:14.179
fit into the universe?

214
00:08:14.390 --> 00:08:15.860
It wasn't because I was trying

215
00:08:16.220 --> 00:08:17.984
to worry when we were worrying about

216
00:08:18.080 --> 00:08:19.560
things outside of data centers.

217
00:08:22.920 --> 00:08:24.537
This was a project that had about

218
00:08:24.750 --> 00:08:26.980
a dozen people on it and

219
00:08:26.990 --> 00:08:27.990
my little.

220
00:08:28.450 --> 00:08:30.139
Yeah, but it was my

221
00:08:30.300 --> 00:08:32.010
my little piece was

222
00:08:32.500 --> 00:08:33.809
what it was was sort of like.

223
00:08:36.210 --> 00:08:37.450
You know, we we we we sort of

224
00:08:37.490 --> 00:08:39.949
figured out that part of the problem

225
00:08:40.580 --> 00:08:41.580
was that.

226
00:08:43.320 --> 00:08:45.739
See has issues.

227
00:08:48.260 --> 00:08:49.260
And and

228
00:08:50.700 --> 00:08:52.289
so I started out so.

229
00:08:52.340 --> 00:08:54.179
So out of this

230
00:08:54.180 --> 00:08:56.490
like large pie of a project,

231
00:08:56.550 --> 00:08:57.720
my slice

232
00:08:58.380 --> 00:08:59.380
was.

233
00:09:01.420 --> 00:09:03.460
To make things a little easier

234
00:09:03.470 --> 00:09:04.969
from a programing language point of

235
00:09:04.970 --> 00:09:06.440
view and fix the

236
00:09:06.980 --> 00:09:08.450
pain points that came from the

237
00:09:08.720 --> 00:09:09.919
programing language part.

238
00:09:10.010 --> 00:09:11.840
And it started out as

239
00:09:13.460 --> 00:09:14.809
kind of do a better see.

240
00:09:15.890 --> 00:09:17.350
And then it and it

241
00:09:18.320 --> 00:09:19.219
got out of control.

242
00:09:19.220 --> 00:09:20.389
The rest of the project

243
00:09:22.130 --> 00:09:23.509
really ended up just providing

244
00:09:23.510 --> 00:09:24.510
context.

245
00:09:25.050 --> 00:09:26.779
You know, the only thing out of that

246
00:09:26.780 --> 00:09:28.669
project that survived was

247
00:09:28.670 --> 00:09:29.670
was Java.

248
00:09:31.670 --> 00:09:34.009
But but but the fact that it was

249
00:09:35.210 --> 00:09:37.309
sort of directed at a set of pain

250
00:09:37.310 --> 00:09:38.310
points.

251
00:09:38.780 --> 00:09:40.201
But it happened to be it, you

252
00:09:40.660 --> 00:09:41.660
know, about

253
00:09:42.740 --> 00:09:44.089
people who were living outside of

254
00:09:44.090 --> 00:09:45.709
data centers and people who were.

255
00:09:49.050 --> 00:09:50.422
Getting shredded by problems

256
00:09:50.880 --> 00:09:53.640
with networking and security

257
00:09:53.730 --> 00:09:54.899
and reliability.

258
00:09:56.850 --> 00:09:58.271
They had to build things that

259
00:09:58.710 --> 00:10:00.510
ran in hostile environments

260
00:10:01.080 --> 00:10:02.549
like in the homes,

261
00:10:05.010 --> 00:10:06.010
which in any home

262
00:10:07.020 --> 00:10:08.550
with a child in it is a

263
00:10:09.090 --> 00:10:11.070
it is a hostile environment.

264
00:10:12.560 --> 00:10:14.639
Speaking of hostile environments,

265
00:10:15.820 --> 00:10:16.820
feral.

266
00:10:20.590 --> 00:10:21.962
What are some of the guiding

267
00:10:22.420 --> 00:10:23.860
principles for designing for a

268
00:10:24.260 --> 00:10:25.260
business?

269
00:10:28.150 --> 00:10:29.150
I love you, Larry.

270
00:10:40.880 --> 00:10:42.739
More and more coming

271
00:10:42.740 --> 00:10:43.740
out of this.

272
00:10:43.760 --> 00:10:44.899
A computer scientist.

273
00:10:45.360 --> 00:10:46.389
So I sort of actively

274
00:10:47.600 --> 00:10:48.972
ignored some of the computer

275
00:10:49.700 --> 00:10:51.289
science literature of the time and

276
00:10:51.290 --> 00:10:53.220
said, well, what can we just talk

277
00:10:53.240 --> 00:10:54.739
together? One part of the brain will

278
00:10:54.740 --> 00:10:56.419
work more like a natural language

279
00:10:58.010 --> 00:10:59.190
instead of. But

280
00:10:59.840 --> 00:11:00.840
instead of

281
00:11:02.180 --> 00:11:03.180
putting on a

282
00:11:05.210 --> 00:11:07.100
campus university campus and

283
00:11:07.190 --> 00:11:08.629
deciding world, the walkways would

284
00:11:08.630 --> 00:11:09.589
go, we're just going to see where

285
00:11:09.590 --> 00:11:10.590
people want to walk.

286
00:11:10.910 --> 00:11:12.139
And they put shortcuts in all those

287
00:11:12.140 --> 00:11:13.610
places and know that more as a

288
00:11:13.940 --> 00:11:15.739
network, not as a terribly

289
00:11:15.800 --> 00:11:18.049
orthogonal or computer

290
00:11:18.050 --> 00:11:19.580
science or mathematical thing.

291
00:11:20.190 --> 00:11:21.649
And then that turned out to be

292
00:11:23.420 --> 00:11:24.829
in the right place at the right time

293
00:11:24.840 --> 00:11:25.840
for bootstrapping a

294
00:11:26.660 --> 00:11:27.660
lot of the web

295
00:11:29.840 --> 00:11:30.840
and.

NOTE 1:00:00 in transcript

296
00:11:31.720 --> 00:11:33.249
It also got used a lot for system

297
00:11:33.250 --> 00:11:35.290
administration and so

298
00:11:37.270 --> 00:11:38.740
various principles that relate

299
00:11:39.190 --> 00:11:40.809
to trying to

300
00:11:41.590 --> 00:11:43.479
provide API eyes to everything.

301
00:11:45.200 --> 00:11:46.629
I'm trying to be both a good text

302
00:11:46.630 --> 00:11:47.630
processing language

303
00:11:48.580 --> 00:11:50.559
linguistically, but also

304
00:11:50.560 --> 00:11:51.649
a quick Google language.

305
00:11:51.920 --> 00:11:52.920
Here's why it was.

306
00:11:53.630 --> 00:11:54.860
Kind of very useful,

307
00:11:55.450 --> 00:11:57.709
was hasty and there was text

308
00:11:57.980 --> 00:12:00.080
and, you

309
00:12:00.300 --> 00:12:02.129
know, database seconds that's that

310
00:12:02.220 --> 00:12:03.109
needs.

311
00:12:03.110 --> 00:12:04.219
And so it was in the right place at

312
00:12:04.220 --> 00:12:05.220
the right time.

313
00:12:06.350 --> 00:12:07.790
We were very fortunate to

314
00:12:08.330 --> 00:12:09.770
have that time because

315
00:12:10.390 --> 00:12:11.615
in the 90s it became very

316
00:12:12.290 --> 00:12:14.509
stable, a lot of a lot of widespread

317
00:12:14.520 --> 00:12:16.250
use. Lot of people were

318
00:12:16.600 --> 00:12:18.090
you know, the language itself was

319
00:12:18.320 --> 00:12:19.447
stabilizing in the form

320
00:12:20.330 --> 00:12:21.330
that it was.

321
00:12:21.410 --> 00:12:22.759
But there were a lot of issues

322
00:12:23.240 --> 00:12:24.240
and salat.

323
00:12:24.620 --> 00:12:26.539
And so in the year 2000,

324
00:12:26.540 --> 00:12:27.618
we we took a step back

325
00:12:28.580 --> 00:12:29.580
and and

326
00:12:30.410 --> 00:12:32.220
and basically said, we're gonna

327
00:12:32.240 --> 00:12:33.619
break everything he's breaking.

328
00:12:33.650 --> 00:12:34.924
So we kind of did the same

329
00:12:35.750 --> 00:12:36.750
thing.

330
00:12:37.450 --> 00:12:39.150
You know, it's done the pipeline to

331
00:12:39.170 --> 00:12:40.934
the step, except instead of breaking

332
00:12:41.330 --> 00:12:43.159
a few things, we decided to

333
00:12:43.160 --> 00:12:44.369
break everything here.

334
00:12:45.200 --> 00:12:46.866
So we came up with a whole raft of

335
00:12:46.880 --> 00:12:48.799
design principles and 50 years since

336
00:12:48.800 --> 00:12:49.729
then.

337
00:12:49.730 --> 00:12:51.169
By the way, forensics did come up

338
00:12:51.170 --> 00:12:52.159
two years ago.

339
00:12:52.160 --> 00:12:53.160
It's getting faster.

340
00:12:54.640 --> 00:12:56.279
And it's

341
00:12:56.640 --> 00:12:58.269
suddenly at the beginning we said we

342
00:12:58.270 --> 00:12:59.149
were going to do something

343
00:12:59.150 --> 00:13:00.559
impossible and and

344
00:13:01.020 --> 00:13:02.629
fail at it could be a very big

345
00:13:02.630 --> 00:13:04.570
failure. And so far, this proved

346
00:13:04.600 --> 00:13:06.669
me to be that. But in the course

347
00:13:06.670 --> 00:13:08.059
of the fail.

348
00:13:08.340 --> 00:13:09.879
Yeah. In the course of

349
00:13:10.220 --> 00:13:12.049
that, we came up some really

350
00:13:12.050 --> 00:13:13.519
interesting list of about 60

351
00:13:13.520 --> 00:13:15.019
different design principles.

352
00:13:19.160 --> 00:13:21.120
When read, even though I think many

353
00:13:21.310 --> 00:13:22.970
of them have already, you know,

354
00:13:23.790 --> 00:13:25.407
you know, kill two birds with one

355
00:13:25.450 --> 00:13:27.190
stone, pick the right people, you

356
00:13:27.220 --> 00:13:28.220
know.

357
00:13:29.640 --> 00:13:30.940
It's possible like that.

358
00:13:31.070 --> 00:13:32.649
But we came up with some cute ones,

359
00:13:32.650 --> 00:13:34.559
like, you think that's cute today?

360
00:13:38.650 --> 00:13:40.299
Oh, oh,

361
00:13:41.030 --> 00:13:42.030
know.

362
00:13:43.210 --> 00:13:44.960
Conserve your brackets because

363
00:13:45.130 --> 00:13:46.649
SFE, even Unicode does not have

364
00:13:47.000 --> 00:13:48.000
enough brackets

365
00:13:51.260 --> 00:13:52.534
that don't reinvent object

366
00:13:53.240 --> 00:13:54.349
orientation queerly,

367
00:13:55.190 --> 00:13:57.199
which arguably provided

368
00:13:59.480 --> 00:14:00.901
no substitute strong features

369
00:14:01.640 --> 00:14:03.169
for weak buttons, late binding

370
00:14:03.170 --> 00:14:04.889
sometimes oposite programs be.

371
00:14:09.060 --> 00:14:10.830
But some of the major ones are,

372
00:14:11.230 --> 00:14:13.159
you know, a lot of the stuff is

373
00:14:13.250 --> 00:14:14.909
just perfect to screw over time.

374
00:14:15.360 --> 00:14:16.980
And, you know,

375
00:14:18.330 --> 00:14:19.649
there were a lot of weird magical

376
00:14:19.650 --> 00:14:21.509
variables I'm going

377
00:14:21.510 --> 00:14:22.510
to.

378
00:14:23.130 --> 00:14:24.990
And so.

379
00:14:27.280 --> 00:14:28.879
A great deal of the redesign was to

380
00:14:28.890 --> 00:14:30.360
say, OK, what is the right peg

381
00:14:30.820 --> 00:14:31.840
to hang everything on?

382
00:14:32.290 --> 00:14:33.370
Is it off to Korea?

383
00:14:33.520 --> 00:14:34.892
Is it something in the works

384
00:14:35.440 --> 00:14:37.450
for school or in a larger school?

385
00:14:37.570 --> 00:14:38.570
Or what

386
00:14:39.570 --> 00:14:41.460
is the right to have each

387
00:14:41.640 --> 00:14:42.970
piece of information on?

388
00:14:43.330 --> 00:14:44.709
And if we don't have that thing, how

389
00:14:44.710 --> 00:14:45.710
do we create it?

390
00:14:46.810 --> 00:14:47.810
Great question.

391
00:14:48.040 --> 00:14:49.363
And since it's such a great

392
00:14:50.080 --> 00:14:51.969
question, Anderson, the spot

393
00:14:52.410 --> 00:14:54.490
and Terrence Parr

394
00:14:54.670 --> 00:14:56.589
had a quiz about while

395
00:14:56.590 --> 00:14:58.540
programmers value simplicity,

396
00:14:58.740 --> 00:15:00.640
they more often value

397
00:15:01.030 --> 00:15:02.590
powerful functionality.

398
00:15:02.680 --> 00:15:04.330
An amazing one liners

399
00:15:04.750 --> 00:15:06.580
incurring the cost of complexity.

400
00:15:07.540 --> 00:15:09.200
So for express.

401
00:15:09.710 --> 00:15:10.499
Yes.

402
00:15:10.500 --> 00:15:12.460
So what are your thoughts about

403
00:15:12.970 --> 00:15:15.250
simplicity versus complexity

404
00:15:15.280 --> 00:15:16.539
and some of the principles that

405
00:15:16.540 --> 00:15:18.360
might guide you, as you know?

406
00:15:18.610 --> 00:15:19.610
Really?

407
00:15:22.850 --> 00:15:23.850
Suppose I.

408
00:15:26.930 --> 00:15:27.930
I was always late

409
00:15:29.280 --> 00:15:31.090
in doing the language, saying

410
00:15:31.100 --> 00:15:31.879
that I've done.

411
00:15:31.880 --> 00:15:33.320
Always tried to

412
00:15:33.770 --> 00:15:35.179
make it so that there was only one

413
00:15:35.180 --> 00:15:36.180
way.

414
00:15:37.010 --> 00:15:38.889
So a lot of languages have

415
00:15:38.920 --> 00:15:40.159
four different ways of doing

416
00:15:40.160 --> 00:15:41.509
something and you pick the wrong

417
00:15:41.510 --> 00:15:43.640
one. And then only

418
00:15:45.080 --> 00:15:46.729
later you realize you went down the

419
00:15:46.730 --> 00:15:48.396
wrong branch and now you've got it

420
00:15:48.580 --> 00:15:49.319
back up.

421
00:15:49.320 --> 00:15:50.549
And so.

422
00:15:50.880 --> 00:15:52.820
So it's not

423
00:15:52.970 --> 00:15:53.970
it's not always easy.

424
00:15:54.530 --> 00:15:55.706
And I think there's like

425
00:15:56.360 --> 00:15:58.279
often we're guilty of creating

426
00:15:58.280 --> 00:15:59.840
what I call

427
00:16:00.140 --> 00:16:01.279
this thing I call simplex.

428
00:16:02.470 --> 00:16:03.880
Take something from blacks,

429
00:16:04.640 --> 00:16:06.012
you grab a single wrapper on

430
00:16:06.570 --> 00:16:08.040
top of it. That's make the

431
00:16:08.050 --> 00:16:09.450
complexity go away, but you're

432
00:16:09.510 --> 00:16:11.119
really creating is simplicity.

433
00:16:11.220 --> 00:16:13.049
You know, it's just like a

434
00:16:13.050 --> 00:16:14.700
bad abstraction on top of

435
00:16:15.510 --> 00:16:17.100
another bad abstraction that is

436
00:16:17.580 --> 00:16:18.580
simpler.

437
00:16:19.140 --> 00:16:20.140
So I don't

438
00:16:21.030 --> 00:16:22.409
know. I mean, it's it's

439
00:16:24.110 --> 00:16:25.869
we think that as picking up correct

440
00:16:25.870 --> 00:16:27.299
default overly complex thing.

441
00:16:28.810 --> 00:16:30.609
Yeah. But but I mean,

442
00:16:31.180 --> 00:16:32.649
the thing about that, about language

443
00:16:32.650 --> 00:16:34.880
design is like like any any decision

444
00:16:35.670 --> 00:16:37.809
you have to live with forever.

445
00:16:38.500 --> 00:16:40.740
I mean that and in languages

446
00:16:40.790 --> 00:16:42.659
you you can always add,

447
00:16:42.700 --> 00:16:43.899
but you can never take away.

448
00:16:44.580 --> 00:16:46.269
And so you've got to actually, as a

449
00:16:46.270 --> 00:16:47.529
language sign up to be very

450
00:16:47.530 --> 00:16:48.530
judicious.

451
00:16:50.920 --> 00:16:52.929
About reasoning over not

452
00:16:52.930 --> 00:16:54.519
so much what to put in, but what not

453
00:16:54.520 --> 00:16:56.309
to put it, an army

454
00:16:56.350 --> 00:16:57.879
because people come to you all the

455
00:16:57.880 --> 00:16:59.259
time. Wouldn't it be great if you

456
00:16:59.260 --> 00:17:00.419
could do this with that?

457
00:17:00.460 --> 00:17:01.389
Are you going to?

458
00:17:01.390 --> 00:17:03.309
Well, yeah, but you can't

459
00:17:03.730 --> 00:17:05.259
fundamentally change the nature of

460
00:17:05.260 --> 00:17:06.519
the beast. If you created an

461
00:17:06.520 --> 00:17:08.209
empirical and an

462
00:17:08.800 --> 00:17:10.239
imperative programing language, you

463
00:17:10.240 --> 00:17:11.469
can't turn it into a functional

464
00:17:11.470 --> 00:17:12.299
program.

465
00:17:12.300 --> 00:17:14.289
Show me you can you can borrow from

466
00:17:14.290 --> 00:17:15.390
functional programing.

467
00:17:16.180 --> 00:17:17.809
But it is what it is.

468
00:17:17.839 --> 00:17:19.269
And you gotta stay true to the

469
00:17:19.270 --> 00:17:20.789
nature of the beast, or you gotta

470
00:17:20.800 --> 00:17:21.800
create a new beast

471
00:17:22.710 --> 00:17:23.710
so-to-speak.

472
00:17:24.990 --> 00:17:25.990
Right.

473
00:17:27.250 --> 00:17:28.250
Any other thoughts?

474
00:17:29.530 --> 00:17:30.530
Excuse me?

475
00:17:30.650 --> 00:17:32.359
Can we interrupt you or you may have

476
00:17:32.790 --> 00:17:34.490
to interrupt any time.

477
00:17:35.960 --> 00:17:36.960
So.

478
00:17:38.040 --> 00:17:39.040
This is sort

479
00:17:39.930 --> 00:17:41.819
of the point about what what

480
00:17:41.820 --> 00:17:42.820
do you leave out?

481
00:17:43.620 --> 00:17:45.479
And the what I

482
00:17:45.480 --> 00:17:47.309
remember in

483
00:17:47.310 --> 00:17:49.169
the early days of Python's design,

484
00:17:50.490 --> 00:17:52.320
there were so many people

485
00:17:52.410 --> 00:17:54.380
who thought they had a good idea.

486
00:17:58.060 --> 00:18:00.190
Is sufficiently critical

487
00:18:00.880 --> 00:18:02.079
to say no.

488
00:18:03.880 --> 00:18:04.958
And so there are a few

489
00:18:05.920 --> 00:18:06.920
things that

490
00:18:08.350 --> 00:18:10.150
didn't work out at the time,

491
00:18:10.230 --> 00:18:11.308
I I haven't learned to

492
00:18:12.190 --> 00:18:13.079
say no.

493
00:18:13.080 --> 00:18:15.039
And I quickly enough

494
00:18:15.040 --> 00:18:16.719
did learn to say no.

495
00:18:16.750 --> 00:18:18.190
And I sort of felt

496
00:18:19.030 --> 00:18:20.030
like several lines

497
00:18:21.460 --> 00:18:23.319
of defense. Like the first line of

498
00:18:23.320 --> 00:18:25.299
defense is, hey,

499
00:18:25.330 --> 00:18:26.799
you think you need a new language

500
00:18:26.800 --> 00:18:28.330
feature, but actually you can

501
00:18:28.390 --> 00:18:30.109
already write that typo

502
00:18:31.930 --> 00:18:33.579
if you needed the locked right to

503
00:18:33.580 --> 00:18:34.720
function or a module.

504
00:18:35.740 --> 00:18:37.210
Well, if they say, yeah, but

505
00:18:37.270 --> 00:18:38.649
everybody needs that.

506
00:18:38.800 --> 00:18:40.900
Well nowadays we say put it on the

507
00:18:41.590 --> 00:18:42.913
package repository in those

508
00:18:43.630 --> 00:18:45.880
days. I said well maybe

509
00:18:46.480 --> 00:18:47.919
you can propose a new standard

510
00:18:47.920 --> 00:18:48.789
library module.

511
00:18:48.790 --> 00:18:50.019
That's a lot cheaper for the

512
00:18:50.020 --> 00:18:51.196
language design than the

513
00:18:51.910 --> 00:18:52.960
new language feature.

514
00:18:54.250 --> 00:18:55.250
And

515
00:18:56.170 --> 00:18:58.299
another line of defense was,

516
00:18:58.330 --> 00:18:59.920
well, you can actually write

517
00:18:59.980 --> 00:19:00.980
extensions in C

518
00:19:01.900 --> 00:19:03.999
or Fortran if you really care to.

519
00:19:04.930 --> 00:19:06.549
And you can help yourself that you

520
00:19:06.550 --> 00:19:07.824
can modify the behavior of

521
00:19:08.470 --> 00:19:10.059
the language in ways that

522
00:19:10.810 --> 00:19:12.329
aren't accessible when you just

523
00:19:12.910 --> 00:19:14.739
write the pure language.

524
00:19:14.770 --> 00:19:16.029
But you can still

525
00:19:16.600 --> 00:19:17.678
sort of you can extend

526
00:19:18.700 --> 00:19:19.876
it in a way that doesn't

527
00:19:19.960 --> 00:19:21.609
fundamentally alter

528
00:19:22.060 --> 00:19:23.260
the core language.

529
00:19:23.980 --> 00:19:25.839
And if you've tried all those

530
00:19:25.840 --> 00:19:27.970
things and and you failed,

531
00:19:29.230 --> 00:19:31.060
then maybe you could

532
00:19:31.150 --> 00:19:32.865
argue for, well, you have to change

533
00:19:33.160 --> 00:19:34.539
something in the core language.

534
00:19:36.140 --> 00:19:37.859
Hi, I'm interrupting.

535
00:19:39.060 --> 00:19:40.109
I find that

536
00:19:41.070 --> 00:19:42.739
whenever I get feedback on

537
00:19:43.230 --> 00:19:45.009
whatever whatever it is, program

538
00:19:45.130 --> 00:19:46.349
makers, I've been working on it.

539
00:19:46.450 --> 00:19:48.269
It's people come to

540
00:19:48.270 --> 00:19:50.460
you with a with a particular

541
00:19:50.520 --> 00:19:52.589
instance of a problem.

542
00:19:52.710 --> 00:19:54.229
And it's your job as a language

543
00:19:54.240 --> 00:19:56.459
assignment to tease out the class

544
00:19:56.550 --> 00:19:58.170
problem they're talking about

545
00:19:58.650 --> 00:20:00.539
and then try to understand whether

546
00:20:00.540 --> 00:20:02.579
there is a feature there underneath

547
00:20:02.580 --> 00:20:03.779
that you could put in place

548
00:20:04.440 --> 00:20:06.299
that is broadly applicable,

549
00:20:06.960 --> 00:20:08.819
not just to that one crawl.

550
00:20:13.770 --> 00:20:16.200
You're putting in place a class,

551
00:20:16.350 --> 00:20:17.390
not an instance suit.

552
00:20:17.520 --> 00:20:19.219
So that damn slide.

553
00:20:20.010 --> 00:20:21.676
And if all you're talking about is

554
00:20:21.750 --> 00:20:23.009
an instance, then you're probably

555
00:20:23.010 --> 00:20:24.010
better off.

556
00:20:28.800 --> 00:20:29.999
Save your money for power.

557
00:20:32.700 --> 00:20:33.809
Actually found this.

558
00:20:35.320 --> 00:20:37.089
In the early design of the Perl 6,

559
00:20:37.090 --> 00:20:38.829
we asked our CS I expected.

560
00:20:40.020 --> 00:20:41.829
Since we got 361,

561
00:20:42.150 --> 00:20:43.630
so we have this in spades

562
00:20:44.670 --> 00:20:45.670
lately.

563
00:20:45.720 --> 00:20:47.189
And, you know, it's completely

564
00:20:47.190 --> 00:20:48.190
overwhelming.

565
00:20:48.260 --> 00:20:49.260
Until I

566
00:20:50.120 --> 00:20:51.660
hit upon this principle,

567
00:20:53.130 --> 00:20:54.809
which Anderson alluded to,

568
00:20:55.170 --> 00:20:56.689
which is basically a lot of the

569
00:20:56.940 --> 00:20:57.949
proposed solution.

570
00:20:58.180 --> 00:20:59.999
Yeah, but there is a pain point

571
00:21:00.000 --> 00:21:01.000
underneath.

572
00:21:01.140 --> 00:21:02.849
And if you look at more enough of

573
00:21:02.870 --> 00:21:04.769
these RFD and they haven't

574
00:21:04.770 --> 00:21:06.485
seen pain point, there is something

575
00:21:06.960 --> 00:21:07.960
you should address.

576
00:21:08.160 --> 00:21:09.809
There's some fundamental unification

577
00:21:09.810 --> 00:21:11.309
you should do underneath or

578
00:21:11.760 --> 00:21:13.589
something that is is

579
00:21:14.220 --> 00:21:15.779
clunky that could be fixed.

580
00:21:16.120 --> 00:21:17.519
And especially if you're doing a

581
00:21:17.520 --> 00:21:18.929
complete redesign that you can think

582
00:21:18.930 --> 00:21:20.170
about the sort of.

583
00:21:34.630 --> 00:21:35.630
To fix it.

584
00:21:36.610 --> 00:21:38.029
How do you know?

585
00:21:38.060 --> 00:21:40.279
There were a lot of problems

586
00:21:40.360 --> 00:21:41.360
where.

587
00:22:00.830 --> 00:22:02.509
Difference that just

588
00:22:02.920 --> 00:22:04.829
won a victory with themselves.

589
00:22:05.170 --> 00:22:06.839
You get like 40 different.

590
00:22:35.350 --> 00:22:36.350
Released.

591
00:22:36.520 --> 00:22:37.870
It was really clear in the way

592
00:22:38.690 --> 00:22:40.569
that this was like a big

593
00:22:40.570 --> 00:22:42.730
issue and

594
00:22:43.090 --> 00:22:45.189
you feel dry and I came

595
00:22:45.190 --> 00:22:46.758
about as close to spilling blood

596
00:22:47.140 --> 00:22:49.029
on this topic as I've

597
00:22:49.030 --> 00:22:50.030
got.

598
00:22:51.330 --> 00:22:52.408
And, you know, I would

599
00:22:53.220 --> 00:22:54.220
much rather.

600
00:22:55.700 --> 00:22:56.700
Do something

601
00:22:57.690 --> 00:22:59.140
much, rather leave something out

602
00:22:59.630 --> 00:23:00.880
there, do something stupid.

603
00:23:02.440 --> 00:23:04.319
One of our principals planned to

604
00:23:04.320 --> 00:23:05.320
visit Margaret later.

605
00:23:05.690 --> 00:23:07.430
Yeah. Well, well,

606
00:23:07.780 --> 00:23:09.940
what that means is that that means

607
00:23:11.410 --> 00:23:12.539
the opening answer is no.

608
00:23:12.720 --> 00:23:13.669
No.

609
00:23:13.670 --> 00:23:14.089
Right.

610
00:23:14.090 --> 00:23:15.409
And one of the things that happened

611
00:23:15.410 --> 00:23:17.179
with like a whole bunch of different

612
00:23:17.260 --> 00:23:19.000
outfits and their job

613
00:23:19.550 --> 00:23:21.167
was that they kind of turned into

614
00:23:21.170 --> 00:23:22.170
competitions

615
00:23:23.710 --> 00:23:25.819
and generics

616
00:23:25.850 --> 00:23:26.950
and closer to work both.

617
00:23:27.110 --> 00:23:28.110
Probably the

618
00:23:29.420 --> 00:23:31.219
most hard fought competition.

619
00:23:31.990 --> 00:23:33.349
People were writing pretty steep

620
00:23:33.350 --> 00:23:34.639
pieces on the topic.

621
00:23:35.120 --> 00:23:36.739
And I was like trying them all out.

622
00:23:38.420 --> 00:23:40.160
I mean, some had no shortage

623
00:23:40.730 --> 00:23:42.619
of smart people at the time, but

624
00:23:43.040 --> 00:23:44.390
it's really hard to be

625
00:23:45.320 --> 00:23:47.269
a planet full

626
00:23:47.270 --> 00:23:48.470
of grad students,

627
00:23:49.330 --> 00:23:51.140
even grad students.

628
00:23:51.350 --> 00:23:52.659
At some point, I think most kids who

629
00:23:52.660 --> 00:23:53.519
were in grad school.

630
00:23:53.520 --> 00:23:54.520
That's correct.

631
00:23:54.860 --> 00:23:56.859
And in linguistics and linguistics.

632
00:23:58.790 --> 00:23:59.790
But it's.

633
00:24:00.480 --> 00:24:01.640
Oh, probably

634
00:24:02.360 --> 00:24:04.519
20, 30 years ago to say

635
00:24:04.940 --> 00:24:06.650
I'm going to write my own language.

636
00:24:07.310 --> 00:24:09.169
And people are very different.

637
00:24:09.410 --> 00:24:10.910
Back then, the

638
00:24:11.390 --> 00:24:12.390
Internet was sort

639
00:24:13.220 --> 00:24:14.209
of starting.

640
00:24:14.210 --> 00:24:16.769
Maybe not that they're single.

641
00:24:17.300 --> 00:24:18.460
I think we have computers.

642
00:24:18.470 --> 00:24:20.700
We have cell phones.

643
00:24:20.930 --> 00:24:22.819
Nobody cared, but nobody

644
00:24:22.820 --> 00:24:23.930
cared. What's all that about?

645
00:24:24.980 --> 00:24:26.599
Nobody cared what you did with the

646
00:24:26.600 --> 00:24:27.600
eighth that.

647
00:24:31.630 --> 00:24:32.630
Decades.

648
00:24:35.180 --> 00:24:37.499
But these are languages

649
00:24:37.530 --> 00:24:38.849
that persisted.

650
00:24:39.300 --> 00:24:41.380
Many changes and

651
00:24:42.010 --> 00:24:44.109
they're on different hardware.

652
00:24:45.300 --> 00:24:47.099
When you started

653
00:24:47.370 --> 00:24:49.210
designing your language, what

654
00:24:49.230 --> 00:24:50.669
was the key goal that you were

655
00:24:50.670 --> 00:24:52.739
trying to meet versus

656
00:24:52.740 --> 00:24:54.640
using some other extinguishment?

657
00:24:56.570 --> 00:24:57.844
Well, I mean, for me, it's

658
00:24:58.460 --> 00:25:00.309
never been people.

659
00:25:01.730 --> 00:25:03.980
And one of the secrets of success

660
00:25:04.010 --> 00:25:05.010
is try to solve as

661
00:25:05.930 --> 00:25:07.749
many problems at once as you can.

662
00:25:08.690 --> 00:25:10.279
And, you know, if there's just

663
00:25:10.550 --> 00:25:12.859
one problem, there's probably

664
00:25:12.860 --> 00:25:13.940
an easier way.

665
00:25:17.350 --> 00:25:18.350
Right.

666
00:25:19.270 --> 00:25:20.270
Right.

667
00:25:20.640 --> 00:25:22.749
You think that way you'll never

668
00:25:22.750 --> 00:25:23.760
have a new program.

669
00:25:25.380 --> 00:25:26.609
Well, there's always a better

670
00:25:26.610 --> 00:25:27.950
solution in the 300.

671
00:25:27.990 --> 00:25:29.970
Well, it depends on how much

672
00:25:30.000 --> 00:25:31.000
you pay.

673
00:25:31.050 --> 00:25:32.339
Chasing down memory,

674
00:25:32.950 --> 00:25:33.950
corruption.

675
00:25:34.050 --> 00:25:35.050
But

676
00:25:35.930 --> 00:25:37.529
the three virtues of the program

677
00:25:37.530 --> 00:25:39.510
are. Blasi This impatience

678
00:25:39.540 --> 00:25:40.589
and cupids

679
00:25:42.350 --> 00:25:43.409
takes Kupers.

680
00:25:46.290 --> 00:25:47.579
The world needs another program

681
00:25:47.920 --> 00:25:48.920
like.

682
00:25:51.150 --> 00:25:52.150
Is that a code?

683
00:25:53.160 --> 00:25:54.299
But the thing about programing

684
00:25:54.300 --> 00:25:56.009
language is that I think a lot of

685
00:25:56.010 --> 00:25:57.989
people don't realize that every

686
00:25:57.990 --> 00:25:59.730
programing language is about

687
00:26:00.240 --> 00:26:01.580
90 percent the same

688
00:26:02.220 --> 00:26:03.869
and maybe 10 percent

689
00:26:04.590 --> 00:26:05.590
if you're lucky.

690
00:26:06.630 --> 00:26:08.249
And a lot of people get very focused

691
00:26:08.250 --> 00:26:09.959
on the 10 percent new and forget

692
00:26:09.960 --> 00:26:11.390
about the 90 percent of the same.

693
00:26:12.060 --> 00:26:13.559
That makes it at great length.

694
00:26:13.770 --> 00:26:14.159
Right.

695
00:26:14.160 --> 00:26:16.140
I mean, there's a lot of hard

696
00:26:16.230 --> 00:26:17.896
or you work in creating programing

697
00:26:18.060 --> 00:26:19.760
language, loss, semantics that

698
00:26:19.910 --> 00:26:21.059
it's got to worry about.

699
00:26:21.120 --> 00:26:23.009
And like everyone and loves

700
00:26:23.010 --> 00:26:24.239
to talk about syntax

701
00:26:25.760 --> 00:26:27.839
and syntax is the easy part.

702
00:26:27.900 --> 00:26:29.849
Semantics is the hard part.

NOTE 01:15:00 in transcript

703
00:26:30.540 --> 00:26:31.979
You know, like how do the types of

704
00:26:31.980 --> 00:26:33.940
work and one of one only dogbone

705
00:26:35.130 --> 00:26:36.689
supported promotions and one of the

706
00:26:36.690 --> 00:26:38.729
different kinds of type constructors

707
00:26:38.730 --> 00:26:40.319
that the language, as etc.

708
00:26:40.380 --> 00:26:41.999
said are these are the hard things

709
00:26:42.000 --> 00:26:44.229
to do to design.

710
00:26:44.330 --> 00:26:46.019
But, but but not the things that

711
00:26:46.020 --> 00:26:48.089
people people love to bike

712
00:26:48.090 --> 00:26:49.539
shed on the syntax, you know,

713
00:26:49.560 --> 00:26:51.510
shouldn't be a colon or or

714
00:26:51.630 --> 00:26:52.239
comma.

715
00:26:52.240 --> 00:26:53.690
You know, it's like, oh my God,

716
00:26:54.160 --> 00:26:55.859
it's have a long thread about that.

717
00:26:58.360 --> 00:27:00.250
So talking about

718
00:27:00.640 --> 00:27:02.799
people having opinions and

719
00:27:02.800 --> 00:27:03.999
syntax and

720
00:27:04.700 --> 00:27:05.700
typing.

721
00:27:08.560 --> 00:27:10.570
These languages don't all

722
00:27:11.140 --> 00:27:13.119
take the same approach to typing.

723
00:27:14.200 --> 00:27:15.359
Maybe we'll start with veto

724
00:27:16.060 --> 00:27:18.309
and talk about typing in

725
00:27:18.760 --> 00:27:20.619
Python and then

726
00:27:20.620 --> 00:27:22.509
kind of work our way around.

727
00:27:30.700 --> 00:27:32.619
I mean, I just

728
00:27:32.620 --> 00:27:34.059
remember that when

729
00:27:35.640 --> 00:27:37.900
Pyfrom first happened,

730
00:27:38.770 --> 00:27:40.630
it was not a class.

731
00:27:41.620 --> 00:27:43.510
It was a little conversion function.

732
00:27:43.780 --> 00:27:45.069
There was an internal

733
00:27:45.760 --> 00:27:47.739
object type, which was really

734
00:27:47.740 --> 00:27:49.630
just kind of a Zite table

735
00:27:50.890 --> 00:27:53.109
that represented integer objects.

736
00:27:55.250 --> 00:27:57.400
And there was a built in function

737
00:27:57.440 --> 00:27:59.660
if you needed to convert the string

738
00:27:59.690 --> 00:28:01.219
to an integer and that

739
00:28:02.000 --> 00:28:03.323
was we had a bunch of those

740
00:28:03.890 --> 00:28:04.890
functions

741
00:28:06.350 --> 00:28:08.420
and we realized that we

742
00:28:09.710 --> 00:28:10.999
we had made a mistake.

743
00:28:11.030 --> 00:28:12.304
We had given users classes

744
00:28:13.040 --> 00:28:14.930
that were

745
00:28:16.040 --> 00:28:17.363
different from the built in

746
00:28:17.900 --> 00:28:19.000
object types.

747
00:28:20.500 --> 00:28:22.439
And for a long time, it was like,

748
00:28:22.560 --> 00:28:23.560
oh, well, like the

749
00:28:24.420 --> 00:28:25.645
real stuff is implemented

750
00:28:26.460 --> 00:28:27.460
in C and the user

751
00:28:28.680 --> 00:28:30.199
writes a little bit of blue top

752
00:28:30.660 --> 00:28:32.009
of that. And when

753
00:28:32.970 --> 00:28:34.489
when I found out that we had 80

754
00:28:35.130 --> 00:28:38.040
different competing web frameworks,

755
00:28:42.270 --> 00:28:44.270
it was sort of time to

756
00:28:44.510 --> 00:28:46.225
to realize that people were writing

757
00:28:46.470 --> 00:28:48.920
much larger programs.

758
00:28:49.020 --> 00:28:51.299
And that's a different approach.

759
00:28:51.300 --> 00:28:52.680
Two types was needed.

760
00:28:52.770 --> 00:28:54.420
And that's where we sort of

761
00:28:55.170 --> 00:28:57.299
reinvented the whole approach

762
00:28:57.300 --> 00:28:58.550
to types in Python.

763
00:29:01.050 --> 00:29:02.680
And there was a bunch of cleanup

764
00:29:02.700 --> 00:29:04.569
that didn't happen until 5 3,

765
00:29:04.570 --> 00:29:05.529
actually.

766
00:29:05.530 --> 00:29:07.799
But one of the things we introduced

767
00:29:07.890 --> 00:29:08.968
and we were lucky that

768
00:29:09.720 --> 00:29:11.700
there weren't actually despite those

769
00:29:11.760 --> 00:29:13.319
80 web frameworks, there weren't

770
00:29:13.380 --> 00:29:15.420
enough users that

771
00:29:16.080 --> 00:29:17.969
we were completely stymied by

772
00:29:17.970 --> 00:29:19.289
backwards compatibility.

773
00:29:19.450 --> 00:29:21.029
Yet like we are now.

774
00:29:22.660 --> 00:29:24.081
So we just one day we changed

775
00:29:24.790 --> 00:29:26.529
the function into

776
00:29:26.940 --> 00:29:28.700
a designated for the class.

777
00:29:30.790 --> 00:29:32.569
And sort of

778
00:29:32.780 --> 00:29:34.660
followed that calling, the class

779
00:29:34.710 --> 00:29:36.789
would be constructing

780
00:29:36.790 --> 00:29:39.009
an instance of the class

781
00:29:39.460 --> 00:29:41.589
so that if you had simple

782
00:29:41.590 --> 00:29:43.480
code that wrote in

783
00:29:43.570 --> 00:29:45.419
left brand some expression right.

784
00:29:47.080 --> 00:29:49.149
And would work exactly the same way

785
00:29:49.600 --> 00:29:51.150
that we have the same effect.

786
00:29:51.190 --> 00:29:53.079
But the workings

787
00:29:53.170 --> 00:29:54.970
inside were completely different.

788
00:29:55.040 --> 00:29:57.010
There was one particular file

789
00:29:57.040 --> 00:29:58.420
that's sort of

790
00:29:59.560 --> 00:30:00.900
implemented the

791
00:30:01.720 --> 00:30:03.309
that's sort of the basic

792
00:30:03.520 --> 00:30:04.941
functionality of types in the

793
00:30:05.170 --> 00:30:06.759
language and it

794
00:30:07.030 --> 00:30:08.402
in for one to it was like 50

795
00:30:09.010 --> 00:30:10.010
lines of code.

796
00:30:10.090 --> 00:30:12.279
And at some point I rewrote

797
00:30:12.280 --> 00:30:13.770
it and it was five thousand by.

798
00:30:16.560 --> 00:30:17.560
That's great.

799
00:30:18.590 --> 00:30:20.359
What about Titus in Dublin?

800
00:30:26.970 --> 00:30:28.779
I got this this this one

801
00:30:28.800 --> 00:30:31.089
history of caring about things

802
00:30:31.090 --> 00:30:32.090
like.

803
00:30:37.820 --> 00:30:39.410
Building robot software

804
00:30:41.340 --> 00:30:43.499
often that that

805
00:30:44.390 --> 00:30:45.900
that comes out about, you know, I'm

806
00:30:45.980 --> 00:30:47.450
much more worried about

807
00:30:47.960 --> 00:30:49.579
what it takes to

808
00:30:50.030 --> 00:30:51.829
build production quality software

809
00:30:52.400 --> 00:30:54.109
than about what it takes, just like

810
00:30:54.770 --> 00:30:55.940
you like a quick thing

811
00:30:56.830 --> 00:30:58.447
and chuckles not a great language

812
00:30:58.940 --> 00:30:59.789
for.

813
00:30:59.790 --> 00:31:00.819
Things, but it's it's

814
00:31:01.610 --> 00:31:03.450
OK, it's not

815
00:31:03.570 --> 00:31:05.609
for me. One of the one of the things

816
00:31:05.610 --> 00:31:07.359
that I love to do and

817
00:31:08.160 --> 00:31:09.240
maybe I'm weird, but

818
00:31:10.740 --> 00:31:12.119
being able to do automatic or

819
00:31:12.150 --> 00:31:14.009
improving on hunks of cup

820
00:31:15.070 --> 00:31:16.420
and and.

821
00:31:17.080 --> 00:31:19.179
Type system is

822
00:31:19.180 --> 00:31:20.950
a really great place

823
00:31:22.000 --> 00:31:24.069
to be, one of the foundations

824
00:31:24.630 --> 00:31:26.080
there begin they're improving

825
00:31:26.470 --> 00:31:27.470
it for.

826
00:31:28.440 --> 00:31:30.329
Four months of code turns out to be

827
00:31:30.330 --> 00:31:31.979
really useful for things like

828
00:31:32.550 --> 00:31:33.550
building optimized.

829
00:31:35.340 --> 00:31:37.109
And doing it

830
00:31:38.040 --> 00:31:39.059
ahead of time.

831
00:31:43.280 --> 00:31:45.299
To be able to hear

832
00:31:45.300 --> 00:31:47.720
improvable way as many

833
00:31:48.080 --> 00:31:50.220
things as possible, so

834
00:31:50.370 --> 00:31:51.650
satellite like like,

835
00:31:52.400 --> 00:31:54.289
you know, one of the not well-known

836
00:31:54.350 --> 00:31:56.329
things about about Java is

837
00:31:56.330 --> 00:31:58.190
that, you know, in the jobless pack,

838
00:31:58.310 --> 00:32:00.019
you know, it says Ouray, subscript

839
00:32:00.020 --> 00:32:01.900
checking is always on.

840
00:32:04.430 --> 00:32:05.930
But, you know, it's only

841
00:32:05.960 --> 00:32:07.490
conceptually all of his all

842
00:32:08.750 --> 00:32:10.279
right. But the truth is

843
00:32:10.660 --> 00:32:12.319
that there are there's more than on

844
00:32:12.320 --> 00:32:14.060
my books or

845
00:32:14.150 --> 00:32:15.470
the compiler

846
00:32:16.970 --> 00:32:18.859
to improve away

847
00:32:18.890 --> 00:32:20.690
almost all index checks.

848
00:32:22.100 --> 00:32:23.521
And same thing with like null

849
00:32:23.810 --> 00:32:25.329
pointer checks and all kinds of

850
00:32:25.640 --> 00:32:27.679
things that look

851
00:32:27.860 --> 00:32:29.029
like they're heavy weight,

852
00:32:30.170 --> 00:32:31.170
but they're really cheap,

853
00:32:32.180 --> 00:32:33.769
you know, so it's a one of the

854
00:32:33.770 --> 00:32:35.289
things that I like.

855
00:32:36.180 --> 00:32:37.400
I really cared about it.

856
00:32:37.430 --> 00:32:39.499
The time was that A plus

857
00:32:39.500 --> 00:32:41.150
B should almost always be

858
00:32:42.280 --> 00:32:43.790
at most one instruction

859
00:32:44.690 --> 00:32:46.779
bonus points for zero instructions.

860
00:32:49.700 --> 00:32:50.700
What about after?

861
00:32:51.450 --> 00:32:52.849
Well, the thing about zero

862
00:32:52.850 --> 00:32:54.619
instructions is often you can like

863
00:32:54.620 --> 00:32:56.749
fold them into some weird addressing

864
00:32:56.750 --> 00:32:57.750
mode.

865
00:32:57.980 --> 00:32:58.980
Right.

866
00:32:59.200 --> 00:33:00.200
And.

867
00:33:00.750 --> 00:33:02.769
And so

868
00:33:02.810 --> 00:33:04.500
in kind of the universe, I'm

869
00:33:04.650 --> 00:33:05.890
right, I tend to live.

870
00:33:07.180 --> 00:33:09.069
Being able to look

871
00:33:09.070 --> 00:33:11.019
at A plus B

872
00:33:11.020 --> 00:33:12.739
and realize it's

873
00:33:12.880 --> 00:33:14.839
that destruction that all

874
00:33:14.860 --> 00:33:16.589
feeds off of the type system.

875
00:33:17.890 --> 00:33:19.779
And sometimes you can you

876
00:33:19.780 --> 00:33:21.969
can do this with sort of optimistic

877
00:33:22.940 --> 00:33:24.810
just-in-time coupon compilation,

878
00:33:26.110 --> 00:33:27.609
depending on how far you want to

879
00:33:27.610 --> 00:33:29.169
push that. So like the

880
00:33:29.860 --> 00:33:31.989
the JavaScript engine

881
00:33:31.990 --> 00:33:33.819
in Chrome is

882
00:33:33.980 --> 00:33:36.220
is absolutely astonishing

883
00:33:36.360 --> 00:33:37.910
for the the

884
00:33:38.440 --> 00:33:40.479
kind of core that they go

885
00:33:40.480 --> 00:33:42.940
through too optimistically.

886
00:33:45.120 --> 00:33:47.249
JavaScript code, but

887
00:33:48.600 --> 00:33:50.609
it's also very hard to do those

888
00:33:50.610 --> 00:33:52.178
kinds of tricks if you're trying

889
00:33:52.650 --> 00:33:54.229
to get into smaller

890
00:33:54.570 --> 00:33:55.570
devices.

891
00:33:56.340 --> 00:33:58.619
So, you know, there are

892
00:33:58.650 --> 00:34:00.240
there are Java compilers that

893
00:34:00.540 --> 00:34:02.130
work for machines that only have

894
00:34:02.160 --> 00:34:03.160
like.

895
00:34:06.660 --> 00:34:08.819
You know, to do that kind of compact

896
00:34:08.980 --> 00:34:10.352
distillation, you need every

897
00:34:11.110 --> 00:34:12.999
kind of hope that gives you

898
00:34:13.030 --> 00:34:14.919
every little every last

899
00:34:14.920 --> 00:34:16.030
drop of information.

900
00:34:16.270 --> 00:34:18.189
And the earlier, you

901
00:34:18.190 --> 00:34:20.079
know, what better job you can do.

902
00:34:20.489 --> 00:34:22.519
Right. So speaking

903
00:34:22.540 --> 00:34:24.339
of lots of information that one

904
00:34:24.340 --> 00:34:25.409
gets, typescript

905
00:34:26.860 --> 00:34:28.539
gives you a lot of flexibility

906
00:34:29.110 --> 00:34:30.110
and power.

907
00:34:30.730 --> 00:34:31.809
What's your general view that

908
00:34:31.810 --> 00:34:32.810
typing?

909
00:34:34.659 --> 00:34:36.227
Well, let me actually it's funny

910
00:34:36.610 --> 00:34:37.650
that you mention

911
00:34:38.550 --> 00:34:39.550
cycles and counting.

912
00:34:39.639 --> 00:34:40.928
How many? And whenever I do, I

913
00:34:40.929 --> 00:34:42.460
remember when I wrote Turbo Pascal,

914
00:34:42.489 --> 00:34:44.110
which was all written in C-A-T

915
00:34:44.260 --> 00:34:45.099
assembly code.

916
00:34:45.100 --> 00:34:46.389
And back in those days,

917
00:34:46.960 --> 00:34:48.428
you could literally look up the

918
00:34:48.429 --> 00:34:50.589
intel manual on the sidewalk

919
00:34:50.590 --> 00:34:51.819
or whatever, you know, and see how

920
00:34:51.820 --> 00:34:53.529
many clock cycles every construction

921
00:34:53.530 --> 00:34:55.239
took. And actually it would work out

922
00:34:55.600 --> 00:34:56.600
just like that.

923
00:34:56.650 --> 00:34:57.850
I remember now

924
00:34:58.690 --> 00:35:00.013
everything takes zero clock

925
00:35:00.730 --> 00:35:02.469
cycles except when it takes a

926
00:35:02.470 --> 00:35:03.470
thousand clocks.

927
00:35:05.200 --> 00:35:06.719
The memory and it is absolutely

928
00:35:07.480 --> 00:35:08.269
impossible.

929
00:35:08.270 --> 00:35:10.059
The reason about what I've seen you

930
00:35:10.060 --> 00:35:11.440
use execute your code today.

931
00:35:11.560 --> 00:35:12.820
That that's just one.

932
00:35:13.300 --> 00:35:14.979
Well, it's not impossible,

933
00:35:15.480 --> 00:35:16.699
but it's much harder.

934
00:35:17.090 --> 00:35:18.969
Yeah, it is a complete pain in

935
00:35:18.970 --> 00:35:19.970
the ass.

936
00:35:21.400 --> 00:35:22.400
And they don't give

937
00:35:23.260 --> 00:35:24.129
you manuals.

938
00:35:24.130 --> 00:35:25.130
Not good at all.

939
00:35:25.140 --> 00:35:26.049
You don't want to actually

940
00:35:26.050 --> 00:35:26.949
understand that.

941
00:35:26.950 --> 00:35:28.209
If you need to get the chip

942
00:35:28.210 --> 00:35:29.210
diagrams.

943
00:35:29.330 --> 00:35:31.659
But but with respect to types,

944
00:35:31.660 --> 00:35:33.369
I've always worked on

945
00:35:33.580 --> 00:35:35.295
programing languages that have type

946
00:35:35.440 --> 00:35:35.979
systems.

947
00:35:35.980 --> 00:35:37.156
But it's interesting how

948
00:35:38.200 --> 00:35:40.479
it's gone from being type systems

949
00:35:40.480 --> 00:35:41.480
far

950
00:35:42.400 --> 00:35:43.919
for the code generators sake or

951
00:35:44.260 --> 00:35:46.079
type systems for four

952
00:35:46.380 --> 00:35:48.549
for, you know, generating

953
00:35:48.550 --> 00:35:49.550
errors too.

954
00:35:50.080 --> 00:35:51.158
Now I almost view type

955
00:35:52.150 --> 00:35:53.277
systems as as a tooling

956
00:35:54.340 --> 00:35:56.109
feature, and that's really sort of

957
00:35:56.320 --> 00:35:57.869
the thing that has been interesting.

958
00:35:57.870 --> 00:35:59.290
High automate for typescript.

959
00:35:59.350 --> 00:36:00.880
It is, you know,

960
00:36:01.660 --> 00:36:03.081
first of all, starting with a

961
00:36:03.100 --> 00:36:04.449
dynamically typed programing

962
00:36:04.450 --> 00:36:05.799
language like JavaScript and then

963
00:36:05.800 --> 00:36:07.869
trying to add a type system on top

964
00:36:07.870 --> 00:36:09.000
of it that

965
00:36:10.320 --> 00:36:11.692
that faithfully reflects the

966
00:36:12.040 --> 00:36:13.363
semantics of the programing

967
00:36:13.480 --> 00:36:14.409
language.

968
00:36:14.410 --> 00:36:16.179
And the reason for doing it is

969
00:36:16.180 --> 00:36:18.159
actually not because we think type

970
00:36:18.160 --> 00:36:19.924
systems are interesting, but because

971
00:36:19.960 --> 00:36:21.381
if you think about what it is

972
00:36:21.820 --> 00:36:24.219
that powers programmer productivity

973
00:36:24.220 --> 00:36:25.689
today, like everyone loves their

974
00:36:25.690 --> 00:36:26.690
I.T. like.

975
00:36:26.710 --> 00:36:28.359
Whatever you're using, I hope it's

976
00:36:28.360 --> 00:36:29.289
Visual Studio code.

977
00:36:29.290 --> 00:36:31.179
But putting that if it's

978
00:36:31.180 --> 00:36:33.249
not, you know, I am sure

979
00:36:33.250 --> 00:36:35.289
you're all like accustomed to things

980
00:36:35.290 --> 00:36:36.515
like state completion and

981
00:36:36.610 --> 00:36:37.933
refactoring code navigation

982
00:36:39.010 --> 00:36:40.629
and go to definition and so forth.

983
00:36:40.630 --> 00:36:42.129
And if you think about it, the

984
00:36:42.130 --> 00:36:44.169
things that the thing that powers

985
00:36:44.170 --> 00:36:45.444
that is semantic knowledge

986
00:36:46.300 --> 00:36:47.529
of your code. And the thing that

987
00:36:47.530 --> 00:36:49.196
provides the semantic knowledge of

988
00:36:49.420 --> 00:36:51.820
your code is a compiler

989
00:36:51.850 --> 00:36:52.850
with a type system.

990
00:36:53.830 --> 00:36:55.480
And once you add types,

991
00:36:55.900 --> 00:36:57.321
you can actually dramatically

992
00:36:58.090 --> 00:36:59.919
increase productivity, which

993
00:37:00.220 --> 00:37:01.690
in some ways seems counter

994
00:37:02.230 --> 00:37:03.099
counter intuitive.

995
00:37:03.100 --> 00:37:03.399
Right.

996
00:37:03.400 --> 00:37:04.929
I thought like dynamic languages

997
00:37:04.930 --> 00:37:06.010
were easier to

998
00:37:07.120 --> 00:37:08.590
to to approach because you got

999
00:37:09.040 --> 00:37:10.479
rid of the types which was like a

1000
00:37:10.480 --> 00:37:11.429
bother all the time.

1001
00:37:11.430 --> 00:37:13.139
Well, look, when it turns out that

1002
00:37:13.360 --> 00:37:15.190
you can actually be more productive

1003
00:37:15.220 --> 00:37:16.249
by by adding types if

1004
00:37:17.170 --> 00:37:19.179
you do it in a non-intrusive manner

1005
00:37:19.210 --> 00:37:20.800
and if you work hard on doing

1006
00:37:21.130 --> 00:37:22.959
good type inference and so forth.

1007
00:37:23.590 --> 00:37:24.590
So, so,

1008
00:37:25.960 --> 00:37:27.249
so, so, so, so.

1009
00:37:27.280 --> 00:37:28.989
So I want to jump in here.

1010
00:37:30.690 --> 00:37:32.610
And then I take

1011
00:37:32.640 --> 00:37:33.640
over.

1012
00:37:34.530 --> 00:37:35.853
So I really, really, really

1013
00:37:36.670 --> 00:37:38.385
believe in that point in the in the

1014
00:37:38.580 --> 00:37:39.580
in the power tools

1015
00:37:40.600 --> 00:37:41.800
for power geeks.

1016
00:37:44.440 --> 00:37:46.659
And one of the things that drives

1017
00:37:46.660 --> 00:37:48.639
me nuts is

1018
00:37:48.640 --> 00:37:50.579
the real man usvi

1019
00:37:51.600 --> 00:37:52.600
movement.

1020
00:37:52.860 --> 00:37:54.669
And, you know, it's it's really

1021
00:37:55.490 --> 00:37:57.219
I just want to punch people who

1022
00:37:57.580 --> 00:37:57.979
are.

1023
00:37:57.980 --> 00:38:00.039
But I'm really inside state.

1024
00:38:00.200 --> 00:38:01.670
I I'm a real developer because

1025
00:38:02.260 --> 00:38:03.309
I use v.i.

1026
00:38:03.420 --> 00:38:05.080
And it's like, you know,

1027
00:38:05.530 --> 00:38:06.530
I do it the hard way.

1028
00:38:06.700 --> 00:38:07.700
I think.

1029
00:38:07.900 --> 00:38:10.270
I think I make language developers

1030
00:38:10.300 --> 00:38:11.300
lazy.

1031
00:38:20.190 --> 00:38:21.219
IBRD Let me get a lot

1032
00:38:22.170 --> 00:38:23.909
more done a lot faster,

1033
00:38:24.100 --> 00:38:25.849
OK? I mean, I'm not I

1034
00:38:26.590 --> 00:38:27.590
I mean, I'm really

1035
00:38:28.560 --> 00:38:30.239
not into proving my manhood.

1036
00:38:30.240 --> 00:38:31.240
I'm good

1037
00:38:32.290 --> 00:38:33.780
at getting things done.

1038
00:38:36.260 --> 00:38:37.949
Case in point, but I

1039
00:38:38.180 --> 00:38:39.869
take the devil's advocate here first

1040
00:38:39.890 --> 00:38:41.899
said he meant something like.

1041
00:38:43.670 --> 00:38:45.360
Yeah, but

1042
00:38:45.490 --> 00:38:47.599
you their notebooks, a lot

1043
00:38:47.600 --> 00:38:48.969
of science people.

1044
00:38:49.010 --> 00:38:51.249
Data scientists get a lot done.

1045
00:38:51.970 --> 00:38:54.519
It's actually a pretty simplistic

1046
00:38:55.410 --> 00:38:56.789
I need

1047
00:38:57.890 --> 00:38:59.270
a dynamic language.

1048
00:38:59.300 --> 00:39:00.300
Most of the time.

1049
00:39:00.820 --> 00:39:02.490
So you could get a lot more than

1050
00:39:03.510 --> 00:39:04.510
that.

1051
00:39:05.600 --> 00:39:06.600
I don't know.

1052
00:39:06.940 --> 00:39:08.309
No, but I think typeset.

1053
00:39:08.420 --> 00:39:09.739
I mean, it's not just

1054
00:39:11.350 --> 00:39:12.669
a type system can help you.

1055
00:39:12.680 --> 00:39:14.479
First of all, if you're uninitiated

1056
00:39:14.700 --> 00:39:16.549
and you want to know here, here's my

1057
00:39:16.550 --> 00:39:18.069
food and now I say food dot and

1058
00:39:18.510 --> 00:39:19.739
then the ivy can show you.

1059
00:39:19.760 --> 00:39:20.909
What can I type next?

1060
00:39:20.930 --> 00:39:22.694
Right. As opposed to I gotta go read

1061
00:39:22.910 --> 00:39:24.400
manuals and figure it out or know

1062
00:39:24.410 --> 00:39:26.329
it. All right. I mean, your original

1063
00:39:26.330 --> 00:39:28.309
developer of whatever piece of code

1064
00:39:28.340 --> 00:39:29.418
might know it all, but

1065
00:39:30.320 --> 00:39:32.179
then people move

1066
00:39:32.180 --> 00:39:33.829
on and new people come in.

1067
00:39:34.190 --> 00:39:35.419
You know, there's here's this

1068
00:39:35.420 --> 00:39:36.409
project.

1069
00:39:36.410 --> 00:39:38.300
There was there was no documentation

1070
00:39:38.330 --> 00:39:39.169
written about it.

1071
00:39:39.170 --> 00:39:40.729
And if you think about it, types or

1072
00:39:40.730 --> 00:39:42.649
documentation to write, I

1073
00:39:42.650 --> 00:39:44.267
mean, and then there just there's

1074
00:39:44.630 --> 00:39:45.630
so many things about

1075
00:39:46.550 --> 00:39:47.550
like

1076
00:39:48.680 --> 00:39:50.329
like adding them that

1077
00:39:50.840 --> 00:39:52.579
down the line give you

1078
00:39:52.670 --> 00:39:54.070
increased productivity.

1079
00:39:54.880 --> 00:39:56.540
So there's productivity

1080
00:39:57.030 --> 00:39:58.309
and then that's that's

1081
00:39:58.310 --> 00:39:59.599
maintainability.

1082
00:40:00.080 --> 00:40:01.080
Oh, well both.

1083
00:40:01.100 --> 00:40:02.570
And maybe

1084
00:40:03.440 --> 00:40:05.340
any different thoughts about how

1085
00:40:05.350 --> 00:40:06.739
you get to talk about types.

1086
00:40:07.110 --> 00:40:08.110
Well, talk.

1087
00:40:09.630 --> 00:40:10.820
The colonel types out.

1088
00:40:11.450 --> 00:40:13.329
Well, that story is very

1089
00:40:13.330 --> 00:40:14.949
different for profile purposes.

1090
00:40:16.920 --> 00:40:18.586
As you know, Girlfight, to sort of

1091
00:40:18.610 --> 00:40:20.679
rural over time and the whole idea

1092
00:40:20.870 --> 00:40:22.179
was sort of particularly idea.

1093
00:40:22.570 --> 00:40:23.619
Everything you can pretend

1094
00:40:23.620 --> 00:40:25.059
everything is a straight even if

1095
00:40:25.060 --> 00:40:26.800
it's a number or floating point

1096
00:40:27.010 --> 00:40:28.909
internally, it's stalled, you

1097
00:40:28.920 --> 00:40:29.920
know, interchangeable.

1098
00:40:30.340 --> 00:40:32.229
And that works out for

1099
00:40:32.620 --> 00:40:34.059
bootstrapping people into a

1100
00:40:34.060 --> 00:40:35.739
language. They're quite nicely.

1101
00:40:37.240 --> 00:40:39.099
And it's it's a feature that we

1102
00:40:39.100 --> 00:40:40.209
wanted to keep in Perl 6.

1103
00:40:40.210 --> 00:40:41.889
But what we discovered in Perl 5 is

1104
00:40:41.890 --> 00:40:43.720
part of the redesign was

1105
00:40:45.010 --> 00:40:46.749
it's it's fine if the

1106
00:40:47.080 --> 00:40:48.999
user is confused about the

1107
00:40:49.000 --> 00:40:50.699
interchangeability to tell more for

1108
00:40:50.720 --> 00:40:52.179
the Muslim to

1109
00:40:52.550 --> 00:40:54.579
take the care of

1110
00:40:54.590 --> 00:40:55.249
your standard.

1111
00:40:55.250 --> 00:40:57.099
Tikes But it's not so

1112
00:40:57.100 --> 00:40:58.809
good if the computer is

1113
00:40:59.170 --> 00:41:00.170
confused about.

1114
00:41:01.930 --> 00:41:02.959
You know, for all the

1115
00:41:03.820 --> 00:41:05.859
girl wanted, it was written way

1116
00:41:05.860 --> 00:41:06.860
back in the dark ages.

1117
00:41:06.960 --> 00:41:08.319
You had a schuberg, a bunch of stuff

1118
00:41:08.320 --> 00:41:09.830
in, and it

1119
00:41:10.600 --> 00:41:11.600
cheated a lot.

1120
00:41:11.980 --> 00:41:14.199
So part of the redesign

1121
00:41:14.350 --> 00:41:16.059
when we did girls school Perl 6

1122
00:41:16.060 --> 00:41:17.170
thing we wanted to do

1123
00:41:18.160 --> 00:41:19.930
object oriented programing

1124
00:41:20.000 --> 00:41:21.568
that is that these languages and

1125
00:41:22.000 --> 00:41:22.959
we wanted to be functional

1126
00:41:22.960 --> 00:41:24.249
programing that are the most

1127
00:41:24.250 --> 00:41:25.449
functional programing languages.

1128
00:41:26.260 --> 00:41:28.149
To do that you have to have a

1129
00:41:28.420 --> 00:41:30.450
fundamental very sound type

NOTE 1:30:00 in the transcript
also note the speaker changes here
but trint did not identify that

1130
00:41:30.460 --> 00:41:32.439
system of a sound object

1131
00:41:32.890 --> 00:41:33.890
model underneath.

1132
00:41:34.330 --> 00:41:36.189
And you really have

1133
00:41:36.190 --> 00:41:38.169
to take seriously these slogans

1134
00:41:38.170 --> 00:41:39.429
like everything is an object.

1135
00:41:39.830 --> 00:41:40.929
Everything is like closure.

1136
00:41:41.250 --> 00:41:43.149
Every feel the process,

1137
00:41:43.390 --> 00:41:45.539
even blocks are closures.

1138
00:41:46.300 --> 00:41:48.189
And he has got to

1139
00:41:48.200 --> 00:41:49.579
optimize. Hard to get up, get them

1140
00:41:49.690 --> 00:41:50.690
away from there.

1141
00:41:50.720 --> 00:41:52.629
But but I I also

1142
00:41:52.630 --> 00:41:54.850
agree with this idea

1143
00:41:54.880 --> 00:41:56.379
that the tikes also

1144
00:41:56.710 --> 00:41:57.710
are cultural.

1145
00:41:58.540 --> 00:41:59.900
I talked about Pei's hanging things

1146
00:41:59.920 --> 00:42:01.629
on pace. They're one of the one of

1147
00:42:01.630 --> 00:42:02.919
the picks. We didn't have something

1148
00:42:02.920 --> 00:42:04.149
to hang on the prettified.

1149
00:42:04.510 --> 00:42:06.149
And now we we can Pangle.

1150
00:42:06.190 --> 00:42:07.869
But all of that information

1151
00:42:08.070 --> 00:42:09.430
and then just run a little program

1152
00:42:10.070 --> 00:42:11.349
there. And then you have to have an

1153
00:42:11.350 --> 00:42:13.090
idea. They say, well, this object.

1154
00:42:13.920 --> 00:42:15.159
You know what methods.

1155
00:42:15.190 --> 00:42:16.719
But what you know, they could do

1156
00:42:16.720 --> 00:42:17.709
their own introspection.

1157
00:42:17.710 --> 00:42:18.710
The whole thing.

1158
00:42:18.730 --> 00:42:20.079
And people just you know, I'm

1159
00:42:21.060 --> 00:42:23.079
baffled by the way we do also

1160
00:42:23.080 --> 00:42:24.080
have an idea.

1161
00:42:27.420 --> 00:42:28.829
Good points all around.

1162
00:42:29.670 --> 00:42:31.649
So most

1163
00:42:31.650 --> 00:42:33.218
of the time programmers actually

1164
00:42:33.480 --> 00:42:35.610
spend maintaining versus

1165
00:42:35.670 --> 00:42:36.969
writing new code

1166
00:42:38.350 --> 00:42:39.690
out in the wild and.

1167
00:42:41.430 --> 00:42:43.189
What things or elements of a

1168
00:42:43.190 --> 00:42:45.020
programing language make it

1169
00:42:45.080 --> 00:42:46.880
easier to maintain?

1170
00:42:47.460 --> 00:42:48.850
And maybe we'll start with Fito.

1171
00:42:52.930 --> 00:42:54.730
Well, I found that.

1172
00:42:58.650 --> 00:42:59.650
TypeScript

1173
00:43:00.660 --> 00:43:02.630
is actually incredibly useful,

1174
00:43:02.640 --> 00:43:05.010
and so we're adding a very similar

1175
00:43:05.100 --> 00:43:06.949
idea to Python, adding it

1176
00:43:07.000 --> 00:43:09.039
in a slightly good way because.

1177
00:43:09.870 --> 00:43:11.460
We are taxed.

1178
00:43:28.950 --> 00:43:29.929
All right.

1179
00:43:29.930 --> 00:43:30.769
Go.

1180
00:43:30.770 --> 00:43:31.770
Anyway,

1181
00:43:33.050 --> 00:43:34.050
we have time.

1182
00:43:36.480 --> 00:43:37.699
I sort of

1183
00:43:38.480 --> 00:43:39.656
learned a painful lesson

1184
00:43:40.550 --> 00:43:42.859
of that before school programs.

1185
00:43:43.160 --> 00:43:45.409
Seafair dynamic typing

1186
00:43:45.410 --> 00:43:47.570
is great for large programs.

1187
00:43:47.600 --> 00:43:48.600
You have to

1188
00:43:49.520 --> 00:43:51.380
have a more disciplined

1189
00:43:51.410 --> 00:43:52.782
approach and it helps if the

1190
00:43:53.300 --> 00:43:54.559
language actually

1191
00:43:55.430 --> 00:43:56.209
gives you that.

1192
00:43:56.210 --> 00:43:58.159
It's rather than telling you,

1193
00:43:58.160 --> 00:43:59.469
well, you can do whatever you want.

1194
00:44:02.030 --> 00:44:02.979
Yeah, yeah.

1195
00:44:02.980 --> 00:44:04.440
That was part of our scale up

1196
00:44:04.990 --> 00:44:06.800
idea for Perl 6,

1197
00:44:07.170 --> 00:44:08.170
the typist would help

1198
00:44:09.100 --> 00:44:10.929
with programing in large because

1199
00:44:10.930 --> 00:44:12.550
we didn't notice those limitations.

1200
00:44:16.600 --> 00:44:18.119
One year ago, I started working

1201
00:44:18.130 --> 00:44:20.300
really heavily on Unbury

1202
00:44:20.310 --> 00:44:21.310
Factoring In.

1203
00:44:22.040 --> 00:44:23.040
And,

1204
00:44:24.140 --> 00:44:25.369
you know, being able to

1205
00:44:26.200 --> 00:44:27.909
do like large scale refactoring

1206
00:44:28.110 --> 00:44:30.219
is unlike millions of lines

1207
00:44:30.220 --> 00:44:31.540
of code at once,

1208
00:44:33.430 --> 00:44:34.749
that really changes the way you

1209
00:44:34.750 --> 00:44:36.309
think about maintaining code.

1210
00:44:37.200 --> 00:44:38.709
Because, you know, the world is

1211
00:44:38.710 --> 00:44:40.179
filled with libraries that have

1212
00:44:40.180 --> 00:44:41.503
things like stupid variable

1213
00:44:42.070 --> 00:44:43.929
names, stupid method

1214
00:44:43.930 --> 00:44:45.760
names, because,

1215
00:44:45.850 --> 00:44:47.049
you know, when you started out, it

1216
00:44:47.050 --> 00:44:48.050
meant one thing.

1217
00:44:48.520 --> 00:44:50.559
But now you realize that

1218
00:44:50.680 --> 00:44:52.269
you really didn't really understand

1219
00:44:52.270 --> 00:44:53.109
what you were doing in the

1220
00:44:53.110 --> 00:44:54.433
beginning. And so this this

1221
00:44:54.940 --> 00:44:57.010
method name really should be

1222
00:44:57.070 --> 00:44:58.070
different.

1223
00:44:58.180 --> 00:45:00.480
But nobody ever Reding's

1224
00:45:00.490 --> 00:45:01.813
methods because it's really

1225
00:45:02.440 --> 00:45:04.659
hard to go over a piece

1226
00:45:05.070 --> 00:45:07.030
of code and

1227
00:45:07.120 --> 00:45:08.780
rename. Exactly.

1228
00:45:08.960 --> 00:45:11.270
That's the right variable.

1229
00:45:11.370 --> 00:45:12.819
But if you've got a good refactoring

1230
00:45:12.820 --> 00:45:14.769
engine, you just you

1231
00:45:14.770 --> 00:45:16.436
just type, you know, medic control

1232
00:45:16.570 --> 00:45:18.340
are typing the new name,

1233
00:45:19.060 --> 00:45:20.829
you know, hundreds and hundreds of

1234
00:45:20.830 --> 00:45:22.660
files later, which takes about,

1235
00:45:23.080 --> 00:45:24.910
you know, maybe 30 seconds

1236
00:45:24.920 --> 00:45:26.289
if it's a lot of files.

1237
00:45:29.100 --> 00:45:30.100
You're done.

1238
00:45:30.710 --> 00:45:32.869
And so you basically be right.

1239
00:45:34.410 --> 00:45:35.909
Seven-part series,

1240
00:45:36.980 --> 00:45:37.980
I

1241
00:45:39.090 --> 00:45:41.070
hope this is this is actually like

1242
00:45:41.220 --> 00:45:42.919
you're describing, that's what

1243
00:45:43.080 --> 00:45:43.599
we what.

1244
00:45:43.600 --> 00:45:44.730
What was the genesis

1245
00:45:45.770 --> 00:45:47.790
of the typescript project, which was

1246
00:45:48.540 --> 00:45:50.590
these enormous JavaScript code basis

1247
00:45:50.640 --> 00:45:52.019
that we were seeing customers.

1248
00:45:52.020 --> 00:45:53.999
Right. And then in-house projects

1249
00:45:54.010 --> 00:45:55.259
that they were getting bigger and

1250
00:45:55.260 --> 00:45:56.789
bigger because JavaScript engines

1251
00:45:56.790 --> 00:45:58.139
were getting good enough for you to

1252
00:45:58.140 --> 00:45:59.669
write really big programs.

1253
00:46:00.200 --> 00:46:02.160
But maintaining them was impossible

1254
00:46:02.790 --> 00:46:04.819
because it turned into write

1255
00:46:04.860 --> 00:46:06.639
only code. You know, you do dare

1256
00:46:06.700 --> 00:46:08.159
not touch it. Once you've written it

1257
00:46:08.190 --> 00:46:10.020
because it worked then and then.

1258
00:46:10.350 --> 00:46:12.059
But, you know, there's this property

1259
00:46:12.060 --> 00:46:13.060
called text.

1260
00:46:13.440 --> 00:46:15.840
And I really want to change it to be

1261
00:46:16.400 --> 00:46:18.329
like some

1262
00:46:18.330 --> 00:46:20.159
other name, except there

1263
00:46:20.160 --> 00:46:21.479
is like a million other properties

1264
00:46:21.480 --> 00:46:23.244
called text. And if I'd just like to

1265
00:46:23.310 --> 00:46:24.639
do a global search and replace for

1266
00:46:24.640 --> 00:46:26.257
tax, then oh my God, it's like at

1267
00:46:26.460 --> 00:46:28.339
all. So I can't change anything.

1268
00:46:28.920 --> 00:46:30.599
But if you have semantic

1269
00:46:30.600 --> 00:46:32.315
understanding up like this property

1270
00:46:32.400 --> 00:46:34.019
call text is different from that

1271
00:46:34.050 --> 00:46:35.099
other property over here.

1272
00:46:35.100 --> 00:46:36.689
Call text. And if you want to rename

1273
00:46:36.690 --> 00:46:37.829
this one, that doesn't mean you want

1274
00:46:37.830 --> 00:46:38.830
to rename that one.

1275
00:46:39.080 --> 00:46:40.080
But but that takes

1276
00:46:41.010 --> 00:46:42.869
for something like a

1277
00:46:42.870 --> 00:46:44.639
type system to be in place.

1278
00:46:44.690 --> 00:46:46.454
And, you know, once you start adding

1279
00:46:46.800 --> 00:46:48.629
that, you add documentation to the

1280
00:46:48.630 --> 00:46:50.420
code and you add these capabilities

1281
00:46:50.430 --> 00:46:52.110
where we're now seeing people

1282
00:46:53.020 --> 00:46:55.139
like with regularity, refactor

1283
00:46:55.140 --> 00:46:56.699
one hundreds of thousands of lines

1284
00:46:56.700 --> 00:46:58.109
of JavaScript code, you know, we'd

1285
00:46:58.110 --> 00:46:59.730
like ten minutes

1286
00:47:00.280 --> 00:47:01.652
and end it and it just works

1287
00:47:02.280 --> 00:47:03.719
afterwards. And people are like

1288
00:47:03.720 --> 00:47:04.980
amazed that it's possible.

1289
00:47:06.130 --> 00:47:07.130
But but but

1290
00:47:08.040 --> 00:47:09.755
it's true. Said it takes a bit more

1291
00:47:10.050 --> 00:47:11.039
work to begin with.

1292
00:47:11.040 --> 00:47:12.420
And it's not maybe not right.

1293
00:47:12.450 --> 00:47:13.919
If you're writing five lines of code

1294
00:47:14.340 --> 00:47:15.989
that maybe more bother than you

1295
00:47:15.990 --> 00:47:16.979
really care to.

1296
00:47:16.980 --> 00:47:18.548
But but, you know, it started to

1297
00:47:18.840 --> 00:47:19.920
pay off pretty quickly.

1298
00:47:21.290 --> 00:47:22.710
There are different types

1299
00:47:23.280 --> 00:47:25.380
of really good lexical scoping

1300
00:47:25.710 --> 00:47:26.829
helps with refactoring.

1301
00:47:27.500 --> 00:47:29.400
It gets

1302
00:47:29.430 --> 00:47:30.430
back into this.

1303
00:47:31.410 --> 00:47:32.449
What's the right pick to hang the

1304
00:47:32.450 --> 00:47:33.820
sign up and off and it's slope.

1305
00:47:34.110 --> 00:47:35.820
It's a lexically skulked or

1306
00:47:35.840 --> 00:47:36.840
dynamically scale.

1307
00:47:36.850 --> 00:47:38.909
So now we're doing those right

1308
00:47:38.940 --> 00:47:40.369
and not interfering with threading.

1309
00:47:43.200 --> 00:47:45.269
So I have one more question

1310
00:47:45.270 --> 00:47:47.149
now before the break in

1311
00:47:47.150 --> 00:47:48.979
three minutes and we made

1312
00:47:49.050 --> 00:47:50.989
sort of a reference to Donald New,

1313
00:47:51.570 --> 00:47:52.920
one of the turning winners

1314
00:47:54.360 --> 00:47:55.739
in this last

1315
00:47:56.310 --> 00:47:57.840
section of conversation.

1316
00:47:57.930 --> 00:47:59.939
And he wrote, Premature

1317
00:47:59.970 --> 00:48:01.489
optimization is the root of all

1318
00:48:01.980 --> 00:48:02.980
evil.

1319
00:48:03.870 --> 00:48:04.997
So what's your approach

1320
00:48:05.880 --> 00:48:07.979
to optimizing code and gaining

1321
00:48:07.980 --> 00:48:09.860
efficiency, saying Python?

1322
00:48:12.520 --> 00:48:14.440
Well, I I'm terrible at that.

1323
00:48:14.470 --> 00:48:15.939
So I leave it to others.

1324
00:48:17.110 --> 00:48:18.110
That is

1325
00:48:19.410 --> 00:48:21.249
a very wise language creator

1326
00:48:21.910 --> 00:48:23.037
and they've lived up to

1327
00:48:23.860 --> 00:48:24.860
the task.

1328
00:48:26.200 --> 00:48:28.820
Piceance hash table implementation

1329
00:48:28.850 --> 00:48:30.980
has been rewritten so many times.

1330
00:48:32.590 --> 00:48:34.239
I don't understand any part of it,

1331
00:48:36.380 --> 00:48:38.229
but it is so much better than the

1332
00:48:38.230 --> 00:48:40.409
thing I actually copied out of truth

1333
00:48:41.230 --> 00:48:42.230
30 years ago.

1334
00:48:44.560 --> 00:48:45.729
Happy to have a clue.

1335
00:48:46.830 --> 00:48:47.870
Okey doke.

1336
00:48:52.790 --> 00:48:54.859
I kind of 50 percent believe that

1337
00:48:56.030 --> 00:48:57.889
because they're, you

1338
00:48:58.020 --> 00:48:59.809
know, the premature optimization in

1339
00:48:59.810 --> 00:49:01.369
terms of algorithms and code.

1340
00:49:01.910 --> 00:49:03.229
That part of it, I believe.

1341
00:49:04.280 --> 00:49:05.960
But people

1342
00:49:06.410 --> 00:49:07.939
often don't really think about the

1343
00:49:07.940 --> 00:49:09.349
role of data structures

1344
00:49:09.970 --> 00:49:11.030
in optimization

1345
00:49:11.900 --> 00:49:13.869
and interpret data structures

1346
00:49:13.940 --> 00:49:14.940
exposed.

1347
00:49:16.730 --> 00:49:17.780
You have to

1348
00:49:18.870 --> 00:49:20.420
know and you decide that, you know,

1349
00:49:20.990 --> 00:49:22.789
this algorithm is a problem.

1350
00:49:23.630 --> 00:49:25.369
But, you know, the reason that it's

1351
00:49:25.370 --> 00:49:27.169
a problem is because,

1352
00:49:27.800 --> 00:49:29.449
you know, this data structure was

1353
00:49:29.450 --> 00:49:30.920
just slapped together and they

1354
00:49:31.310 --> 00:49:32.719
thought, oh, they'd only ever be

1355
00:49:32.720 --> 00:49:34.610
like 10 items in the list

1356
00:49:35.090 --> 00:49:36.550
or 10 items in the set.

1357
00:49:37.220 --> 00:49:38.869
And then and then you start,

1358
00:49:39.130 --> 00:49:41.329
you know, having, you know, million

1359
00:49:41.330 --> 00:49:43.159
eight million items sets all the

1360
00:49:43.160 --> 00:49:44.160
time.

1361
00:49:44.300 --> 00:49:45.770
And all of a sudden, you know,

1362
00:49:45.800 --> 00:49:47.170
you're a little linkedlist

1363
00:49:47.660 --> 00:49:49.520
is. And that is not gonna happen.

1364
00:49:50.270 --> 00:49:50.819
Right.

1365
00:49:50.820 --> 00:49:52.600
Ex-directory or a unit

1366
00:49:52.640 --> 00:49:53.640
or. Yeah.

1367
00:49:53.910 --> 00:49:55.989
And and

1368
00:49:56.030 --> 00:49:57.980
so and so if you can you know,

1369
00:49:58.190 --> 00:50:00.320
it's like you know hyd

1370
00:50:00.350 --> 00:50:01.219
early hide.

1371
00:50:01.220 --> 00:50:02.220
Often.

1372
00:50:02.400 --> 00:50:03.400
Right.

1373
00:50:03.590 --> 00:50:05.449
Never tell the truth about what your

1374
00:50:05.450 --> 00:50:06.649
data structures are.

1375
00:50:09.320 --> 00:50:11.179
And on that note, we're

1376
00:50:11.180 --> 00:50:12.180
going to take

1377
00:50:13.100 --> 00:50:14.300
a quick look.

1378
00:50:14.330 --> 00:50:16.579
I just want to say that I

1379
00:50:16.580 --> 00:50:18.230
think like

1380
00:50:18.460 --> 00:50:20.479
with anything, the important thing

1381
00:50:20.480 --> 00:50:22.097
is to make sure that you actually

1382
00:50:22.580 --> 00:50:24.429
have the right data before

1383
00:50:24.430 --> 00:50:26.200
you you start optimizing.

1384
00:50:26.210 --> 00:50:28.099
Right. By too often you have

1385
00:50:28.100 --> 00:50:29.900
a hunch that, oh,

1386
00:50:29.930 --> 00:50:31.789
I think finally I need to do

1387
00:50:31.790 --> 00:50:32.219
this.

1388
00:50:32.220 --> 00:50:33.469
And I need to change this data

1389
00:50:33.470 --> 00:50:34.969
structure from being a linkedlist to

1390
00:50:34.970 --> 00:50:36.349
being overweight or whatever.

1391
00:50:36.980 --> 00:50:38.899
But but if you

1392
00:50:38.900 --> 00:50:40.129
measure, then you discover it

1393
00:50:40.130 --> 00:50:41.119
doesn't matter.

1394
00:50:41.120 --> 00:50:42.949
And but really, there's this other

1395
00:50:42.950 --> 00:50:44.269
thing that you haven't even thought

1396
00:50:44.270 --> 00:50:46.099
about that matters enormously.

1397
00:50:46.100 --> 00:50:47.509
And maybe you should go look over

1398
00:50:47.510 --> 00:50:49.489
there. So get perf.,

1399
00:50:49.550 --> 00:50:51.479
get it, use a profile or

1400
00:50:51.480 --> 00:50:53.149
figure out where it is, the time

1401
00:50:53.150 --> 00:50:54.079
actually spent.

1402
00:50:54.080 --> 00:50:55.909
I am continually surprised and

1403
00:50:55.910 --> 00:50:57.439
I write compilers for a living and

1404
00:50:57.440 --> 00:50:59.239
I'm always surprised by where

1405
00:50:59.330 --> 00:51:00.330
time is spent.

1406
00:51:00.370 --> 00:51:01.039
All right.

1407
00:51:01.040 --> 00:51:02.960
And that is that

1408
00:51:03.110 --> 00:51:04.670
we're out of time for this section.

1409
00:51:05.030 --> 00:51:06.739
We're going to take a quick 15

1410
00:51:06.740 --> 00:51:07.740
minute break.

1411
00:51:09.080 --> 00:51:10.999
And when we come

1412
00:51:11.000 --> 00:51:12.029
back, the really hard

1413
00:51:12.920 --> 00:51:14.210
questions are going to come out

1414
00:51:14.270 --> 00:51:16.280
because the audience questions

1415
00:51:16.890 --> 00:51:18.709
that have been submitted are

1416
00:51:18.710 --> 00:51:20.260
going to be the ones that believe in

1417
00:51:20.380 --> 00:51:21.380
you.

1418
00:51:22.020 --> 00:51:23.020
So thank you.
[No audio until 51:55]


NOTE BEGIN PART 2
2:00:55 in the transcript

1419
00:51:55.280 --> 00:51:56.899
And teach in every one of you has a

1420
00:51:56.900 --> 00:51:58.419
community that is formed around

1421
00:51:58.760 --> 00:52:00.079
the languages that you've created.

1422
00:52:00.110 --> 00:52:02.389
I'm curious if you were talk two

1423
00:52:02.390 --> 00:52:03.979
ways in which you've seen the design

1424
00:52:03.980 --> 00:52:05.059
choices you've made for your

1425
00:52:05.060 --> 00:52:07.109
language, so shaping the communities

1426
00:52:07.110 --> 00:52:08.110
that are formed around.

1427
00:52:11.440 --> 00:52:12.449
It really helps.

1428
00:52:12.580 --> 00:52:14.379
Well, good.

1429
00:52:14.590 --> 00:52:15.610
It really helps

1430
00:52:16.870 --> 00:52:18.040
if you're trying to

1431
00:52:19.270 --> 00:52:21.190
make everybody look

1432
00:52:21.240 --> 00:52:22.690
what's here with your slogan here.

1433
00:52:24.380 --> 00:52:25.970
Everybody comfortable?

1434
00:52:28.070 --> 00:52:29.860
We're doing that right now.

1435
00:52:30.600 --> 00:52:32.349
They think they can make

1436
00:52:32.840 --> 00:52:34.769
all of all Americans, I heard.

1437
00:52:37.920 --> 00:52:39.299
You know, feel like comfortable

1438
00:52:39.360 --> 00:52:40.360
programing.

1439
00:52:40.870 --> 00:52:42.050
You see it for all.

1440
00:52:42.150 --> 00:52:43.969
Yes. To make sense of

1441
00:52:43.980 --> 00:52:45.059
it. I guess you know

1442
00:52:46.120 --> 00:52:47.120
what?

1443
00:52:48.190 --> 00:52:49.600
The sign was a clue.

1444
00:52:51.250 --> 00:52:52.769
It's not see us for all of us, just

1445
00:52:52.770 --> 00:52:53.770
Americans.

1446
00:52:54.140 --> 00:52:55.590
No. There are many underserved

1447
00:52:56.520 --> 00:52:57.639
populations in the world.

1448
00:52:57.960 --> 00:52:59.879
And what we've really discovered

1449
00:52:59.910 --> 00:53:01.979
that it really helps

1450
00:53:01.980 --> 00:53:03.390
build international

1451
00:53:06.540 --> 00:53:08.070
camaraderie and community

1452
00:53:08.700 --> 00:53:10.499
to have world class

1453
00:53:11.010 --> 00:53:12.650
literally support for unico.

1454
00:53:13.440 --> 00:53:15.155
Yes. I would encourage any language

1455
00:53:15.330 --> 00:53:16.349
designers out there.

1456
00:53:16.920 --> 00:53:18.579
Don't do this halfway stuff with,

1457
00:53:18.830 --> 00:53:20.760
you know, code points.

1458
00:53:21.200 --> 00:53:22.200
Go all the graphics.

1459
00:53:23.160 --> 00:53:25.469
I completely agree.

1460
00:53:26.850 --> 00:53:28.109
Guido, you had something you were

1461
00:53:28.110 --> 00:53:29.110
going to say.

1462
00:53:29.940 --> 00:53:31.950
I was trying to say.

1463
00:53:33.280 --> 00:53:34.590
I how my

1464
00:53:35.160 --> 00:53:37.360
design choices necessary

1465
00:53:38.390 --> 00:53:39.390
to make

1466
00:53:40.570 --> 00:53:42.520
choices about how to use

1467
00:53:43.000 --> 00:53:44.200
energy certainly have.

1468
00:53:46.260 --> 00:53:47.260
Did

1469
00:53:48.250 --> 00:53:49.250
I?

1470
00:53:49.390 --> 00:53:51.309
I was somehow a dude

1471
00:53:51.510 --> 00:53:53.269
with the importance of

1472
00:53:54.260 --> 00:53:56.119
user centered based something

1473
00:53:56.630 --> 00:53:58.219
because by comparison.

1474
00:53:59.000 --> 00:54:01.189
That was ABC long

1475
00:54:01.370 --> 00:54:02.370
lost,

1476
00:54:04.260 --> 00:54:06.409
its authors had a of things

1477
00:54:06.410 --> 00:54:08.209
right, and one of those was

1478
00:54:08.780 --> 00:54:10.219
listen to the users

1479
00:54:10.850 --> 00:54:12.560
asking users what they

1480
00:54:12.710 --> 00:54:14.559
think. And then again, you making

1481
00:54:14.680 --> 00:54:16.052
a new one of their suggested

1482
00:54:16.580 --> 00:54:17.359
solutions.

1483
00:54:17.360 --> 00:54:18.360
And you first figure

1484
00:54:19.190 --> 00:54:20.699
out what nobody spends.

1485
00:54:21.530 --> 00:54:23.330
You gave users a voice.

1486
00:54:23.450 --> 00:54:24.699
And I said,

1487
00:54:25.610 --> 00:54:26.884
take that back and let the

1488
00:54:27.440 --> 00:54:28.879
users help.

1489
00:54:30.680 --> 00:54:32.829
And now they're raising of.

1490
00:54:36.790 --> 00:54:38.379
I was going to say I always felt

1491
00:54:38.380 --> 00:54:39.380
like

1492
00:54:40.630 --> 00:54:42.140
it's important as a language society

1493
00:54:42.170 --> 00:54:43.787
that you don't create unnecessary

1494
00:54:44.140 --> 00:54:45.904
work for your community because it's

1495
00:54:46.030 --> 00:54:47.451
it's very easy to in the name

1496
00:54:47.860 --> 00:54:49.149
of purity or whatever.

1497
00:54:49.180 --> 00:54:51.309
Go change something

1498
00:54:51.310 --> 00:54:53.169
or or strive

1499
00:54:53.170 --> 00:54:54.500
for the perfect language.

1500
00:54:54.560 --> 00:54:55.870
And I've always said, like, I mean,

1501
00:54:57.770 --> 00:54:59.109
show me the perfect language and

1502
00:54:59.110 --> 00:55:00.249
I'll show you your language with no

1503
00:55:00.250 --> 00:55:01.250
use or shoot over.

1504
00:55:01.300 --> 00:55:03.189
Every every language

1505
00:55:03.190 --> 00:55:04.500
has its functions.

1506
00:55:05.280 --> 00:55:06.620
Has them because.

1507
00:55:07.260 --> 00:55:09.150
Because it has evolved.

1508
00:55:09.240 --> 00:55:10.955
And the hard part of language aside

1509
00:55:11.040 --> 00:55:12.689
is actually not version one.

1510
00:55:12.960 --> 00:55:14.219
It's like the older versions that

1511
00:55:14.220 --> 00:55:16.230
come after you

1512
00:55:16.580 --> 00:55:18.719
try to sneak things

1513
00:55:18.720 --> 00:55:20.969
in without actually

1514
00:55:21.210 --> 00:55:23.249
causing extra burdens on

1515
00:55:23.250 --> 00:55:24.689
people who don't necessarily care

1516
00:55:24.690 --> 00:55:25.829
about the things that you're trying

1517
00:55:25.830 --> 00:55:26.699
to sneak in. Right.

1518
00:55:26.700 --> 00:55:27.629
And so.

1519
00:55:27.630 --> 00:55:29.550
So that's the hard part

1520
00:55:29.910 --> 00:55:31.590
of language design, I think

1521
00:55:31.770 --> 00:55:33.719
is is the evolution

1522
00:55:33.810 --> 00:55:34.810
of the language and

1523
00:55:35.730 --> 00:55:37.396
keeping your community with you as

1524
00:55:37.500 --> 00:55:38.500
you have all.

1525
00:55:40.880 --> 00:55:42.630
Yeah. And so and so with with

1526
00:55:42.930 --> 00:55:43.930
Jarbawi like always

1527
00:55:45.150 --> 00:55:47.099
followed that we've

1528
00:55:47.100 --> 00:55:48.717
maybe been a little bit excessive

1529
00:55:49.590 --> 00:55:50.590
about it.

1530
00:55:51.420 --> 00:55:52.890
I mean, I've got binaries that

1531
00:55:53.250 --> 00:55:54.570
I compiled like

1532
00:55:55.680 --> 00:55:56.820
20 years ago,

1533
00:55:57.550 --> 00:55:58.650
but still run.

1534
00:55:59.490 --> 00:56:01.349
And and it's it's, you

1535
00:56:01.350 --> 00:56:02.967
know, on machines that were never

1536
00:56:03.120 --> 00:56:05.159
invented when the code was compiled.

1537
00:56:07.200 --> 00:56:09.380
And you only get there with

1538
00:56:09.390 --> 00:56:11.070
with crazy discipline

1539
00:56:11.640 --> 00:56:12.669
and and a a tolerance

1540
00:56:14.010 --> 00:56:15.750
for living with your mistakes.

1541
00:56:17.420 --> 00:56:19.260
But but it but it makes the

1542
00:56:19.420 --> 00:56:20.420
you.

1543
00:56:20.460 --> 00:56:21.329
It also help.

1544
00:56:21.330 --> 00:56:23.159
It also ends up selecting who

1545
00:56:23.160 --> 00:56:24.659
your users are

1546
00:56:25.210 --> 00:56:26.789
because in the in the Java universe,

1547
00:56:26.820 --> 00:56:28.143
pretty much everybody is is

1548
00:56:28.680 --> 00:56:30.599
really disciplined because,

1549
00:56:30.630 --> 00:56:31.829
you know, it's kind of like like

1550
00:56:31.920 --> 00:56:32.920
mountain climbing.

1551
00:56:32.950 --> 00:56:34.077
You know, you you don't

1552
00:56:34.770 --> 00:56:36.599
get there, get

1553
00:56:36.600 --> 00:56:38.217
sloppy with your gear when you're

1554
00:56:38.340 --> 00:56:39.340
mountain climbing

1555
00:56:40.700 --> 00:56:42.599
because it has a clear price.

1556
00:56:45.030 --> 00:56:47.039
And, you

1557
00:56:47.040 --> 00:56:48.167
know, when you you said

1558
00:56:48.930 --> 00:56:49.930
something about

1559
00:56:51.380 --> 00:56:52.619
Larry, you said it said something

1560
00:56:52.620 --> 00:56:53.796
about Unicode Java's had

1561
00:56:54.570 --> 00:56:56.099
Unicode since ninety two.

1562
00:56:56.700 --> 00:56:58.500
So, you know, it's,

1563
00:56:58.580 --> 00:56:59.990
you know, trying to be inclusive.

1564
00:57:01.110 --> 00:57:03.030
Yeah. And and that

1565
00:57:03.090 --> 00:57:05.070
really made a difference.

1566
00:57:05.850 --> 00:57:07.565
I mean lots of folks thought it was

1567
00:57:07.590 --> 00:57:09.419
like the lamest thing ever.

1568
00:57:10.200 --> 00:57:11.200
But

1569
00:57:12.240 --> 00:57:14.099
that was that that was during

1570
00:57:14.100 --> 00:57:15.330
my my, you know,

1571
00:57:16.230 --> 00:57:17.230
Lord of Java

1572
00:57:18.060 --> 00:57:19.060
phase which.

1573
00:57:20.190 --> 00:57:21.190
Yeah.

1574
00:57:21.460 --> 00:57:23.224
Yeah. But that that phase of my life

1575
00:57:23.400 --> 00:57:24.829
ended in about ninety five.

1576
00:57:25.800 --> 00:57:26.619
I'd like to say something

1577
00:57:26.620 --> 00:57:27.790
possibilitythat.

1578
00:57:30.150 --> 00:57:32.159
And that is we did this

1579
00:57:32.900 --> 00:57:34.769
perfect try to perfect

1580
00:57:35.310 --> 00:57:36.310
compatibility.

1581
00:57:36.540 --> 00:57:37.540
All the way through.

1582
00:57:37.560 --> 00:57:38.630
I didn't really

1583
00:57:40.200 --> 00:57:41.200
think.

1584
00:57:45.360 --> 00:57:47.099
That eventually runs into

1585
00:57:47.850 --> 00:57:49.569
the problem. But, you know, you

1586
00:57:49.620 --> 00:57:51.300
can't I can't fix your mistakes.

1587
00:57:51.930 --> 00:57:53.909
So one of the mistakes

1588
00:57:54.120 --> 00:57:55.640
that we made was was kind of a

1589
00:57:55.740 --> 00:57:57.389
mistake and that was assuming

1590
00:57:58.200 --> 00:58:00.179
that you had to always have

1591
00:58:00.180 --> 00:58:02.219
that kind of stability

1592
00:58:02.460 --> 00:58:03.630
and not change anything.

1593
00:58:04.890 --> 00:58:07.020
How would it be to design a language

1594
00:58:07.500 --> 00:58:09.089
where the language itself

1595
00:58:10.180 --> 00:58:12.120
could be evolved

1596
00:58:12.270 --> 00:58:13.270
from within

1597
00:58:14.290 --> 00:58:16.140
by the community

1598
00:58:16.620 --> 00:58:17.699
and the language we want?

1599
00:58:17.930 --> 00:58:19.639
Twenty five or 50 years from now,

1600
00:58:19.650 --> 00:58:21.449
this ties into what Polygram was

1601
00:58:21.450 --> 00:58:22.559
talking about, a hundred year

1602
00:58:22.590 --> 00:58:24.209
language. We kind of took that

1603
00:58:24.210 --> 00:58:25.210
seriously.

1604
00:58:26.040 --> 00:58:27.040
What is it that

1605
00:58:27.990 --> 00:58:29.879
prevents languages from evolving,

1606
00:58:30.490 --> 00:58:32.589
not being not having things still

1607
00:58:32.610 --> 00:58:33.610
thrive?

1608
00:58:34.710 --> 00:58:35.710
But

1609
00:58:36.630 --> 00:58:38.296
fundamentally, the Perl 6 compiler

1610
00:58:38.760 --> 00:58:40.050
is written in Perl 6.

1611
00:58:40.080 --> 00:58:41.789
Most the runtime system is written

1612
00:58:41.790 --> 00:58:42.769
in Perl 6.

1613
00:58:42.770 --> 00:58:44.099
The users can extend it.

1614
00:58:45.150 --> 00:58:47.219
Perl 6 itself does not care

1615
00:58:47.580 --> 00:58:49.260
whether things are built in or not.

1616
00:58:49.890 --> 00:58:51.099
It all works the same.

1617
00:58:51.450 --> 00:58:53.189
So most of the buildings are written

1618
00:58:53.190 --> 00:58:54.300
as if they were used or code.

1619
00:58:55.020 --> 00:58:57.269
So users, even if we say

1620
00:58:57.460 --> 00:58:58.460
no, we don't

1621
00:58:59.280 --> 00:59:00.420
want that language right now.

1622
00:59:00.450 --> 00:59:02.069
But put it in a module

1623
00:59:03.090 --> 00:59:04.090
you can do.

1624
00:59:04.140 --> 00:59:05.904
You can turn classes in two monitors

1625
00:59:06.090 --> 00:59:07.419
with about that much code.

1626
00:59:07.770 --> 00:59:09.499
You can put into implement actor

1627
00:59:09.520 --> 00:59:10.990
model classes with about about

1628
00:59:11.550 --> 00:59:12.550
how much code

1629
00:59:14.070 --> 00:59:15.269
you can you take.

1630
00:59:15.510 --> 00:59:17.399
Adding operators is just trivial.

1631
00:59:18.690 --> 00:59:20.579
When it does, it

1632
00:59:20.820 --> 00:59:22.979
does everything perfectly one pass

1633
00:59:22.980 --> 00:59:24.530
and drops into the sub compiler.

1634
00:59:24.960 --> 00:59:26.590
And with the end, it comes back out.

1635
00:59:26.880 --> 00:59:28.869
So there's never, ever

1636
00:59:28.900 --> 00:59:30.517
any ambiguity as to what language

1637
00:59:31.020 --> 00:59:32.020
you are in.

1638
00:59:32.820 --> 00:59:34.469
And so

1639
00:59:35.430 --> 00:59:37.210
because this scoping

1640
00:59:37.350 --> 00:59:38.879
of the exact language you're in is

1641
00:59:38.880 --> 00:59:40.889
always like if we scoped

1642
00:59:41.010 --> 00:59:42.510
any lexical scope can say

1643
00:59:43.290 --> 00:59:45.119
use whatever language I want a

1644
00:59:45.120 --> 00:59:46.760
feature version of Perl 6.

1645
00:59:46.860 --> 00:59:49.079
You could say use COBOL, use Python.

1646
00:59:49.380 --> 00:59:50.909
If someone actually people have

1647
00:59:50.910 --> 00:59:52.739
written Python parsers in

1648
00:59:52.870 --> 00:59:54.480
April 6, I don't think anybody's

1649
00:59:54.510 --> 00:59:55.619
going to call for sure yet,

1650
00:59:57.390 --> 00:59:58.899
but we'd like to think that, you

1651
00:59:58.900 --> 01:00:00.820
know, a lot of language

1652
01:00:00.890 --> 01:00:02.279
stuff like languages and Perl 6.

1653
01:00:03.030 --> 01:00:04.050
So I think that

1654
01:00:06.190 --> 01:00:08.149
I think the third module

1655
01:00:08.160 --> 01:00:09.440
of semantics would

1656
01:00:10.350 --> 01:00:11.350
likely.

1657
01:00:11.970 --> 01:00:12.970
But I think

1658
01:00:13.800 --> 01:00:14.800
there is a way us.

1659
01:00:15.590 --> 01:00:16.889
I think there is a way forward to

1660
01:00:16.890 --> 01:00:18.030
get the stability and

1661
01:00:18.040 --> 01:00:19.040
rehabilitation.

1662
01:00:19.310 --> 01:00:20.310
And I think.

1663
01:00:23.200 --> 01:00:24.200
Awesome.

1664
01:00:24.400 --> 01:00:26.115
Do we want to take another audience

1665
01:00:26.170 --> 01:00:27.170
question?

1666
01:00:28.960 --> 01:00:29.960
All right.

1667
01:00:31.650 --> 01:00:32.650
Thanks.

1668
01:00:36.650 --> 01:00:38.089
Can you talk into the mike like

1669
01:00:38.090 --> 01:00:38.899
you're eating the mike?

1670
01:00:38.900 --> 01:00:39.900
Hi. Yeah.

1671
01:00:39.980 --> 01:00:40.980
Can you hear me?

1672
01:00:41.190 --> 01:00:42.190
CAROL Yeah.

1673
01:00:42.380 --> 01:00:43.879
[Participant 2]
Great. There's been a lot of talk tonight

1674
01:00:43.880 --> 01:00:45.559
about the evolvability of

1675
01:00:45.560 --> 01:00:46.999
languages and

1676
01:00:47.510 --> 01:00:48.931
the trouble with implementing

1677
01:00:49.820 --> 01:00:51.169
things that might not be backwards

1678
01:00:51.170 --> 01:00:52.170
compatible.

1679
01:00:52.220 --> 01:00:53.899
What's something that you wish you

1680
01:00:53.900 --> 01:00:55.009
could have implemented with your

1681
01:00:55.010 --> 01:00:56.869
language? But for that, or

1682
01:00:56.870 --> 01:00:58.039
[Participant 2]
maybe another reason that wasn't

1683
01:00:58.040 --> 01:00:59.040
possible.

1684
01:01:02.160 --> 01:01:03.160
A

1685
01:01:06.300 --> 01:01:07.449
word,

1686
01:01:11.090 --> 01:01:12.090
right?

1687
01:01:13.370 --> 01:01:14.719
As all the

1688
01:01:15.820 --> 01:01:16.820
words

1689
01:01:19.070 --> 01:01:20.829
or these

1690
01:01:22.540 --> 01:01:23.689
or other forms of

1691
01:01:28.440 --> 01:01:29.440
naming is hard.

1692
01:01:31.690 --> 01:01:33.039
Well, I'll I'll I'll go.

1693
01:01:33.040 --> 01:01:34.989
I've. I mean I my favorite is always

1694
01:01:35.200 --> 01:01:37.059
like d that the two billion dollar

1695
01:01:37.060 --> 01:01:38.060
mistake of the

1696
01:01:38.920 --> 01:01:40.269
billion dollar mistake of having

1697
01:01:40.270 --> 01:01:41.379
Nall in the language.

1698
01:01:41.400 --> 01:01:42.400
And since JavaScript

1699
01:01:43.540 --> 01:01:45.009
has both null and undefined, it's

1700
01:01:45.010 --> 01:01:46.630
the two billion dollar mistake.

1701
01:01:48.730 --> 01:01:49.730
But

1702
01:01:50.770 --> 01:01:51.820
I mean if it did.

1703
01:01:52.320 --> 01:01:54.219
And I mean what's done is done

1704
01:01:54.220 --> 01:01:56.169
right. And and now now,

1705
01:01:56.230 --> 01:01:58.539
you know, I spend a significant

1706
01:01:58.540 --> 01:02:00.699
portion of my time actually working

1707
01:02:00.700 --> 01:02:02.910
on ways to

1708
01:02:04.110 --> 01:02:06.310
to make code null safe

1709
01:02:06.880 --> 01:02:07.880
to and who's

1710
01:02:08.770 --> 01:02:10.749
never who in here has ever had

1711
01:02:10.750 --> 01:02:11.949
a null pointer exception.

1712
01:02:12.010 --> 01:02:13.179
Right. There you go.

1713
01:02:14.530 --> 01:02:16.780
It's it is by far the most

1714
01:02:16.960 --> 01:02:17.960
problematic part of

1715
01:02:18.820 --> 01:02:20.439
language design. It it's a single

1716
01:02:20.440 --> 01:02:21.440
value that

1717
01:02:22.560 --> 01:02:24.280
that if only that wasn't there.

1718
01:02:25.060 --> 01:02:26.679
Imagine all the problems we wouldn't

1719
01:02:26.680 --> 01:02:28.199
have. Right. If if type systems

1720
01:02:28.570 --> 01:02:30.139
were designed that way and some type

1721
01:02:30.140 --> 01:02:31.757
systems are and some type systems

1722
01:02:32.020 --> 01:02:33.279
are getting there. But boy, trying

1723
01:02:33.280 --> 01:02:35.139
to retrofit that on top of a

1724
01:02:35.140 --> 01:02:36.806
type system that has normal in the

1725
01:02:36.820 --> 01:02:38.739
first place is

1726
01:02:39.400 --> 01:02:40.400
quite an undertaking.

1727
01:02:41.740 --> 01:02:43.357
The features I wanted to add were

1728
01:02:43.360 --> 01:02:44.360
negative features.

1729
01:02:46.660 --> 01:02:48.081
I think all of us as language

1730
01:02:48.580 --> 01:02:50.739
designers have borrowed

1731
01:02:50.740 --> 01:02:51.729
things.

1732
01:02:51.730 --> 01:02:53.259
You know, we all steal from each

1733
01:02:53.260 --> 01:02:54.550
other's languages all the time.

1734
01:02:55.300 --> 01:02:57.230
And and

1735
01:02:57.920 --> 01:02:59.096
it's often we steal good

1736
01:02:59.920 --> 01:03:00.279
things.

1737
01:03:00.280 --> 01:03:02.229
But for some reason, we also steal

1738
01:03:02.230 --> 01:03:03.230
bad things.

1739
01:03:05.580 --> 01:03:06.580
You know, like what?

1740
01:03:07.160 --> 01:03:08.750
Like regular expression syntax.

1741
01:03:08.850 --> 01:03:10.570
Oh, yeah.

1742
01:03:11.480 --> 01:03:12.379
I did that one.

1743
01:03:12.380 --> 01:03:14.150
Like the C precedence table.

1744
01:03:14.750 --> 01:03:15.750
OK. Another.

1745
01:03:15.870 --> 01:03:16.870
OK.

1746
01:03:16.880 --> 01:03:18.049
These are things that I could not

1747
01:03:18.050 --> 01:03:19.090
fix in four or five.

1748
01:03:20.450 --> 01:03:21.289
And we did fix it.

1749
01:03:21.290 --> 01:03:22.290
Burleson's.

1750
01:03:24.620 --> 01:03:25.620
Awesome.

1751
01:03:25.940 --> 01:03:26.940
All right.

1752
01:03:27.140 --> 01:03:28.806
We do tend to be doing pretty well

1753
01:03:29.060 --> 01:03:29.989
with the audience. So how would

1754
01:03:29.990 --> 01:03:31.249
another audience question?

1755
01:03:36.680 --> 01:03:37.680
Hi. Hello there.

1756
01:03:38.280 --> 01:03:40.399
So, well, what I wanted to ask

1757
01:03:40.400 --> 01:03:42.319
is you can sort of know what is

1758
01:03:42.770 --> 01:03:44.690
these sort of pendulum effect

1759
01:03:45.200 --> 01:03:47.269
in regards to tech and programing,

1760
01:03:47.500 --> 01:03:48.530
throw it out. Time

1761
01:03:49.430 --> 01:03:50.929
to Vinick's several different

1762
01:03:50.930 --> 01:03:51.859
paradigms.

1763
01:03:51.860 --> 01:03:53.219
There's a certain time where people

1764
01:03:53.230 --> 01:03:55.039
were like, are we going to do things

1765
01:03:55.040 --> 01:03:55.959
imperatively?

1766
01:03:55.960 --> 01:03:57.439
Are we going to go object oriented

1767
01:03:57.440 --> 01:03:58.440
or functional?

1768
01:03:59.770 --> 01:04:01.729
And for example, right now there's

1769
01:04:01.730 --> 01:04:03.130
a whole bunch of language used are

1770
01:04:03.140 --> 01:04:04.610
sort of like very aggressively

1771
01:04:05.480 --> 01:04:07.280
taking that standpoint of

1772
01:04:07.820 --> 01:04:09.589
being very pro,

1773
01:04:09.890 --> 01:04:12.409
being friendly with concurrency

1774
01:04:12.410 --> 01:04:14.270
or being bery

1775
01:04:14.720 --> 01:04:16.429
regardless of what immutability with

1776
01:04:16.430 --> 01:04:17.430
memory.

1777
01:04:17.600 --> 01:04:19.070
And I think that that kind of

1778
01:04:19.160 --> 01:04:20.287
pendulum effect happens

1779
01:04:21.080 --> 01:04:22.999
because technology has actually

1780
01:04:23.000 --> 01:04:24.339
been evolving throughout time.

1781
01:04:24.920 --> 01:04:26.749
Our machines are like beefier and

1782
01:04:26.750 --> 01:04:28.010
we have more memory now.

1783
01:04:28.520 --> 01:04:29.520
So that waste on we

1784
01:04:30.380 --> 01:04:32.239
design programing languages now

1785
01:04:32.790 --> 01:04:34.099
are probably a lot different than

1786
01:04:34.100 --> 01:04:36.079
there were 20 or 30 years

1787
01:04:36.080 --> 01:04:37.429
ago. So

1788
01:04:37.970 --> 01:04:39.919
considering that, where do you think

1789
01:04:39.980 --> 01:04:42.019
or how do you think the programing

1790
01:04:42.020 --> 01:04:43.579
languages of 10 years into the

1791
01:04:43.580 --> 01:04:44.630
future are going to look like?

1792
01:04:45.290 --> 01:04:46.999
Or in your opinion,

1793
01:04:48.070 --> 01:04:50.239
where where is language you design

1794
01:04:50.510 --> 01:04:52.489
headed to in the future?

1795
01:04:53.220 --> 01:04:54.079
So. So.

1796
01:04:54.080 --> 01:04:56.210
So my favorite candidate for that

1797
01:04:57.020 --> 01:04:58.909
is that, you

1798
01:04:58.910 --> 01:05:00.919
know, these, you know, major breaks

1799
01:05:00.920 --> 01:05:02.299
and things like that always happen

1800
01:05:02.300 --> 01:05:03.672
because, you know, something

1801
01:05:04.280 --> 01:05:05.799
major happens in the underlying

1802
01:05:05.870 --> 01:05:06.870
technology.

1803
01:05:07.550 --> 01:05:09.469
And, you know, one

1804
01:05:09.470 --> 01:05:10.579
of the areas

1805
01:05:11.540 --> 01:05:12.540
of underlying

1806
01:05:13.550 --> 01:05:15.069
technology that is kind of like

1807
01:05:15.650 --> 01:05:17.750
a computer that

1808
01:05:17.930 --> 01:05:20.420
I think is really underserved is

1809
01:05:20.900 --> 01:05:22.880
is writing code for GP use.

1810
01:05:24.000 --> 01:05:25.520
Right. I mean, right now there are

1811
01:05:26.270 --> 01:05:28.190
there are no languages

1812
01:05:28.550 --> 01:05:29.550
worth that crap.

1813
01:05:30.500 --> 01:05:32.719
And that's a technical for trademark

1814
01:05:35.010 --> 01:05:37.070
that work well

1815
01:05:37.160 --> 01:05:38.389
on GP use.

1816
01:05:38.960 --> 01:05:41.020
And, you know, maybe because there's

1817
01:05:41.030 --> 01:05:42.030
like a limited

1818
01:05:43.160 --> 01:05:44.929
number of algorithms that people are

1819
01:05:44.930 --> 01:05:46.498
willing to are really interested

1820
01:05:46.820 --> 01:05:48.049
in, but they're really important

1821
01:05:48.050 --> 01:05:48.889
ones.

1822
01:05:48.890 --> 01:05:50.479
Or maybe it's because they're

1823
01:05:50.540 --> 01:05:52.640
they're inherently all mathematical.

1824
01:05:53.330 --> 01:05:55.039
And the number of people who care

1825
01:05:55.040 --> 01:05:56.929
about writing numerical code

1826
01:05:56.960 --> 01:05:58.699
is small.

1827
01:06:00.080 --> 01:06:01.670
I'm one of those, though.

1828
01:06:02.160 --> 01:06:03.826
And you know, and so every now and

1829
01:06:03.980 --> 01:06:05.749
then, I spend time thinking about,

1830
01:06:06.350 --> 01:06:07.569
you know, what would you do to make

1831
01:06:07.570 --> 01:06:08.900
GPE you programing

1832
01:06:09.530 --> 01:06:10.530
really easy.

1833
01:06:17.550 --> 01:06:19.429
Yeah, but but but it's more than

1834
01:06:19.430 --> 01:06:20.749
just vector operators.

1835
01:06:20.800 --> 01:06:22.639
I mean that that's kind of my my

1836
01:06:22.640 --> 01:06:24.109
problem with the whole thing is that

1837
01:06:24.110 --> 01:06:26.239
most of these things are just

1838
01:06:26.360 --> 01:06:28.309
libraries of hand coded

1839
01:06:28.310 --> 01:06:30.840
assembly that do vector operators.

1840
01:06:31.850 --> 01:06:33.889
And and surely there's

1841
01:06:33.890 --> 01:06:35.510
something more clever than that.

1842
01:06:38.960 --> 01:06:40.439
Maybe I'll just add one with

1843
01:06:40.460 --> 01:06:41.509
language design. You know, one of

1844
01:06:41.510 --> 01:06:42.409
one of the things that's interesting,

1845
01:06:42.410 --> 01:06:43.879
you look at like all of us old

1846
01:06:43.880 --> 01:06:45.209
geezers sitting up here and there

1847
01:06:45.230 --> 01:06:46.969
they were, we're proof positive that

1848
01:06:46.970 --> 01:06:48.219
languages move slowly.

1849
01:06:49.860 --> 01:06:51.800
And I mean, it's not

1850
01:06:51.890 --> 01:06:52.890
you know, like

1851
01:06:53.930 --> 01:06:55.339
a lot of people make the mistake of

1852
01:06:55.340 --> 01:06:56.839
thinking that languages move at the

1853
01:06:56.840 --> 01:06:58.159
same speed as like

1854
01:06:58.730 --> 01:07:00.619
hardware or like all of the other

1855
01:07:00.620 --> 01:07:02.480
technologies that that that

1856
01:07:02.600 --> 01:07:04.190
we live with. But but

1857
01:07:04.670 --> 01:07:06.379
languages are much more like math

1858
01:07:06.410 --> 01:07:08.329
and much more like the human brain.

1859
01:07:08.360 --> 01:07:10.049
And, you know, and they all evolve

1860
01:07:10.250 --> 01:07:11.879
slowly. And we're still programing

1861
01:07:11.880 --> 01:07:13.595
and languages that were invented 50

1862
01:07:13.880 --> 01:07:15.019
years ago, like functional.

1863
01:07:15.020 --> 01:07:16.159
But all of the principles of

1864
01:07:16.160 --> 01:07:17.599
functional programing were thought

1865
01:07:17.600 --> 01:07:19.070
of more than 50 years ago.

1866
01:07:20.420 --> 01:07:22.369
I do think one of the things that

1867
01:07:22.370 --> 01:07:24.469
that is luckily happening is

1868
01:07:24.470 --> 01:07:26.140
that like as Larry said, everyone's

1869
01:07:26.150 --> 01:07:27.739
boring from everyone and languages

1870
01:07:27.740 --> 01:07:29.760
are becoming more multi paradigm.

1871
01:07:29.810 --> 01:07:31.699
Yeah, I think it's wrong to talk

1872
01:07:31.700 --> 01:07:33.559
about oh, I only

1873
01:07:33.560 --> 01:07:35.079
like object oriented programing

1874
01:07:35.150 --> 01:07:37.099
languages or I only like imperative

1875
01:07:37.100 --> 01:07:38.960
programing or functional programing.

1876
01:07:38.990 --> 01:07:40.520
I mean, it's important to

1877
01:07:41.300 --> 01:07:42.829
look at where's the research or

1878
01:07:42.850 --> 01:07:44.149
where the new where's the new

1879
01:07:44.150 --> 01:07:45.709
thinking and where where are new

1880
01:07:45.710 --> 01:07:47.419
paradigms that are interesting and

1881
01:07:47.420 --> 01:07:49.400
then try to try to incorporate them.

1882
01:07:49.430 --> 01:07:51.320
But do so tastefully

1883
01:07:51.680 --> 01:07:53.809
in a sense and and work them in to

1884
01:07:53.900 --> 01:07:55.615
work them into whatever is is there

1885
01:07:56.060 --> 01:07:56.669
already.

1886
01:07:56.670 --> 01:07:58.336
And and I think we're all learning

1887
01:07:58.730 --> 01:08:00.109
a lot from functional programing

1888
01:08:00.110 --> 01:08:00.979
languages these days.

1889
01:08:00.980 --> 01:08:03.139
I certainly feel like I am because

1890
01:08:03.140 --> 01:08:05.089
a lot of interesting research

1891
01:08:05.090 --> 01:08:06.090
has happened there.

1892
01:08:06.290 --> 01:08:07.440
But functional programing

1893
01:08:08.210 --> 01:08:09.337
is imperfect and no one

1894
01:08:10.070 --> 01:08:11.690
writes pure functional programs.

1895
01:08:11.750 --> 01:08:13.429
I mean because they they don't

1896
01:08:13.430 --> 01:08:14.606
exist. So it's all about

1897
01:08:15.260 --> 01:08:17.270
like how can you tastefully

1898
01:08:17.399 --> 01:08:18.399
sneak in mutation in

1899
01:08:19.340 --> 01:08:21.289
ways that you can better reason

1900
01:08:21.290 --> 01:08:23.119
about as opposed to

1901
01:08:23.750 --> 01:08:25.879
mutation and free threading for

1902
01:08:25.880 --> 01:08:26.818
everyone, you know?

1903
01:08:26.819 --> 01:08:28.579
And that's like just a

1904
01:08:28.970 --> 01:08:30.259
recipe for disaster.

1905
01:08:30.290 --> 01:08:30.659
Right.

1906
01:08:30.660 --> 01:08:31.660
So,

1907
01:08:32.740 --> 01:08:33.740
yes.

1908
01:08:40.560 --> 01:08:41.999
All languages are imperfect.

1909
01:08:42.029 --> 01:08:43.439
Let's just settle it right there.

1910
01:08:43.540 --> 01:08:44.540
You know.

1911
01:09:04.760 --> 01:09:06.619
Yes, we've been

1912
01:09:06.620 --> 01:09:07.959
thinking a lot about the

1913
01:09:08.470 --> 01:09:10.550
the the parallelism

1914
01:09:10.580 --> 01:09:11.959
concurrency issue.

1915
01:09:12.770 --> 01:09:14.599
This is one of those issues where we

1916
01:09:14.600 --> 01:09:15.739
were waiting to be smarter.

1917
01:09:17.229 --> 01:09:18.229
Uh, we we

1918
01:09:19.130 --> 01:09:20.130
did borrow.

1919
01:09:21.050 --> 01:09:22.399
I mean, steal a bunch of

1920
01:09:23.029 --> 01:09:24.529
ideas from Haskell

1921
01:09:24.920 --> 01:09:26.194
in terms of lazy lists and

1922
01:09:26.750 --> 01:09:28.579
things like that. But in

1923
01:09:28.580 --> 01:09:30.649
terms of how you actually

1924
01:09:30.890 --> 01:09:32.059
are going to deal with the end of

1925
01:09:32.060 --> 01:09:33.060
Moore's Law

1926
01:09:34.130 --> 01:09:35.389
or if not the end of Moore's Law, at

1927
01:09:35.390 --> 01:09:37.759
least, you know, multi, multi,

1928
01:09:38.990 --> 01:09:41.090
multiple C.P.S., many, many cores.

1929
01:09:41.540 --> 01:09:42.688
When you have a thousand cores, what

1930
01:09:42.689 --> 01:09:43.689
are you gonna do with them?

1931
01:09:44.180 --> 01:09:46.309
And so a lot of our

1932
01:09:46.760 --> 01:09:48.648
early design on Perl 6

1933
01:09:48.649 --> 01:09:50.449
was we didn't understand how

1934
01:09:50.510 --> 01:09:52.309
we wanted to do it yet, but

1935
01:09:52.550 --> 01:09:54.140
we knew very

1936
01:09:54.620 --> 01:09:56.539
we knew very deeply that we didn't

1937
01:09:56.540 --> 01:09:57.409
want to do anything. It would

1938
01:09:57.410 --> 01:09:58.410
prevent it.

1939
01:09:59.060 --> 01:10:00.899
So a lot of the early design

1940
01:10:00.980 --> 01:10:02.930
decisions were just saying, well,

1941
01:10:03.290 --> 01:10:04.290
we have to keep this open.

1942
01:10:05.210 --> 01:10:06.778
Then when the right smart people

1943
01:10:06.950 --> 01:10:08.600
came along, they said, look,

1944
01:10:09.230 --> 01:10:10.896
what you really have to look at is

1945
01:10:11.210 --> 01:10:12.209
things that are composed.

1946
01:10:12.210 --> 01:10:13.210
All threads are

1947
01:10:14.060 --> 01:10:15.169
not composed of, well, we can't take

1948
01:10:15.170 --> 01:10:16.489
two threads and turn it into a third

1949
01:10:16.490 --> 01:10:17.490
thread.

1950
01:10:18.440 --> 01:10:19.819
What you want is things like

1951
01:10:20.360 --> 01:10:21.529
we stole from go from.

1952
01:10:21.530 --> 01:10:23.689
We stole promises and channels

1953
01:10:24.290 --> 01:10:26.029
from C-sharp. We stole functional

1954
01:10:26.030 --> 01:10:27.039
reaction will programing.

1955
01:10:29.270 --> 01:10:31.279
And with these sort of primitives

1956
01:10:32.510 --> 01:10:34.078
which end up being duels of your

1957
01:10:34.340 --> 01:10:35.719
regular list processing kind of

1958
01:10:35.720 --> 01:10:37.459
primitives, they just happened to be

1959
01:10:37.460 --> 01:10:38.460
working on events.

1960
01:10:38.890 --> 01:10:40.939
And you're you can have loops, event

1961
01:10:40.940 --> 01:10:42.139
loops that were just like regular

1962
01:10:42.140 --> 01:10:43.979
loops that they're running on and

1963
01:10:44.510 --> 01:10:45.379
same control flow.

1964
01:10:45.380 --> 01:10:46.700
So that is easy to learn.

1965
01:10:47.370 --> 01:10:48.938
There are ways of sneaking these

1966
01:10:49.310 --> 01:10:51.710
things into a language,

1967
01:10:53.000 --> 01:10:54.109
at least in the ways that we

1968
01:10:54.110 --> 01:10:55.220
understand right now

1969
01:10:55.940 --> 01:10:58.100
that I heard.

1970
01:10:59.750 --> 01:11:00.750
Was it.

1971
01:11:01.250 --> 01:11:02.869
Who is talking about adding fibers

1972
01:11:02.870 --> 01:11:04.420
to their runtime?

1973
01:11:04.640 --> 01:11:06.109
I think you'd think it was a Java

1974
01:11:06.230 --> 01:11:07.230
news item,

1975
01:11:08.600 --> 01:11:10.099
but like lightweight threads that

1976
01:11:10.100 --> 01:11:11.178
can you, as Erlang has

1977
01:11:12.110 --> 01:11:13.689
done, has shown, you know, you can

1978
01:11:13.690 --> 01:11:14.839
run a hundred thousand threads in

1979
01:11:14.840 --> 01:11:15.840
your process.

1980
01:11:15.990 --> 01:11:17.089
That's another language that's been

1981
01:11:17.090 --> 01:11:18.049
around for a long.

1982
01:11:18.050 --> 01:11:19.420
Oh yeah. And we stole from them too.

1983
01:11:19.600 --> 01:11:20.600
Yeah.

1984
01:11:21.080 --> 01:11:22.219
Should we. We like their pattern

1985
01:11:22.220 --> 01:11:23.099
matching.

1986
01:11:23.100 --> 01:11:24.919
But one of the many therms is

1987
01:11:24.920 --> 01:11:26.389
that all of that language design is

1988
01:11:26.390 --> 01:11:27.589
theft of course.

1989
01:11:28.230 --> 01:11:29.209
Yeah. So yeah.

1990
01:11:29.210 --> 01:11:30.380
I mean we're thinking about it.

1991
01:11:30.410 --> 01:11:31.549
We don't know have all the answers

1992
01:11:31.550 --> 01:11:32.550
yet but

1993
01:11:34.370 --> 01:11:35.899
I think all these languages, you

1994
01:11:35.900 --> 01:11:37.468
know, long term tend to converge

1995
01:11:37.790 --> 01:11:38.899
on similar solutions.

1996
01:11:39.620 --> 01:11:40.747
So very sadly we are at

1997
01:11:41.540 --> 01:11:43.580
the last question.

1998
01:11:44.990 --> 01:11:46.910
I know you went so quickly

1999
01:11:46.940 --> 01:11:48.770
and unfortunately

2000
01:11:49.190 --> 01:11:51.319
all good things must end

2001
01:11:52.130 --> 01:11:53.899
except for maybe good languages.

2002
01:11:56.150 --> 01:11:58.520
So I'm going to start and

2003
01:11:59.180 --> 01:12:00.920
maybe quickly wrap up

2004
01:12:01.400 --> 01:12:03.259
as you look back over

2005
01:12:03.320 --> 01:12:04.839
the long lives of the languages

2006
01:12:05.390 --> 01:12:06.390
that you've written.

2007
01:12:06.830 --> 01:12:08.630
What have you found most rewarding?

2008
01:12:10.150 --> 01:12:11.029
The people?

2009
01:12:11.030 --> 01:12:12.590
The people by far

2010
01:12:12.880 --> 01:12:13.880
after the.

2011
01:12:16.810 --> 01:12:17.810
Oh, yeah.

2012
01:12:18.630 --> 01:12:20.269
Yeah. I mean, that that's what keeps

2013
01:12:20.270 --> 01:12:21.800
me coming back, that to work

2014
01:12:21.830 --> 01:12:23.899
everyday is like the incredible

2015
01:12:23.900 --> 01:12:25.700
excitement that

2016
01:12:25.790 --> 01:12:27.409
the community shows.

2017
01:12:27.740 --> 01:12:29.839
You and I I am so grateful

2018
01:12:29.840 --> 01:12:31.550
for all the people who

2019
01:12:32.590 --> 01:12:33.600
who have used that.

2020
01:12:34.190 --> 01:12:35.509
I mean, it there's nothing more

2021
01:12:35.510 --> 01:12:36.784
rewarding than have people

2022
01:12:37.550 --> 01:12:39.430
just like on Twitter or

2023
01:12:39.440 --> 01:12:41.204
wherever it is, go, oh, my God, this

2024
01:12:41.270 --> 01:12:42.990
is so great. And it's like, yeah,

2025
01:12:43.430 --> 01:12:44.749
that makes you want to just keep

2026
01:12:44.750 --> 01:12:45.559
doing more of it.

2027
01:12:45.560 --> 01:12:46.049
I mean.

2028
01:12:46.050 --> 01:12:47.716
Oh, yeah. You know, somebody comes

2029
01:12:48.050 --> 01:12:49.618
up to me in the street somewhere

2030
01:12:50.120 --> 01:12:51.629
and says, you know,

2031
01:12:52.370 --> 01:12:54.080
thanks for giving me a career.

2032
01:12:54.500 --> 01:12:55.640
Can I have a selfie?

2033
01:12:57.370 --> 01:12:58.791
You know, that's like this is

2034
01:12:59.270 --> 01:13:01.189
kind of like, you know, light and

2035
01:13:01.190 --> 01:13:02.089
dark thing.

2036
01:13:02.090 --> 01:13:03.266
But being this is really

2037
01:13:03.950 --> 01:13:04.939
wonderful.

2038
01:13:04.940 --> 01:13:05.929
This is all great.

2039
01:13:05.930 --> 01:13:07.369
I mean, I have to be in Silicon

2040
01:13:07.370 --> 01:13:09.289
Valley at a movie theater, the

2041
01:13:09.290 --> 01:13:10.290
tire shop.

2042
01:13:14.350 --> 01:13:15.739
JAMES Yeah. Well, what's really weird is

2043
01:13:15.740 --> 01:13:18.410
JAMES when you're when your kids are with
you when that happens.

2045
01:13:20.430 --> 01:13:23.239
LARRY But yes, it's -
It's when the when people say, "you

2047
01:13:23.240 --> 01:13:25.990
LARRY gave me a career, you gave me 
a livelihood.

2049
01:13:26.480 --> 01:13:30.739
LARRY You fed my family."
That's the best.
[CROSSTALK]

2052
01:13:32.150 --> 01:13:36.289
LARRY It's.
And you know, and
and not just.

2055
01:13:36.990 --> 01:13:39.390
LARRY And we're not talking about a star 
relationship here,

2057
01:13:39.920 --> 01:13:43.369
LARRY of each of us to those individuals;
we're talking about the second

2059
01:13:43.370 --> 01:13:47.010
order, Community 2.0
kind of effects.

2061
01:13:47.300 --> 01:13:50.800
LARRY It's it's one
thing to see them, you know,

2063
01:13:51.950 --> 01:13:54.229
LARRY getting help from you;
It's another thing to see them

2065
01:13:54.230 --> 01:13:55.230
LARRY helping each other.

2066
01:13:56.660 --> 01:14:00.980
LARRY That's a whole whole level above 
even the individual relationships

2068
01:14:01.280 --> 01:14:05.539
LARRY I feel, to see a community
that is learning how to love,

2070
01:14:06.230 --> 01:14:08.340
LARRY you know, kind of building a little 
bit of heaven on earth.

2072
01:14:10.460 --> 01:14:11.460
LARRY There's just nothing like it.
[APPLAUSE]

2073
01:14:13.850 --> 01:14:15.020
CAROL Completely agree.
Guido, you get the last word

2074
01:14:15.480 --> 01:14:16.849
GUIDO I love learning from the community.

2075
01:14:22.250 --> 01:14:26.180
CAROL And we love learning from you as
well. And with that,

2078
01:14:26.810 --> 01:14:29.876
CAROL I want to say we've all
learned so much from all

2080
01:14:30.560 --> 01:14:34.357
CAROL of our wonderful panelists tonight.
I want to thank you for

2082
01:14:35.120 --> 01:14:38.600
CAROL allowing me to share and celebrate
this moment with you.

2084
01:14:39.260 --> 01:14:42.829
CAROL The little fifth-grade kid in me
that was plinking away on BASIC at

2086
01:14:42.830 --> 01:14:47.929
CAROL Bell Labs who later
watch the PC
cell phone, Internet, Web and cloud

2089
01:14:47.930 --> 01:14:53.510
CAROL computing and change our world,
never dreamed that I would be
chatting with four individuals

2092
01:14:53.780 --> 01:14:57.979
CAROL who profoundly impacted our world
by doing what they love, creating

2094
01:14:58.100 --> 01:14:59.239
CAROL and building languages.

2095
01:14:59.540 --> 01:15:03.529
CAROL Please join me in
a huge round of applause to thank

2097
01:15:03.620 --> 01:15:09.180
CAROL Anders
for inspiring this evening.

2099
01:15:15.990 --> 01:15:16.990
ANDERS Thank you, Carol.

NOTE Should the following section be modified?
There are several speakers who cannot
be identified and lots of 
crosstalk. It may be sufficient to add
[CROSSTALK] or [UNIDENTIFIED SPEAKER]
to line 8429 and the remaining segments
but otherwise leave the remaining 
autogenerated captions

2100
01:15:18.050 --> 01:15:19.050
OK.

2101
01:15:19.320 --> 01:15:19.629
OK.

2102
01:15:19.630 --> 01:15:20.720
Was that epic?

2103
01:15:22.200 --> 01:15:24.539
So, Marina, I have an idea.

2104
01:15:24.840 --> 01:15:26.457
If you guys could step down right

2105
01:15:26.820 --> 01:15:28.109
here, we're going to get a giant

2106
01:15:28.110 --> 01:15:29.110
picture of everybody.

2107
01:15:30.720 --> 01:15:32.130
You guys stand right there.

2108
01:15:32.770 --> 01:15:33.629
We're going to get up here with your

2109
01:15:33.630 --> 01:15:34.630
camera.

2110
01:15:38.610 --> 01:15:39.869
All right. We're getting a picture

2111
01:15:39.870 --> 01:15:40.920
of everybody in that room.

2112
01:15:42.300 --> 01:15:44.189
All right. I need some excitement.

2113
01:15:47.070 --> 01:15:48.600
Oh, my God. I met my idol.

2114
01:15:50.460 --> 01:15:51.359
She can stand on the chair.

2115
01:15:51.360 --> 01:15:52.360
You can use my shoulder.

2116
01:15:53.220 --> 01:15:54.059
OK.

2117
01:15:54.060 --> 01:15:55.628
All right. Show some excitement,

2118
01:15:55.680 --> 01:15:56.760
people. Let's see it.

2119
01:16:01.350 --> 01:16:02.350
Monahan's.

2120
01:16:05.480 --> 01:16:07.180
Three times it's kinda weird and

2121
01:16:07.190 --> 01:16:08.190
lumpy.

2122
01:16:11.630 --> 01:16:12.630
OK,

2123
01:16:14.480 --> 01:16:16.189
thank you, everyone, for coming.

2124
01:16:16.220 --> 01:16:18.199
Thank you to all of our speakers.

2125
01:16:18.230 --> 01:16:19.790
Thank you to our organizers.

2126
01:16:19.820 --> 01:16:22.039
Let's do something equally awesome

2127
01:16:22.040 --> 01:16:23.040
next year.

2128
01:16:23.780 --> 01:16:24.780
And

2129
01:16:25.640 --> 01:16:26.939
if you want a robot there, 100

2130
01:16:26.960 --> 01:16:28.070
bucks. So come see me.

2131
01:16:29.000 --> 01:38:00.000
[No Audio]