Monday, October 10, 2016

LoneStarRuby Conf 2013 - Elixir: Power of Erlang, Joy of Ruby by Dave Thomas

D.T.: Fata- cold?
D.T.: Then come up here and move your arms
around a bit. It's great.
Let me start off by explaining something, right. I
love Ruby. I've been doing Ruby since 1998. I
have read the Pick Axe book four times. And,
in fact, the latest revision went into print, well,
came off the presses on Friday, last week, a
week ago. And there's actually four copies to be
won lying around somewhere.
OK, so, I have a commitment to Ruby. Don't
get me wrong, right. So, I'm not here to
say, all you Ruby programmers are, you know, shouldn't
be here, it's crap. That's not the case.
However, I also think that we owe it to
ourselves, all of us, to keep an eye on
what's going on. Because if we don't, as Sandi
and I were discussing over lunch, if you don't
keep an eye on what's going on, then eventually
you will be out of date.
I, how many people here used to go to\ \
the No Fluff, Just Stuff Conference things? A few.\
\ Not many. I'm actually kind of proud of the\
\ fact that I've actually been banned from speaking
at\ \
those. Because I used to go to a Java,\ \
Java Conferences to talk about Ruby. And they said,\
\ oh, you can't do that anymore. Go talk about\
\ Java. And I said, no, I'm not gonna talk\
\ about Java anymore.\
\ So here I am in a Ruby conference. I\
\ am gonna talk about Elixir. So, why, why I'm\
\ gonna talk about Elixir. Elixir, to me, is
of the gateway to what I used to do\ \
to what I will be doing. And I've been\ \
looking for this for ten years and I've finally\ \
found it.\ \
So, the reality, I think, is that the future\ \
is functional and the future is concurrent. Those two\
\ technologies combined are pretty much the
only way we\ \
can continue to develop in, say, develop in terms\
\ of like, carry on, get things better, in software.\
\ We're kind of running into all sorts of obstacles.\
\ I'm sure you've all heard of. That's not gonna\
\ work. All right. Don't play. You've all heard
of\ \
Moore's Law. For the longest time I couldn't work\
\ out what he had to do with it. But,\
\ you know. Something like, you know, every
year you\ \
get twice as ugly or something. I don't know\ \
what it is.\ \
But, does he or does he not look like\ \
- this, this is Gordon Moore, right. He looks,\ \
to me, like Wendy's Dave Thomas guy, don't you\
\ think? But, and he's cheery. The reason he's
cheery\ \
is he's worth four billion dollars. But, he's cheery.\
\ And he came up with the famous Moore's Law.\
\ Now, Moore's Law is not actually what most
people\ \
say it is.\ \
Moore's Law says that the most cost-effective density for\
\ putting transistors onto a certain chip doubles
every two\ \
years. You know? It's nothing to do with performance\ \
or anything else. It just says that, Moore's Law\
\ says I can pack twice as many transistors
onto\ \
a chip and hit the most optimal point for\ \
cost-effectiveness.\ \
And it is pretty well born out. I, so\ \
I'll show in a moment - I copied the\ \
slide - what the first micro-persistor I used was.\
\ And it was definitely, I've definitely used
8008. And\ \
I've used the 6502, which is the, the last\ \
one on the bottom there. So both of them\ \
down there have maybe three thousand transistors on them,\
\ right.\
\ The axis on the y-axis here is exponential,
just\ \
as it was on Sandi's graph in her books.\ \
So, nowadays, my laptop has a core 07 in\ \
it, so it's up there. Unbelievably, I just like,\
\ last night when I actually got these numbers,
I\ \
couldn't believe it. My laptop has roughly a billion\
\ transistors in its processor. A billion transistors.\
\ And so I've actually measured the size of
the\ \
processor chip, cause Intel don't actually release the specs.\
\ And worked out that it's 121 square millimeters,
right.\ \
Which means that I have eight and a quarter\ \
million transistors per square millimeter in my processor. That's\
\ almost unimaginable. To learn how big a square
millimeter\ \
is, well, I finally found a use for this\ \
guy, because if you take the cross-section of a\
\ typical paperclip, it is about a square millimeter.
So,\ \
eight point six, or sorry, eight point two-five million\
\ transistors in that little space there.\
\ But what that really means in practice is
that\ \
you can't use, you know, a hundred, sorry, a\
\ billion transistors, like you would use ten
thousand. If\ \
you have ten thousand transistors, you basically have them\
\ all sitting as, you know, here's addition,
no go\ \
do this, no go do that. When you have\ \
that many, then you cannot coordinate them and have\
\ them all working on the same problem at the\
\ same time in any kind of meaningful way, right.\
\ If nothing else, the speed of light gets in\
\ your way.\
\ So, as a result, we're moving more and more\
\ towards multi-core and then hyper-threading
inside the cores to\ \
give us parallelism in our processors. We've all heard\
\ this, right. And we've all kind of like hoped\
\ the problem will go away, because this doesn't
really\ \
effect us right now. We can sort of get\ \
away with writing our Rails code on our servers,\ \
and we'll run n-processors and it kind of works.\
\ Except it kind of doesn't. You all know the\
\ issues that we face when we try and run\
\ that way.\
\ So, for those reasons, I think that we're
looking\ \
at a C-change, now. In order to continue to\ \
exploit this growing curve that's Moore's Law, we are\
\ going to have to find new ways of writing\
\ software, and I believe that functional programming
is one\ \
of those new ways.\ \
So, Elixir, I think, is functional, concurrent, pragmatic and\
\ fun. Let's spend a bit of time looking at\
\ the functional aspects. Now, initially I had,
like, I\ \
don't know, a hundred slides or something. I thought,\
\ screw that, let me just code instead. So I'm\
\ just gonna try and code some stuff up. You'll\
\ get to see how bad a typist I am,\
\ and between us, we may learn something.\
\ So, the first thing I'm gonna do is I'm\
\ actually gonna go into the shell, and Elixir
has\ \
a shell just like IRB, that's called IEX. I\
\ say it's just like IRB. It's nothing like
IRB.\ \
Let me just do something quickly here. OK. So,\
\ first of all, right now, I am orchestrating
roughly\ \
twenty-five independent processors in my Elixir shell here.\
\ So, the compiler is running in a separate
process.\ \
The Elixir command line, so the IEX command line\
\ is running in a separate shell, a separate
process.\ \
And they're all communicating back and forth. Which is\
\ kind of cool, and it's totally transparent.
Just like\ \
with IRB, I can do things like, you know,\ \
type an expression and it works. I can assign\ \
to a variable, right. So now a has the\ \
value one.\ \
Except I'm not assigning to a variable because this\
\ is a functional programming language, right.
All I'm doing\ \
when I say a equals one is asserting that\ \
a and one both have the same value. So,\ \
because of that, I can also say one equals\ \
a. And it says yup, no problem at all.\ \
Now what happens if I was to say two\ \
equals a. Well that now gets you something called\
\ a NoMatchError. And the reason for that is
that\ \
Elixir will only bind new values to variables if\
\ they're on the left-hand side of an equal
sign.\ \
So here, a already has a value one. So\ \
Elixir says, I'm sorry, but whatever is over here\
\ has to be a one as well. You know.\
\ Now, I told you about pattern matching. This
is\ \
an absolutely critical thing in any functional programming language.\
\ So let's have a quick look at pattern matching.\
\ We've already seen a equals one. Well, here,
we\ \
have a pattern which is some variable and a\
\ scalar. And that's gonna make, the variable
has the\ \
same value as the scalar. I could use Elixir\ \
list instead, OK. So now my variable a has\ \
been associated with that list one, two, three. But\
\ pattern matching goes pretty deep. So I can
actually\ \
say, [d,e,f] equals a. And what this is gonna\ \
do is say, OK, on both sides I'm expecting\ \
to see a list of three elements. Well, it\ \
just so happens that a is a list of\ \
three elements. So this is gonna match, and it's\
\ gonna assign one to d, two to e, and\
\ three to f. Yeah?\
\ All right. We can take it a bit further\
\ than that. I can, for example, take a list\
\ like a, and split it into its head and\
\ its tail. So I can say, head followed by\
\ tail in a list equals a. And if I\
\ do that, then a, so the head is one.\
\ The first element. And the tail is the rest\
\ of the list. Yeah? So pattern matching allows
me\ \
to construct and deconstruct my data. And as you'll\
\ see a bit later on, it also allows me\
\ to write really large programs without any
conditional logic.\ \
Typically also without any loops.\ \
All I'm doing is declaring stuff in my code.\ \
So let's start off and actually write some Elixir\
\ code, like, outside this thing. Let's go into
an\ \
editor here. So I'll just use Sublime. And let's\
\ just write a really quick bit of Elixir. Let's\
\ do defmodule, let's call it Sequence because
eventually it\ \
will be. Oops.\ \
All right. So, all Elixir code goes into a\ \
module. So, that's not quite true. When you want\
\ to compile Elixir code, you have to put it\
\ into a module. And the reason for that is\
\ that Elixir runs on the Erlang virtual machine,
and\ \
that's how the Erlang virtual machine manages its code.\
\ You may have heard the Erlang/Elixir do a
hot-swapping\ \
of code. Well, I can hot swap?? Elixir code\ \
as well. So when you're running an application, I\
\ can replace it. I replace things at the module\
\ level. So that's why I have to put my\
\ code in a module.\
\ So I can write a method, let's say hello,\
\ and we'll just say IO. All right. And then\
\ down here I can call it by saying Sequence\
\ dot hello. Yeah? If I run that code, wow.\
\ So, if you are squinting, you will think that,\
\ that was Ruby, right. Very, very similar.
And, in\ \
fact, Elixir takes a whole bunch of things from\
\ Ruby and runs with them.\
\ For example, here, you think that is a function\
\ definition. Which it kind of is. But actually,
it's\ \
alo a function call. Because, in Elixir, def is\
\ a function call. So is defmodule. And this
function\ \
takes two parameters. The name of the method and\
\ the body of the method. And it just so\
\ happens, in Elixir, there's a bit of syntactic
sugar\ \
that lets me write the body as do end.\ \
Underneath the covers, that actually gets translated into, oops,\
\ that.\
\ That do colon there is actually just like
a\ \
Ruby hash. It's just a named parameter that I'm\
\ passing in. So, for example, an if statement
in\ \
Elixir, if I have an if statement like if\ \
a equals b, sorry, do something, else something else.\
\ Yeah? Actually, again, it's just syntactic
sugar. Because if\ \
is also just a method. Actually it's a macro,\ \
effectively. It's a method. It takes three parameters. It\
\ takes the expression, it takes the do block,
and\ \
it takes the else block.\ \
And that means that I can rewrite the entire\ \
language if I want to, right. Everything is soft.\
\ We always talk about metaprogramming. This
is serious metaprogramming.\ \
And unlike other languages you might name, it's actually\
\ done in a very controlled style. First of
all,\ \
metaprogramming in Elixir is done by manipulating the parse\
\ tree, which actually happens to be an Elixir
data\ \
structure. It is done in a hygienic way, so\ \
it will not overwrite any of your local values\ \
accidentally. And it is done in a scoped way.\ \
In Elixir, every change you make to the environment,\ \
whether it's defining macros or including a module, whatever\
\ it might be, is lexically scoped. So, for
example,\ \
I can say apply this differentiation of a particular\
\ operator only to the parameters of this method
call,\ \
which can come in very, very useful.\ \
However, that's not why we're here. So let's have\
\ a look at something slightly different. Let's
have a\ \
look at lists. So what I want to be\ \
able to do is say something like IO dot\ \
puts Sequence, oops, dot length of some list. Yeah?\
\ So, how am I about, how am I gonna\
\ go about doing that? Well, obviously there
are built-in\ \
ways of doing this, but let's do it the\ \
hard way.\ \
So I need to find a method called length\ \
that's gonna give me the length of the list.\ \
In a functional programming language, I'm gonna use pattern\
\ matching to do most of the work for me.\
\ So I'm gonna say, the length of an empty\
\ list, k, I'm pattern matching here, an empty
list.\ \
And what's the length of an empty list? Zero.\ \
So, that's taken care of one case. The other\ \
case is a non-empty list. So let's have a\ \
look at a list which has a head or\ \
a tail. And tail is potentially empty. So what's\
\ the list of that?\
\ Well, we know the head contributes one. And
the length of the tail is simply length(tail). That\
\ make sense? Nice little recursive definition.
If I run\ \
that, it blows up.\ \
AUDIENCE: [laughter]\ \
D.T.: But that's OK. Oh. Kernel dot length conflicts\
\ with local function. Look at that. I actually
have,\ \
although it's untyped, it actually does a fair amount\
\ of work to make sure that, for example, I\
\ haven't mispelt function names, that I haven't
accidentally overwritten\ \
function names. So let's call that len rather than\
\ length. There we go. So the problem is here.\
\ First of all, there's my result, five. Also
notice,\ \
variable head is unused, because here, when I did\
\ the pattern match, I said head to be the\
\ head of the list, tail to be the tail\
\ of the list, but then my code, I only\
\ ever use tail. So, Elixir has a convention
that\ \
if I've got an underscore in front of a\ \
variable name, it says, I don't plan to use\ \
this variable.\ \
So by doing that, my warning goes away. I\ \
could, in fact, just say underscore here, and again,\
\ that just works. But I personally prefer to
leave\ \
the variable name there because it documents what I'm\
\ doing. That's a typical recursive definition.
You don't find\ \
yourself doing those too often. K, but one of\
\ the things I like about this, from a, an\
\ aesthetic point of view, is that my definitions
here\ \
are pretty much the specification of what I want\
\ to write. So the length of an empty list\
\ is zero. That's, you know, someone could tell
you\ \
that. Right. The length of any other list is\
\ one plus the length of the rest of the\
\ list. OK, it's a specification.\
\ Let me show you another specification. This
one is\ \
like, it's almost like you'll groan when you see\
\ it, right, but it's actually useful for my
purposes.\ \
Fibonacci numbers, right. One, one, two, three, five, eight,\
\ whatever. Each number is the sum of the previous\
\ two numbers. If you go to a math book\
\ or Wikipedia or whatever else, they will tell
you,\ \
the definition of a function of fibonacci is the\
\ fibonacci of zero is one. The fibonacci of
one,\ \
oops, is also one. And then the fibonacci of\
\ some number n is the fibonacci of n minus\
\ one, woops, plus the fibonacci of n minus
two.\ \
That is the mathematical definition of fibonacci, yeah.\
\ But, if we were to convert that into running\
\ Elixir code, all I have to do is say,\
\ right, and now I have code. So now I\
\ have to say Sequence dot fib(20). There it
is.\ \
Is that cool? Yeah. Now, at this point, you're\ \
probably saying, but, Dave, that's all very well, but\
\ how many times a day to you actually calculate\
\ the fibonacci sequence, right? Eh, OK. No
more than\ \
five or six, I have to tell you.\ \
But that's OK, because the reality is that this\
\ form of expressing code I do all the time,\
\ in fact, when I'm writing Ruby code, I will\
\ sit there and I'll typically work out the
various\ \
definitions, and I'll write them down, not as tidy\
\ as this, but I'll write, you know, zero is\
\ one. And I'll write, you know, empty request
means\ \
I send back a form or whatever. You know,\ \
all those kind of things I would write down.\ \
I do the same when I'm writing Elixir, but\ \
it just turns out, coincidentally, I'm writing code, right.\
\ That's really cool.\
\ So let's look at a few more things in\
\ here. First of all, let's define a map function.\
\ It's the same as the Ruby map function in\
\ that I want to take a collection and apply\
\ a function to each element in that collection
and\ \
return the resulting set of values. So, I'm gonna\
\ say, let's start with a use because that's
always\ \
a good way of doing things. So in this\ \
case, I will do IO dot puts Sequence dot\ \
map(). OK, so I'm gonna map, in this case,\ \
a list. So let's go four, five, six, seven,\ \
and I'm gonna map it through my fibonacci function,\
\ all right. So my fibonacci sequence is gonna
be.\ \
Oh, actually I'll do it this way.\ \
OK. So I want to say, Sequence dot fib\ \
of some number n for each successive element. So\
\ how do I do that? Well, the long-winded way\
\ is to say, I'm gonna write an anonymous function,\
\ that's just like a lambda in Ruby, that takes\
\ a parameter n and its body is, will calculate\
\ the sequence dot fib of n, yeah. So my\
\ map function takes the collection and it takes
the\ \
function. So let's start off with that.\ \
So the map function takes a collection and it\
\ takes a function. So what are my special cases\
\ here? Well I think it's the same as the,\
\ the counting one, the length one we did, and\
\ that is, if I want to map over an\
\ empty collection, then what's gonna come back
is also\ \
an empty collection, yeah. That's all I need to\
\ do. So now I'm dealing with the case of\
\ mapping over a non-empty collection. So we're
gonna do\ \
that head-tail thing again. And we get the function.\
\ So now, what are we gonna do here?\
\ Well I have to apply the function to the\
\ head of the list, because that's part of what\
\ I have to do. And then I have to\
\ generate the rest of the list, which is being\
\ mapped. So I can express that by saying, OK,\
\ fun dot head, so, let's clone the function
on\ \
the head, and then I'm gonna build the rest\ \
of the list by mapping on the tail. And\ \
that map function is the same map function here.\
\ So what's gonna happen? Let me shrink the
if I shrink the font down, can you see\ \
it in the back? Is that OK? OK, thank\ \
you. All right. So, what's gonna happen here is,\
\ each time around this recursion, it's gonna
take one\ \
extra element of the list. It's gonna convert that\
\ by running it through the function and build
a\ \
new list where that element's there. Then at the\
\ very end, I'm gonna run out of list. So\
\ tail will be empty, so then it comes with\
\ this clause and it's gonna terminate the recursion.\
\ So now in theory, if I run this code,\
\ oh. Ah! OK. Let me change this to be\
\ an inspect. All right. I've just run afoul
of\ \
some Erlang disgustingness. All right. But anyway. So there\
\ is me mapping my list. Yeah. That's not bad\
\ is it. It's really quite straight-forward.
It's really understandable\ \
code. It is recursive.\ \
But so what? So here I've written a map\ \
function and a fibonacci function, and in both cases,\
\ I have no explicit loops, I have no explicit\
\ conditional logic, which is absolutely fantastic.
OK.\ \
So, you say to me, but Dave you're doing,\ \
you know, testing by inspection. Can we do better\
\ than that? You bet we can. So let's do\
\ something like class TestIt. No, not class.
Oh, dear,\ \
dear, so that's the problem. You switch back and\
\ forth between the two and you are totally
lost.\ \
It's horrible. All right. So.\ \
So there's my test module. I need to use\ \
ExUnit, which is a testing framework, and in fact,\
\ I actually have to start the testing framework.
Cause\ \
Erlang's about lots of processes. It actually runs in\
\ its own separate process. So I'm gonna run
its\ \
own process. So now I can write some tests.\ \
Test "Basic fib function works" do. All right. So\
\ now assert Sequence dot fib(10) equals one
oh nine\ \
four six, end. Test "Basic Map works" do. So\
\ now I'm just gonna cut and paste over here,\
\ so I have. OK, so. Five, eight, thirteen,
oops.\ \
I made a typo. Bad me.\ \
Okee dokes. So, folks, that's a basic unit test\
\ written using the ExUnit framework. So let's
see what\ \
happens when I run that. Function test two undefined.\
\ I'm sorry, where?\
\ AUDIENCE: [indecipherable]\
\ D.T.: To get the fun thing done. Yeah. I\
\ think that's, that's not the error I'm actually
seeing\ \
here. I'm seeing a problem that test two is\ \
undefined, which means that this test method here, which\
\ probably means that I've forgotten something.
Give me a\ \
quick second. I had a cheat sheet just to\ \
make sure I don't do this. Oh, I'm sorry,\ \
it's not use ExUnit, it's use ExUnit dot case.\ \
That's right. Oh, there you go. There worked loads,\
\ a whole lot better.\
\ Cannot invoke remote function Sequence dot
fib/1.\ \
AUDIENCE: [indecipherable]\ \
D.T.: Cause it is.\ \
AUDIENCE: [indecipherable]\ \
D.T.: Is that what it is? Oh, sorry. It'll\ \
be eighty-nine in that case, isn't it? All right.\
\ I'm having a really bad time here. What's
wrong\ \
with that?\ \
What on earth is wrong with that?\ \
It is not the most efficient way of doing\ \
that, I have to tell you. Oh, oh, oh.\ \
OK. Here we go. Here we go. Thank you.\ \
Thank you.\ \
All right. So, my test failed. Notice the error.\
\ Expected five, eight, thirteen, twenty-one,
but got five, eight,\ \
thirteen, twenty-two. In most of the testing frameworks, I'd\
\ have to say assert_equals that comma result.
Right. Remember\ \
I said that metaprogramming is wonderful in Elixir. Here\
\ my test function actually intercepts calls
to equals equals\ \
inside that block. It runs the original equals equals,\
\ but it still has access to the two operands,\
\ as a code fragment. So it can take that\
\ code fragment, convert it back into text,
and then\ \
use it here to show me the two values.\ \
That is pretty damn cool. Just think of the\ \
cool stuff you could do with that. All right.\ \
So, let's push the boat out. I talked about\ \
it being a parallel, a concurrent language, so let's\
\ go back to our map sequence here, and let's\
\ see if we can make this parallel. And I\
\ have, five minutes? Is that right? Somebody?
God help\ \
me. All right.\ \
So, I'm gonna write a parallel version of map.\
\ So let's call it pmap, all right. it's gonna\
\ take a collection and a function, yeah. And
what\ \
it's gonna do is this is what I'm gonna\ \
show you how I think about functional programming. It's\
\ gonna take the collection, it's going to run
each\ \
of the elements of this collection in a separate\ \
sub-process. So we're gonna call it spawn_children(), let's call\
\ it fun rather than function just to save myself\
\ a little bit here. And then after we've done\
\ that we're gonna collect the results.\
\ So what's this pipe operator here? It's actually
a\ \
pipe line. Just as if we're in the Unix\ \
shell and I can pipe the output of one\ \
thing into the next. Here, I've taken my collection.\ \
I am piping it to spawn children, and then\ \
the results will come back. I'm gonna pipe it\
\ to collect the results. The reason this is
significant\ \
is not that it saves me a little bit\ \
of typing. The reason it's significant is it changes\
\ the way I think about programming. Because
now I\ \
think about programming as mapping from a to b,\
\ and not from doing something. So here I have\
\ a collection. I'm gonna map it into a set\
\ of processes. And then I'm gonna map those
processes\ \
into some results. So it's a transformation. It's not\
\ a explicit imperative way of coding.\
\ So how do we code those things up? Well,\
\ I know I'm running out of time so I'm\
\ gonna do this pretty quickly. So, I have my\
\ spawn_children function, and it's gonna take
a collection and\ \
a function, and it's gonna take the collection and\
\ it's gonna map it, so again I'm gonna transform\
\ the collection by mapping it through, that's
spawn_child- yeah.\ \
A function called spawn_child. So it's going to spawn\
\ one independent child, K.\
\ And clearly I need to pass it the element\
\ from the collection and the function. So what
I\ \
could do is to say something like this, elem\ \
comma fun, and then wrap this in an anonymous\ \
function, just as you would in Ruby. So I'd\ \
say function, so I'm gonna receive an element. And\
\ I'm gonna do that. But because that's such
a\ \
common pattern, Elixir has a short cut for this.\
\ I could say map it through spawn_children,
where the\ \
first element is the first parameter passed in, and\
\ Elixir automatically is gonna wrap that into
an anonymous\ \
function. It's then gonna say, but wait a minute,\
\ this anonymous function is nothing more than
a call\ \
to a named function, so it's gonna rewrite that\
\ again into a direct function call to that
named\ \
function, once for each element in the collection. That's\
\ pretty sweet. Right.\
\ So what does my spawn_child look like? It's
gonna\ \
look like something like it receives a value and\
\ it receives a function, and this is where
it\ \
gets a little bit tricky, because what we're gonna\
\ do is we're gonna spawn ourselves and make
ourselves\ \
run in a separate process. To run in a\ \
separate process in Elixir, I'm gonna call spawn, and\
\ there are many different ways of calling spawn.
The\ \
one I'm gonna choose is actually slightly verbose, but\
\ it's actually easiest to explain. I'm gonna
spanw a\ \
function in my current module that's gonna be called\
\ :child, and I'm gonna pass it three parameters.
So\ \
the parameters I'm gonna pass it are the value,\ \
I'm gonna pass it the function and I'm gonna\ \
pass it myself. Self is a, a current process.\ \
And the reason I want to do that is\ \
that my child is gonna send me a message\ \
back to say here's the value.\ \
So now I'm gonna run my child function. Do\ \
I really write all these one-line functions when I'm\
\ writing Elixir? Actually I do. And typically
them on the same line just because it's like\ \
easier to do that. I find this kind of\ \
decomposition. Step at a time, just decomposing it to\
\ single-line functions. It's really a powerful
way to think\ \
about it.\ \
So my child method, what that's gonna do is\ \
actually send - so, OK, so the child method\ \
receives a value, a function, and a pointer to\
\ self, which is actually the parent of the
child,\ \
yeah. SO what it's gonna do is send a\ \
message back to the parent. And I'm using Elixir,\
\ oops, arrow thing here, which means send a
message.\ \
And what message am I gonna send? I am\ \
gonna respond with my own process id, and the\
\ value. And the value is fun dot. So I'm\
\ calling the, the function on the current value.
OK.\ \
So here, what I've done is come back to\ \
my original pmap up top there. I've done the\ \
first part. I've spawned a separate process for each\
\ value in my incoming collection, and it's
now running\ \
independently somewhere. So what you do now is collect\
\ your results back. And to collect your results
back,\ \
I'm just gonna go down some lines so you\ \
can see it, it's gonna look like this. Collect_results.\ \
Now, the result of running all of this mapping\ \
is that I'm gonna take a set of values\ \
and convert it to a set of processes. So\ \
my collect_results is gonna get passed a set of\
\ process ids. And what I need to do is\
\ to get the results from those processes in
order.\ \
So, I'm gonna use a, a receive block, and\ \
that means go and receive a message, and pattern\ \
matching allows me to control what message I receive,\
\ when.\
\ In this case, what I want to do is\
\ to receive the message from a particular pid.
OK.\ \
So, I haven't got that yet. I'll get it\ \
in a sec. And it's gonna come back with\ \
a pid and a value, and out of all,\ \
the only part of that I'm interested in is\ \
the value. So, my problem here is that my\ \
collect_results has got a whole list of pids. I\
\ want the result to be a particular pid, so\
\ I'm gonna break that into two methods here.
So\ \
I'm gonna say pids transform them by mapping them\
\ through the function collect_result_for_pid.
And, again, it takes the\ \
pid as a parameter, and now I'm gonna have\ \
a method called def collect_results_for_pid, it takes a pid,\
\ and away we go.\
\ AUDIENCE: You have a naming, naming problem
on line\ \
twenty-six.\ \
D.T.: Line twenty-six. That. Oh, look at that. Collect.\
\ Yes. Thank you. I mean, the chances of it\
\ actually running the first time are about
zero. But\ \
we'll, we'll give it a go. All right. I\ \
mean, given my luck this afternoon, it's gonna be\
\ like, it'll blow my machine up or something.\
\ So, what I've done here is actually a fair\
\ amount of typing. I don't have to make it\
\ so big. but what I'm doing, what I'm doing\
\ instead is spelling it out so I can show\
\ you my thought process. So if we run this,\
\ and it blows up. What a surprise. Function
collection_results\ \
does not exist. So somewhere I've used collection_results.\ \
AuDIENCE: [indecipherable]\ \
D.T.: Should be collect_results, right. And it fails again.\
\ You see, it's very. Oh, that's my test failing.\
\ All right. Fix that up. Yay! All right.\
\ So I changed that form being map to pmap.\
\ Still works. Cool. Just to make sure that
can actually see that, and I know I'm about\ \
to get bum rushed, right. OK, I'm just gonna\ \
show you just one cool thing. And that is,\ \
if I take that file, which is called what?\ \
T. My god I'm imaginative. OK.\ \
Oh, that's the wrong t. Hmm. Oh, there it\ \
is. If there are any Brits you'll know why\ \
I called this directory BluePeter, because here's some examples\
\ I prepared for earlier. All right. So now
I've\ \
loaded that file into my Elixir, so I can\ \
call Sequence. Actually I can't call Sequence. Actually I\
\ can't call Sequence. Import Sequence. All
right, so now\ \
I can say fib(30), say, and there's the result.\ \
So, if I now run fib, if I run\ \
regular map. So map, say, thir- oh, let's make\
\ it thirty-five. Five, thirty-six, thirty-seven,
thirty-eight, thirty-nine, and forty.\ \
And I map it through the fibonacci thing. Hey,\
\ everybody's picky. All right. And it blows
up. Why\ \
did it blow up? Oh. Thank you. OK. So,\ \
if, while it's doing this, I call up my\ \
activity monitor, you'll see that I'm running on one\
\ CPU, you know. And there's the results. If
I\ \
change this to use pmap, I'm running on all\ \
That cool? All right. I've got, I could go\ \
on for days, but then you won't let me\ \
come back. So, I want to come back. So.\ \
V.O.: Thank you Dave.\ \
D.T.: I'm done. Thank you.}