1
00:00:01,260 --> 00:00:03,260
welcome everybody to

2
00:00:03,260 --> 00:00:06,440
another episode of functional
programming one along

3
00:00:06,440 --> 00:00:10,410
and today we're going to talk about

4
00:00:10,410 --> 00:00:15,059
declaring types and classes ourselves up
to now

5
00:00:15,059 --> 00:00:19,699
we have only in the and lecture on
parsers

6
00:00:19,699 --> 00:00:23,570
we have defined dives ourselves an

7
00:00:23,570 --> 00:00:28,500
but for the rest we gotta have issue
data types are there for example we have

8
00:00:28,500 --> 00:00:29,070
issue

9
00:00:29,070 --> 00:00:33,540
that the dye Bliss Texas then we have
assumed that doubles exists

10
00:00:33,540 --> 00:00:37,940
but of course just like we can define
new function ask our

11
00:00:37,940 --> 00:00:40,950
we also want to define new types

12
00:00:40,950 --> 00:00:44,100
and well we're going to do in

13
00:00:44,100 --> 00:00:47,649
dis lectures we're going to show an

14
00:00:47,649 --> 00:00:52,399
how to declare new types and how to dig
their new type gloss ish

15
00:00:52,399 --> 00:00:56,010
but also I'm going to compare and
contrast this

16
00:00:56,010 --> 00:00:59,570
with typical object-oriented programming

17
00:00:59,570 --> 00:01:03,070
am how these things work there and what
you will see

18
00:01:03,070 --> 00:01:07,659
is that although this index is a little
bit different a lot of the ideas are

19
00:01:07,659 --> 00:01:08,580
very common

20
00:01:08,580 --> 00:01:12,510
and this is a general thing that I like
to do instead of emphasizing the

21
00:01:12,510 --> 00:01:13,590
differences

22
00:01:13,590 --> 00:01:17,830
I think it's often much more productive
to emphasize

23
00:01:17,830 --> 00:01:21,280
commonalities okay and and and

24
00:01:21,280 --> 00:01:24,330
maybe you know you should try dish do it

25
00:01:24,330 --> 00:01:28,640
alas programmers we often gotta go well
but no

26
00:01:28,640 --> 00:01:32,200
it's much more productive to say yes and

27
00:01:32,200 --> 00:01:35,470
and emphasize am you now

28
00:01:35,470 --> 00:01:39,409
but this common am and so I'm going to
try to do that here

29
00:01:39,409 --> 00:01:42,700
and and please take that as a metal ass
and

30
00:01:42,700 --> 00:01:46,500
for this lecture alright

31
00:01:46,500 --> 00:01:50,010
am and Haskell an weekend

32
00:01:50,010 --> 00:01:53,190
the final alias for an existing type

33
00:01:53,190 --> 00:01:56,330
an by saying type strain

34
00:01:56,330 --> 00:01:59,490
equals another time are

35
00:01:59,490 --> 00:02:03,310
type you now new name equals another
type

36
00:02:03,310 --> 00:02:06,729
and in this case we are defining string

37
00:02:06,729 --> 00:02:09,820
as an ailish for a list of characters

38
00:02:09,820 --> 00:02:14,130
and this is it and an idea

39
00:02:14,130 --> 00:02:17,820
that we have used throughout all the
lectures right every time we saw

40
00:02:17,820 --> 00:02:18,650
straying

41
00:02:18,650 --> 00:02:24,220
we treated it as a list of characters
but n-type three of them wrote strength

42
00:02:24,220 --> 00:02:25,470
biggest

43
00:02:25,470 --> 00:02:29,430
that's what we want to do convey but
when we were going against range

44
00:02:29,430 --> 00:02:33,150
we want to see them as a list of
characters and

45
00:02:33,150 --> 00:02:36,760
and here this type declaration achieved
that

46
00:02:36,760 --> 00:02:40,080
this is just another name fordice died

47
00:02:40,080 --> 00:02:45,470
good and

48
00:02:45,470 --> 00:02:48,959
what we often do is reused this just
like which train

49
00:02:48,959 --> 00:02:52,690
reused type aliases to convey the
meaning

50
00:02:52,690 --> 00:02:58,340
our intent of what we want to do for
example instead of having a pair of

51
00:02:58,340 --> 00:02:59,110
enter chess

52
00:02:59,110 --> 00:03:02,730
we can see a while we declared this as

53
00:03:02,730 --> 00:03:06,250
diaper position and then instead of

54
00:03:06,250 --> 00:03:09,900
in our writing in Tacoma and everywhere

55
00:03:09,900 --> 00:03:13,750
we can ride position showed the Orijen
thats

56
00:03:13,750 --> 00:03:18,190
at cornered 00 and then we want to go if
they are

57
00:03:18,190 --> 00:03:22,950
move a position to the laughter is a
function that takes a position

58
00:03:22,950 --> 00:03:26,380
every there's a new position and it
takes a bear

59
00:03:26,380 --> 00:03:29,880
XY and it subtracts Guam from

60
00:03:29,880 --> 00:03:33,910
X alright and now by looking at this
dive

61
00:03:33,910 --> 00:03:37,070
you get a better idea of what it's doing

62
00:03:37,070 --> 00:03:41,910
and the fact that we're using bears to
represent positions

63
00:03:41,910 --> 00:03:48,540
is really just an implementation detail
another example here

64
00:03:48,540 --> 00:03:52,290
is a parametrized type

65
00:03:52,290 --> 00:03:56,140
alias so here we define the type of
bearish

66
00:03:56,140 --> 00:03:58,380
of Taipei

67
00:03:58,380 --> 00:04:01,420
as bears of Taipei

68
00:04:01,420 --> 00:04:04,420
see so the type of bears

69
00:04:04,420 --> 00:04:08,080
a value of type A are bears values

70
00:04:08,080 --> 00:04:12,770
of Taipei and now we can define
functions

71
00:04:12,770 --> 00:04:17,070
over debt type for example we can take a
pair of integers

72
00:04:17,070 --> 00:04:21,440
and return an integer by multiplying the
two

73
00:04:21,440 --> 00:04:24,530
or we can take a single value

74
00:04:24,530 --> 00:04:27,950
and copy that into a pair

75
00:04:27,950 --> 00:04:31,530
and there's the function got be which
takes annex

76
00:04:31,530 --> 00:04:35,610
and returns a tuple that contains
another copy of Mac

77
00:04:35,610 --> 00:04:39,010
type decorations

78
00:04:39,010 --> 00:04:42,140
of course can be nested saw it

79
00:04:42,140 --> 00:04:45,650
if we define position as a pair of
integers

80
00:04:45,650 --> 00:04:49,680
then weekend define another type Shin am

81
00:04:49,680 --> 00:04:53,080
to me position to position so

82
00:04:53,080 --> 00:04:57,950
a transformation other position as a
function that takes one position

83
00:04:57,950 --> 00:05:02,620
an takes every turn another position so
in this case here

84
00:05:02,620 --> 00:05:07,190
we could have defined this function here
to have typed

85
00:05:07,190 --> 00:05:11,960
trench now there's one important thing
here is that these

86
00:05:11,960 --> 00:05:16,790
and type synonyms I really just synonyms
they're just abbreviations

87
00:05:16,790 --> 00:05:21,510
and which means that they cannot be
recursive so in this case

88
00:05:21,510 --> 00:05:24,820
where you define type3 to be

89
00:05:24,820 --> 00:05:28,070
a pair of an event analyst of threes

90
00:05:28,070 --> 00:05:31,440
this is not a lot and when

91
00:05:31,440 --> 00:05:34,870
you can t find and we will see that
later its

92
00:05:34,870 --> 00:05:38,090
no problem to define recursive types in
Haskell

93
00:05:38,090 --> 00:05:41,300
but they always have to go be a normal
type

94
00:05:41,300 --> 00:05:44,500
an and this is a restriction that

95
00:05:44,500 --> 00:05:48,530
many languages have if you have an
object-oriented

96
00:05:48,530 --> 00:05:51,630
a language typically it has a

97
00:05:51,630 --> 00:05:55,760
a nominal type system and that makes it
easier to check

98
00:05:55,760 --> 00:06:00,350
whether to died sorry call our mom died
this is sharp dive bar of another

99
00:06:00,350 --> 00:06:03,610
and if you have purely structural dives
that the gums

100
00:06:03,610 --> 00:06:07,669
a little bit more difficult so it's them
that this

101
00:06:07,669 --> 00:06:11,330
typically the reason for that an but if
you're interested

102
00:06:11,330 --> 00:06:16,419
programming language in type systems the
field of structural verses nominal types

103
00:06:16,419 --> 00:06:18,370
is it very interesting

104
00:06:18,370 --> 00:06:23,900
a field and you know you can spend years
researching that am reading papers on

105
00:06:23,900 --> 00:06:24,550
that and

106
00:06:24,550 --> 00:06:31,550
you know get a PhD in that area right I
said this when we want to

107
00:06:32,479 --> 00:06:36,659
make a recursive dive we have to go
through a nominal typed

108
00:06:36,659 --> 00:06:40,449
and the way we define nominal types in
Haskell

109
00:06:40,449 --> 00:06:43,940
is the I data declarations

110
00:06:43,940 --> 00:06:48,800
so here we're defining a new type named
bull

111
00:06:48,800 --> 00:06:52,259
and it has two constructors

112
00:06:52,259 --> 00:06:55,590
through and false K

113
00:06:55,590 --> 00:07:00,430
and these constructors are don't don't
take argument

114
00:07:00,430 --> 00:07:05,779
they are at the same time values if you

115
00:07:05,779 --> 00:07:09,270
no skyline you will see that skyline

116
00:07:09,270 --> 00:07:12,360
case classes are very similar to algebra

117
00:07:12,360 --> 00:07:15,469
datatypes an but if you did

118
00:07:15,469 --> 00:07:20,580
you know even if you don't know scholar
but say you know a job our C sharp

119
00:07:20,580 --> 00:07:23,960
one way to look at these data
decorations is as follows:

120
00:07:23,960 --> 00:07:28,460
this is an abstract base class ball and
then

121
00:07:28,460 --> 00:07:32,029
it dives here the under right hand side

122
00:07:32,029 --> 00:07:36,039
are the subject show falls extensible

123
00:07:36,039 --> 00:07:41,779
and through extensible and since bullets
elvis abstract you cannot create an in

124
00:07:41,779 --> 00:07:42,779
search of bull

125
00:07:42,779 --> 00:07:46,419
you can only create instances of false
and through

126
00:07:46,419 --> 00:07:50,199
but both have died bull and Haskell

127
00:07:50,199 --> 00:07:55,419
and true and false don't have their own
types they always have died bull

128
00:07:55,419 --> 00:08:00,839
so they that even simpler than having an
abstract base class

129
00:08:00,839 --> 00:08:06,620
and two iraq types as I mentioned

130
00:08:06,620 --> 00:08:10,490
the their values on the right hand side
there to run for a false

131
00:08:10,490 --> 00:08:13,729
are typically gold constructors an

132
00:08:13,729 --> 00:08:16,930
and yeah with explanation I gave

133
00:08:16,930 --> 00:08:19,710
with the abstract I band

134
00:08:19,710 --> 00:08:22,790
subtypes that make sense they correspond

135
00:08:22,790 --> 00:08:25,860
two constructors in an object-oriented
language

136
00:08:25,860 --> 00:08:30,600
and and Haskell died and constructor
names

137
00:08:30,600 --> 00:08:34,130
must begin with an uppercase letter SL

138
00:08:34,130 --> 00:08:38,390
functions and died very bones began with
a lower case letter

139
00:08:38,390 --> 00:08:41,870
and dives and construct their names must
begin

140
00:08:41,870 --> 00:08:45,860
million a British latter and these
algebra

141
00:08:45,860 --> 00:08:50,390
datatypes another way to look at them
are as context-free grammars

142
00:08:50,390 --> 00:08:53,720
where the thing on the left hand side is
known on terminal

143
00:08:53,720 --> 00:08:57,510
and then their right hand side can an
contain

144
00:08:57,510 --> 00:09:02,880
and terminals which are the constructors
that themselves can recursively

145
00:09:02,880 --> 00:09:06,760
contain other types like a grammar
contain

146
00:09:06,760 --> 00:09:11,020
and recursively on the right hand side
unknown terminals as well

147
00:09:11,020 --> 00:09:15,280
Heerden an example

148
00:09:15,280 --> 00:09:19,260
a slightly more interesting as you brag
datatype

149
00:09:19,260 --> 00:09:22,730
and its like a boolean thing red

150
00:09:22,730 --> 00:09:26,590
three values three valuable ensure we
have yes

151
00:09:26,590 --> 00:09:30,860
no and don't know so this is maybe like
sequel

152
00:09:30,860 --> 00:09:34,600
where this is like you now know so is it
revalued

153
00:09:34,600 --> 00:09:39,190
logic and now weekend use this thing

154
00:09:39,190 --> 00:09:44,020
Tuesday I want to have a list of answers
yes no and no

155
00:09:44,020 --> 00:09:49,550
and now I'm going to define another
function that flips the answers i buy

156
00:09:49,550 --> 00:09:52,740
this is like the generalization of
negation

157
00:09:52,740 --> 00:09:57,070
and if I have yes and I let that becomes
no

158
00:09:57,070 --> 00:10:01,750
if I have no and I let that it becomes
yes

159
00:10:01,750 --> 00:10:05,390
and if I had if I don't know and I
flatbed

160
00:10:05,390 --> 00:10:10,050
I still don't know writer if you're
familiar with chic wall

161
00:10:10,050 --> 00:10:13,200
dish will be very similar where'd

162
00:10:13,200 --> 00:10:17,720
unknown plays the role not

163
00:10:17,720 --> 00:10:24,149
here's the and standards oooh tombaugh
love circles and rectangles and shapes

164
00:10:24,149 --> 00:10:27,170
an ascot looks as follows: we haven't

165
00:10:27,170 --> 00:10:31,700
abstract base class shape and Duceppe
type circle

166
00:10:31,700 --> 00:10:35,310
and rectangle both of them are shapes

167
00:10:35,310 --> 00:10:38,970
and in Haskell circle and rectangle
don't have their own types

168
00:10:38,970 --> 00:10:42,430
so when you create a sure goal using the
circle constructor

169
00:10:42,430 --> 00:10:46,910
you get the shape and if you grade the
rectangle using the rectangle

170
00:10:46,910 --> 00:10:47,800
constructor

171
00:10:47,800 --> 00:10:52,990
you also get a shape alright that's the
main difference between

172
00:10:52,990 --> 00:10:56,010
algebra david ives in Haskell an

173
00:10:56,010 --> 00:10:59,670
the way you'd typically rides
inheritance hierarchies

174
00:10:59,670 --> 00:11:03,579
in or language now we can

175
00:11:03,579 --> 00:11:07,660
creates use this constructor to create

176
00:11:07,660 --> 00:11:11,029
particular shape a square so

177
00:11:11,029 --> 00:11:14,380
we take a value and for

178
00:11:14,380 --> 00:11:18,440
the a first gordon et there and weaker
in the shape

179
00:11:18,440 --> 00:11:22,520
I just duplicating that value here and
we can

180
00:11:22,520 --> 00:11:25,920
computer the Ariana Ave shape

181
00:11:25,920 --> 00:11:30,350
and that will return afloat as a year
you see that Ariana

182
00:11:30,350 --> 00:11:34,190
is defined on shape okay and then we do

183
00:11:34,190 --> 00:11:37,709
a better match here which is very much
like doing

184
00:11:37,709 --> 00:11:41,170
a dynamic dispatch so if

185
00:11:41,170 --> 00:11:45,040
the shape happens to be a sure call then

186
00:11:45,040 --> 00:11:48,459
II take by times are square

187
00:11:48,459 --> 00:11:51,670
if the shape is a rectangle

188
00:11:51,670 --> 00:11:55,380
I just multiply sites okay

189
00:11:55,380 --> 00:11:59,050
but so better matching got of course
bonds

190
00:11:59,050 --> 00:12:02,339
to virtual dispatch and when you think

191
00:12:02,339 --> 00:12:06,140
of algebra data dives as object
hierarchies

192
00:12:06,140 --> 00:12:11,440
now

193
00:12:11,440 --> 00:12:14,899
when we look at these an construct arse

194
00:12:14,899 --> 00:12:18,610
and that we define in here

195
00:12:18,610 --> 00:12:22,000
they implicitly defined constructor

196
00:12:22,000 --> 00:12:26,440
functions that they could float and
return to shape

197
00:12:26,440 --> 00:12:27,810
and

198
00:12:27,810 --> 00:12:30,500
second function that takes two floats

199
00:12:30,500 --> 00:12:34,740
and returns shape and these functions
are Gold Circle

200
00:12:34,740 --> 00:12:38,580
and rectangle so if you look at it I
believe you ask

201
00:12:38,580 --> 00:12:41,780
goal on TV of circle haskell the EU

202
00:12:41,780 --> 00:12:45,100
float arrow shape and if you ask

203
00:12:45,100 --> 00:12:49,210
what is a diaper rectangle it will tell
you it's a function that takes two

204
00:12:49,210 --> 00:12:49,920
floats

205
00:12:49,920 --> 00:12:53,090
and returns shape okay

206
00:12:53,090 --> 00:12:57,070
so again the anthology red and objects

207
00:12:57,070 --> 00:13:00,640
is still intact an you define

208
00:13:00,640 --> 00:13:03,840
your ship dives and really what you get
for free

209
00:13:03,840 --> 00:13:10,630
is constructor to construct the stocks
we mentioned

210
00:13:10,630 --> 00:13:15,310
maybe type a couple of times and maybe
dive is like a list

211
00:13:15,310 --> 00:13:18,920
that's either empty or has a single
value and

212
00:13:18,920 --> 00:13:22,560
usually convenient to just use list but

213
00:13:22,560 --> 00:13:25,650
if you really want to use the maybe died
maybe

214
00:13:25,650 --> 00:13:30,940
here in F sharp person where are ml
person where maybe is used a lot

215
00:13:30,940 --> 00:13:34,000
well you just defined in the large iraq
datatype

216
00:13:34,000 --> 00:13:37,770
may be okay is either not think are just

217
00:13:37,770 --> 00:13:41,589
and a and all we can define am

218
00:13:41,589 --> 00:13:46,330
lawler functions and that people are
kinda in a proud of

219
00:13:46,330 --> 00:13:50,600
because we can take a list and we can
take the heads

220
00:13:50,600 --> 00:13:54,339
but if the list is empty the return not
think

221
00:13:54,339 --> 00:13:58,530
and if the list is not empty we take the
hat

222
00:13:58,530 --> 00:14:02,570
Mb return just I'm not a big fan

223
00:14:02,570 --> 00:14:07,130
up this kind of definition of had
because we're not still stuck with this

224
00:14:07,130 --> 00:14:08,030
maybe

225
00:14:08,030 --> 00:14:11,070
so you really have now reduced to list

226
00:14:11,070 --> 00:14:15,190
to a singleton list budget yet not
really gotten the hat out and in order

227
00:14:15,190 --> 00:14:17,150
to gonna be a go for men may be a

228
00:14:17,150 --> 00:14:20,710
a2a you will still run in the same
difficulties

229
00:14:20,710 --> 00:14:24,390
as you had when you and from a list from
an empty list

230
00:14:24,390 --> 00:14:27,690
to a WYD proper head okay

231
00:14:27,690 --> 00:14:31,280
so that's why I am I typically don't
care about

232
00:14:31,280 --> 00:14:34,410
this maybe die bitch its it stands in
the way

233
00:14:34,410 --> 00:14:38,260
and it doesn't help you because she at
some point you have to get rid of it

234
00:14:38,260 --> 00:14:39,620
and you

235
00:14:39,620 --> 00:14:42,130
alright that was the end of Part wrong

236
00:14:42,130 --> 00:14:45,880
let's take a short break and I'll see
you shown

237
00:14:45,880 --> 00:14:46,530
for part two