1
00:00:00,000 --> 00:00:00,500


2
00:00:00,500 --> 00:00:03,510
DAVID MALAN: --indeed things just
get easier in many ways to do.

3
00:00:03,510 --> 00:00:04,730
So we transitioned to C--

4
00:00:04,730 --> 00:00:10,920
DOUG LLOYD: So, after nine years
of CS50 and eight years of people

5
00:00:10,920 --> 00:00:16,649
asking us to do Python instead of
PHP we finally made the switch.

6
00:00:16,649 --> 00:00:18,690
DAVID MALAN: Yeah, it just
felt like it was time.

7
00:00:18,690 --> 00:00:24,279
I mean Python has much of the momentum
in terms of web programming these days.

8
00:00:24,279 --> 00:00:26,070
But you know what put
me over the edge was,

9
00:00:26,070 --> 00:00:30,010
increasingly with the year have we
refined our approach to web programming

10
00:00:30,010 --> 00:00:34,350
in PHP, and we gave students a bit
more framework or representative code.

11
00:00:34,350 --> 00:00:37,610
We introduced MVC some
years ago, we gave them

12
00:00:37,610 --> 00:00:42,125
a library to facilitate SQL
queries in recent years, too.

13
00:00:42,125 --> 00:00:44,250
And this past year what I
was really inclined to do

14
00:00:44,250 --> 00:00:46,620
was give them a bit more
of distribution code

15
00:00:46,620 --> 00:00:49,980
for pset7, CS50 finance, that
would also handle routing,

16
00:00:49,980 --> 00:00:52,680
so that routing would get handled
through some one central file,

17
00:00:52,680 --> 00:00:56,400
like index.php, as is often the
case with a lot of PHP frameworks

18
00:00:56,400 --> 00:00:57,390
these days.

19
00:00:57,390 --> 00:01:00,150
And honestly, at that
point the closer we

20
00:01:00,150 --> 00:01:03,682
got to taking over routing from the
language, which historically had been

21
00:01:03,682 --> 00:01:06,390
done with individual files in
different directories and so forth,

22
00:01:06,390 --> 00:01:10,800
like we did in CS50, at that point we're
really losing the advantages of PHP

23
00:01:10,800 --> 00:01:15,810
and how there were so little
distance between the file system,

24
00:01:15,810 --> 00:01:19,860
and the files themselves, and the
controlling role that they played.

25
00:01:19,860 --> 00:01:21,767
So it felt to me, like
at that point, if I'm

26
00:01:21,767 --> 00:01:23,850
going to be giving them
this much boilerplate code

27
00:01:23,850 --> 00:01:25,558
and abstracting away
those details, let's

28
00:01:25,558 --> 00:01:27,990
just go to where most other
people have been skating.

29
00:01:27,990 --> 00:01:29,989
DOUG LLOYD: But it really
gave us an opportunity

30
00:01:29,989 --> 00:01:33,420
to dig deep into the equivalent
of week eight in the past

31
00:01:33,420 --> 00:01:37,377
and reinvent a lot of stuff
in terms of the staff.

32
00:01:37,377 --> 00:01:39,210
A lot of the staff was
familiar with Python,

33
00:01:39,210 --> 00:01:42,810
but some folks weren't, and
so it was a radical change.

34
00:01:42,810 --> 00:01:43,560
DAVID MALAN: Yeah.

35
00:01:43,560 --> 00:01:47,340
No, I mean I had to pick it up myself
certainly and try to map things

36
00:01:47,340 --> 00:01:49,670
past to the present, figure
out what still made since,

37
00:01:49,670 --> 00:01:51,856
what didn't, redo all
of the examples, think

38
00:01:51,856 --> 00:01:53,730
about how we would
implement pset7 and pset8,

39
00:01:53,730 --> 00:01:56,160
and so forth, but it's a wonderful--

40
00:01:56,160 --> 00:01:59,740
I think it's really strengthened
this part of the course

41
00:01:59,740 --> 00:02:02,370
now because Python just as
a practical matter is just

42
00:02:02,370 --> 00:02:04,050
so useful in other contexts too.

43
00:02:04,050 --> 00:02:06,420
I mean I really liked
this year how in pset6 we

44
00:02:06,420 --> 00:02:09,389
made that no longer
C, but Python as well

45
00:02:09,389 --> 00:02:12,180
to explore machine learning and
sentiment analysis, because now you

46
00:02:12,180 --> 00:02:15,540
had an opportunity to use the language
at the command line too before we

47
00:02:15,540 --> 00:02:17,169
transition in pset7 to web programming.

48
00:02:17,169 --> 00:02:18,960
So you sort of get the
best of both worlds.

49
00:02:18,960 --> 00:02:20,910
DOUG LLOYD: Yeah, we never
had the equivalent of a PHP

50
00:02:20,910 --> 00:02:22,410
at the command line, which
you could do, but we never

51
00:02:22,410 --> 00:02:23,250
had a problem set about it.

52
00:02:23,250 --> 00:02:25,170
DAVID MALAN: And was just never
the right language for command line

53
00:02:25,170 --> 00:02:27,503
scripts, like you could use
it, and I used it for years,

54
00:02:27,503 --> 00:02:30,690
but it's just not really done as much.

55
00:02:30,690 --> 00:02:32,940
So this was a win, and I was
nervous, because we never

56
00:02:32,940 --> 00:02:36,270
made this non-trivial of a change
like ripping out the whole language,

57
00:02:36,270 --> 00:02:40,050
most everything else has been
pretty marginal and incremental

58
00:02:40,050 --> 00:02:44,705
like switching to SQL lite
away from MySQL for instance.

59
00:02:44,705 --> 00:02:46,830
But this was impactful,
but it seemed to work well.

60
00:02:46,830 --> 00:02:49,290
We got through it, and
it was a enough time.

61
00:02:49,290 --> 00:02:52,835
It turns out, out of, I think
perhaps a bit of nerves,

62
00:02:52,835 --> 00:02:57,347
our Python lecture was the longest,
I think, of our lectures this year.

63
00:02:57,347 --> 00:02:59,430
I was just so excited to
talk about, but make sure

64
00:02:59,430 --> 00:03:01,987
we really hit every mark we could.

65
00:03:01,987 --> 00:03:04,570
DOUG LLOYD: But I think that
also, since you brought up pset6,

66
00:03:04,570 --> 00:03:08,190
we used to have the web server which
was the final cpset, we instead as you

67
00:03:08,190 --> 00:03:11,790
mentioned had a Python
command pset were also

68
00:03:11,790 --> 00:03:14,840
got to reintroduce some of the
problem from pset1 and pset2--

69
00:03:14,840 --> 00:03:16,506
DAVID MALAN: Yeah, which I really liked.

70
00:03:16,506 --> 00:03:19,590
DOUG LLOYD: --which is a great
way to allow students to get

71
00:03:19,590 --> 00:03:23,102
used to the language without having
to also solve a brand new problem.

72
00:03:23,102 --> 00:03:26,310
That was only half of it, the other half
what's actually solve a new problem,

73
00:03:26,310 --> 00:03:30,655
but it really took that
barrier to entry away as well.

74
00:03:30,655 --> 00:03:31,530
DAVID MALAN: I agree.

75
00:03:31,530 --> 00:03:34,071
But it's stuff like that I was
a little nervous about because

76
00:03:34,071 --> 00:03:38,760
unlike C to JavaScript where there
really is more of a line for line map--

77
00:03:38,760 --> 00:03:39,960
DOUG LLOYD: Or C to PHP.

78
00:03:39,960 --> 00:03:43,740
DAVID MALAN: Or C the PHP, whose
documentation too I always loved.

79
00:03:43,740 --> 00:03:46,915
I think php.net is speller and
Python's I think is less so.

80
00:03:46,915 --> 00:03:49,540
DOUG LLOYD: There's no equivalent
to [INTERPOSING VOICES] that.

81
00:03:49,540 --> 00:03:51,250
DAVID MALAN: It doesn't
line up quite as well.

82
00:03:51,250 --> 00:03:53,580
And so it's a bit more of a
mental leap, and that's OK,

83
00:03:53,580 --> 00:03:54,870
and I do think it worked well.

84
00:03:54,870 --> 00:03:56,578
And it's inevitable
if students are going

85
00:03:56,578 --> 00:03:59,310
to go pick up the language on
their own, so it was just perhaps

86
00:03:59,310 --> 00:04:02,152
a little overly conservative for
some time, but it worked out.

87
00:04:02,152 --> 00:04:04,860
DOUG LLOYD: Now in addition to
the question of PHP versus Python,

88
00:04:04,860 --> 00:04:09,290
there was then the follow up question
which is Python 2 or Python 3.

89
00:04:09,290 --> 00:04:12,720
DAVID MALAN: Yeah, so I did my
googling and my due diligence,

90
00:04:12,720 --> 00:04:15,420
talked to some folks, and
it felt like now there's

91
00:04:15,420 --> 00:04:18,810
enough momentum behind Python 3 that if
we're going to do this let's go all in

92
00:04:18,810 --> 00:04:21,690
and not have to upgrade
from Python 2 to Python 3

93
00:04:21,690 --> 00:04:25,120
in a year or two once we've
put out all these materials.

94
00:04:25,120 --> 00:04:27,120
And the reality is, so
far as CS50 is concerned,

95
00:04:27,120 --> 00:04:29,328
there's terribly few
differences between the language

96
00:04:29,328 --> 00:04:32,336
that our students might encounter
so early on, print for instance

97
00:04:32,336 --> 00:04:35,460
might be one such example, and there
are others they might encounter later,

98
00:04:35,460 --> 00:04:37,140
but I was comfortable with that.

99
00:04:37,140 --> 00:04:40,140
And there's just so many nice
new features syntactically

100
00:04:40,140 --> 00:04:42,714
and also functionally in
the library that comes

101
00:04:42,714 --> 00:04:44,130
with Python 3 that was compelling.

102
00:04:44,130 --> 00:04:44,430
DOUG LLOYD: Yeah.

103
00:04:44,430 --> 00:04:46,530
And I would say for
anybody who's watching this

104
00:04:46,530 --> 00:04:48,600
and is learning Python
themselves, I'd say

105
00:04:48,600 --> 00:04:51,490
that the one tip I would
give you for what I was doing

106
00:04:51,490 --> 00:04:54,974
was, when you're searching for
things in Python search for Python 3

107
00:04:54,974 --> 00:04:57,640
how to do something, not necessarily
Python how to do something.

108
00:04:57,640 --> 00:05:00,550
It's the one catch,
but I agree, if we had

109
00:05:00,550 --> 00:05:03,880
used Python 2 we would
have had to been playing

110
00:05:03,880 --> 00:05:06,067
catch up probably in 2017 instead of--

111
00:05:06,067 --> 00:05:06,900
DAVID MALAN: Agreed.

112
00:05:06,900 --> 00:05:09,200
We leap-frogged that.

113
00:05:09,200 --> 00:05:10,473