0:00:17.190,0:00:19.160
STEPHAN HAGEMANN: Thanks for coming.

0:00:19.160,0:00:21.130
You came to Big Rails.

0:00:21.130,0:00:23.190
If you did not intend to go to Big Rails.

0:00:23.190,0:00:24.890
My wife's actually currently in a plane,

0:00:24.890,0:00:27.630
and so for some reason I had to think of a[br]plane metaphor.

0:00:27.630,0:00:29.350
And her flight just got canceled.

0:00:29.350,0:00:30.360
If you did not intend to go to Big

0:00:30.360,0:00:33.140
Rails, which I can understand. It can be messy.

0:00:33.140,0:00:37.750
Better get out now, the flight might be bumpy.

0:00:37.750,0:00:39.640
So I want to talk about really large Rails

0:00:39.640,0:00:41.840
apps, and my buddy Austin suggested, a couple[br]days

0:00:41.840,0:00:45.730
ago, let's use GitHub. Search for Rails. Sort[br]by

0:00:45.730,0:00:49.050
size. This is officially the biggest Rails[br]app you've

0:00:49.050,0:00:52.059
never heard of on GitHub. It's open source.[br]It

0:00:52.059,0:00:55.269
also only has eight models. Which is because[br]they

0:00:55.269,0:00:56.260
upload all their assets.

0:00:56.260,0:00:58.050
It's a hundred and fifty megabytes big, and[br]that's

0:00:58.050,0:01:01.230
not the kind of big I'm talking about. I

0:01:01.230,0:01:05.430
want to talk about component-based Rails architectures[br]and why

0:01:05.430,0:01:09.430
you should refactor towards them. And how[br]to do

0:01:09.430,0:01:11.770
that.

0:01:11.770,0:01:15.130
Component-based Rails architectures is a term[br]I use. I

0:01:15.130,0:01:19.020
tag my Tweets about it as #cbra, and I'm

0:01:19.020,0:01:20.360
the only person in the world doing it. But

0:01:20.360,0:01:24.590
if you use that too, I'm certainly replying.[br]And,

0:01:24.590,0:01:27.510
here's a memo for that.

0:01:27.510,0:01:32.410
If I can get one idea in straight off

0:01:32.410,0:01:35.720
the bat, then it is that this is, this

0:01:35.720,0:01:38.510
stuff is easy. If you can think about your

0:01:38.510,0:01:41.500
application by writing, by painting boxes[br]and drawing arrows

0:01:41.500,0:01:43.920
for things that depend on each other and how

0:01:43.920,0:01:47.040
they interact, then you know everything else[br]you need

0:01:47.040,0:01:50.040
to know about component-based architectures[br]and you just need

0:01:50.040,0:01:52.770
to start doing them to what I think will

0:01:52.770,0:01:55.560
be improving the quality of your application[br]and the

0:01:55.560,0:01:57.680
quality of your code.

0:01:57.680,0:02:01.390
Since I called this talk refactoring towards[br]component-based Rails

0:02:01.390,0:02:03.360
architectures, this talk assumes that you[br]all know what

0:02:03.360,0:02:06.420
that is. So, no. I will actually explain what

0:02:06.420,0:02:07.450
that is to you. You're not screwed if you

0:02:07.450,0:02:09.758
don't know what that is.

0:02:09.758,0:02:14.810
So, I started talking and actually writing[br]applications pretty

0:02:14.810,0:02:19.310
much exclusively in this kind of odd way,[br]so,

0:02:19.310,0:02:23.480
like, two, three years ago. And it, I wrote

0:02:23.480,0:02:26.090
this sample app that you find at this address.

0:02:26.090,0:02:29.900
I'm shageman with one n on GitHub. And there's

0:02:29.900,0:02:32.590
just the_next_big_thing sample app.

0:02:32.590,0:02:35.510
It does nothing. Only it shows you the structure

0:02:35.510,0:02:37.450
and how to hook certain things up. And let

0:02:37.450,0:02:40.060
me quickly go through what that means. So[br]this

0:02:40.060,0:02:42.540
is the root of the project. And, as you

0:02:42.540,0:02:45.410
can see, it's not a Rails application. But[br]it

0:02:45.410,0:02:47.810
is. There's actually no particular prescribed[br]way on how

0:02:47.810,0:02:50.150
to do this. In many apps where I've done

0:02:50.150,0:02:51.870
this, the Rails application is still at the[br]root

0:02:51.870,0:02:54.120
of the project. But in this particular one,[br]I

0:02:54.120,0:02:56.700
chose the other version, where Rails moves[br]into a

0:02:56.700,0:02:58.940
sub-folder and, as you can see here, so this

0:02:58.940,0:03:02.550
looks pretty much like Rails. But it doesn't[br]have

0:03:02.550,0:03:05.080
an app folder. There, and you might be asking,

0:03:05.080,0:03:06.880
is there nothing happening in this app?

0:03:06.880,0:03:10.209
Well, there is. Here, we see there's, at least

0:03:10.209,0:03:12.240
one thing is mounted. There is an engine that's

0:03:12.240,0:03:18.250
mounted. So, Teaser::Engine. I don't know[br]if anyone has

0:03:18.250,0:03:21.260
ever downloaded that from RubyGems. Well,[br]don't try. I

0:03:21.260,0:03:22.640
don't know what that gem would be. But I

0:03:22.640,0:03:23.990
didn't upload it there.

0:03:23.990,0:03:25.430
Let's look at the gem file to find out

0:03:25.430,0:03:28.280
what that gem is. So this is the, the

0:03:28.280,0:03:30.270
only kind of trick that you need to employ

0:03:30.270,0:03:33.700
when you want to write component-based architectures.[br]And that

0:03:33.700,0:03:36.310
is, you use gems as not a, not as

0:03:36.310,0:03:39.060
a distribution mechanism, but just as a packaging[br]mechanism

0:03:39.060,0:03:41.430
for parts of your application.

0:03:41.430,0:03:43.400
You leave them in the same source tree and

0:03:43.400,0:03:45.620
you just reference them. And then you can[br]do

0:03:45.620,0:03:47.700
everything you're used to with those things.

0:03:47.700,0:03:49.489
So, as you can see here, if we go

0:03:49.489,0:03:51.430
to the components folder, we expect to see[br]a

0:03:51.430,0:03:53.489
teaser folder. Ah. Which we do. And there[br]is

0:03:53.489,0:03:56.020
an engine. So that's just a Rails. Ah, sorry.

0:03:56.020,0:03:58.700
A RubyGem that is pimped a little bit so

0:03:58.700,0:04:02.900
it has Active star. All the Actives. So Rails

0:04:02.900,0:04:05.650
in it. And this, indeed, now looks like an

0:04:05.650,0:04:06.670
application.

0:04:06.670,0:04:09.730
If you don't know what an engine is, it's

0:04:09.730,0:04:12.099
essentially Rails minus everything you need[br]to actually get

0:04:12.099,0:04:15.800
it booted. So, if we look in here, what

0:04:15.800,0:04:18.279
we can see, you know, there's a routes file

0:04:18.279,0:04:21.199
and this engine is doing something, right.[br]There's. It's

0:04:21.199,0:04:23.960
normal. Here we'll look in apps. So the controller.

0:04:23.960,0:04:26.400
So you can see this is namespaced under the

0:04:26.400,0:04:28.930
name, which is nice. But it's just, you know,

0:04:28.930,0:04:32.789
your ordinary controller. But, again, everything's[br]namespaced.

0:04:32.789,0:04:38.220
There's assets in here. Yeah. This is pretty[br]normal,

0:04:38.220,0:04:42.620
standard stuff. We even have migrations down[br]here in

0:04:42.620,0:04:45.510
this engine. What else would I talk about?[br]Well,

0:04:45.510,0:04:48.190
ah, yes. Maybe the coolest part. There's specs[br]in

0:04:48.190,0:04:49.990
here. I like to use rSpec. That could be

0:04:49.990,0:04:52.220
tests in here if you're doing this and liking

0:04:52.220,0:04:53.330
those better.

0:04:53.330,0:04:57.980
But there's, there's a spec here for this[br]controller

0:04:57.980,0:05:00.080
that I just looked at, and the cool thing

0:05:00.080,0:05:02.540
about that, and I don't know how to emphasize

0:05:02.540,0:05:04.000
it more than to come up to the front

0:05:04.000,0:05:07.340
of the stage and say it's very important that

0:05:07.340,0:05:09.240
I can run these tests, and the only thing

0:05:09.240,0:05:11.300
that these tests are gonna load is the code

0:05:11.300,0:05:13.600
from this engine, and not from all the other

0:05:13.600,0:05:16.490
stuff in my Big Rails application.

0:05:16.490,0:05:20.090
So, just by there being a spec folder in

0:05:20.090,0:05:21.800
here, I have a part of which I can

0:05:21.800,0:05:24.650
prove that it is small. Or, at least, as

0:05:24.650,0:05:26.620
big as only this part, and not depending on

0:05:26.620,0:05:28.710
anything else.

0:05:28.710,0:05:33.100
I, for, what I typically do, then, to kind

0:05:33.100,0:05:36.100
of bind these components together - and there's[br]more

0:05:36.100,0:05:37.840
in this folder, as you can see. There's this

0:05:37.840,0:05:40.300
event counter signup and whatnot. So I write[br]a

0:05:40.300,0:05:42.290
little script here that binds all the tests.[br]As

0:05:42.290,0:05:43.840
you can see there's some request specs and[br]some

0:05:43.840,0:05:46.320
Jasmine specs, and if you look into the other

0:05:46.320,0:05:49.490
folders, you find more of these test files.[br]And

0:05:49.490,0:05:54.169
then in the root. I typically put this, this

0:05:54.169,0:05:56.500
build script here. And what this does is loops

0:05:56.500,0:05:58.449
over all the folders and tries to find test

0:05:58.449,0:05:59.320
files and runs them.

0:05:59.320,0:06:01.330
So you can still run this as the app,

0:06:01.330,0:06:03.449
as the specs to one application. But if you

0:06:03.449,0:06:05.229
just change one part, you can just test that

0:06:05.229,0:06:08.699
one part and still be sure that it works.

0:06:08.699,0:06:11.590
So, now that you know what these applications[br]are,

0:06:11.590,0:06:13.190
you don't need to think about all those little

0:06:13.190,0:06:14.780
things. How to hook them up. You'll find out

0:06:14.780,0:06:16.410
about that when you try to do it. What's

0:06:16.410,0:06:17.630
important about them is that you can now talk

0:06:17.630,0:06:22.280
about your application as a sum of smaller[br]parts.

0:06:22.280,0:06:24.030
For example, the app that I just showed you

0:06:24.030,0:06:26.759
is a empty Rails container that mounts an[br]engine

0:06:26.759,0:06:28.259
that uses two other engines that I didn't[br]talk

0:06:28.259,0:06:30.110
about and another gem.

0:06:30.110,0:06:31.900
This is a project that it recently, it o-

0:06:31.900,0:06:34.389
that was a tiny project, so this was Tiny

0:06:34.389,0:06:37.650
Rails. Empty Rails container. Two engines.[br]The one was

0:06:37.650,0:06:40.660
using two gems to talk to APIs, and the

0:06:40.660,0:06:45.290
other one was just straight-up, pretty much[br]normal application.

0:06:45.290,0:06:47.320
Or this site that was a travel site. It

0:06:47.320,0:06:49.780
was so small, we didn't even bother to use

0:06:49.780,0:06:51.620
engines. We just put a couple gems in there

0:06:51.620,0:06:53.630
that were talking to APIs.

0:06:53.630,0:06:55.889
This site was a bit bigger. It started out

0:06:55.889,0:06:59.080
like this. We had six engines, one gem, and

0:06:59.080,0:07:01.139
it ended up about like this. But this is

0:07:01.139,0:07:03.150
a lie, because there were about twice as many

0:07:03.150,0:07:05.440
engines, only the other ones are abstracted[br]away for

0:07:05.440,0:07:07.060
clarity in this picture. As you can see, it's

0:07:07.060,0:07:08.150
a very clear picture.

0:07:08.150,0:07:12.210
Now, this looks chaotic. But the fact that[br]I

0:07:12.210,0:07:14.770
can draw such a chaotic picture about your[br]application,

0:07:14.770,0:07:16.680
but if you, if you were to draw the

0:07:16.680,0:07:19.220
arrows a bit straighter, you would see that[br]the

0:07:19.220,0:07:23.009
dependencies all go in one direction, and[br]that there

0:07:23.009,0:07:25.500
are parts that other parts are based upon,[br]right.

0:07:25.500,0:07:28.330
So this app is composed of parts, and it's

0:07:28.330,0:07:31.360
no longer a ball of mud.

0:07:31.360,0:07:32.690
The fact that you can see that this is

0:07:32.690,0:07:36.539
a complicated domain is an improvement, despite[br]this picture

0:07:36.539,0:07:38.220
looking a bit weird. This picture is from[br]Ben

0:07:38.220,0:07:41.180
Smith's talk about how he architected his[br]Big Rails

0:07:41.180,0:07:43.630
app for success, which he gave at Rocky Mountain

0:07:43.630,0:07:46.630
Ruby last year. And I recommend you watch[br]that

0:07:46.630,0:07:47.550
talk.

0:07:47.550,0:07:51.139
So, if you're like, what? Now? Why would I

0:07:51.139,0:07:53.610
do that? As I just said. Because it helps.

0:07:53.610,0:07:57.160
When applications get big, I believe it's[br]fundamentally important

0:07:57.160,0:07:58.949
to be able to think about the parts of

0:07:58.949,0:08:04.919
the application independently, as much as[br]possible.

0:08:04.919,0:08:08.259
There's a bunch of resources that, you know.[br]I've

0:08:08.259,0:08:10.430
given talks about this. Ben's given talks[br]about this.

0:08:10.430,0:08:13.300
There's blog posts about this. And there's,[br]most importantly,

0:08:13.300,0:08:16.729
that repository that I recommend you look[br]at.

0:08:16.729,0:08:22.509
OK. Back to Big Rails. How big? Does anyone

0:08:22.509,0:08:25.039
here think they're working on a big Rails[br]application?

0:08:25.039,0:08:29.910
OK. That's about a third. So, I've, I think

0:08:29.910,0:08:31.590
I've worked on a couple, but I didn't know

0:08:31.590,0:08:33.818
what to show, because I work for Pivatol Labs

0:08:33.818,0:08:36.198
and we consult and the code is our clients',

0:08:36.198,0:08:38.509
so I can't show it here, so I won't.

0:08:38.509,0:08:41.130
But I went onto this thing called Google and

0:08:41.130,0:08:43.289
I searched for open source Rails apps, and[br]there

0:08:43.289,0:08:45.350
they were. So these claim to be big Rails

0:08:45.350,0:08:45.940
apps.

0:08:45.940,0:08:50.060
How big, I wondered. Well, if you have, currently,

0:08:50.060,0:08:53.660
on your laptop, your company's code, please,[br]execute this

0:08:53.660,0:08:57.279
script and Tweet the result with a #cbra hashtag,

0:08:57.279,0:08:58.930
cause this is just gonna open how many lines

0:08:58.930,0:09:00.630
of code your app has.

0:09:00.630,0:09:03.420
And here's the result for this list of applications.

0:09:03.420,0:09:06.580
So, as you can see, they're not all simi-

0:09:06.580,0:09:09.890
not all the same interpretation of big. So[br]we're

0:09:09.890,0:09:13.910
going from about 400,000 lines of code to[br]still

0:09:13.910,0:09:17.000
in the thousands. If you look at the files,

0:09:17.000,0:09:20.770
same picture. Thousands of files for the biggest[br]apps,

0:09:20.770,0:09:24.660
and still hundreds for the smaller ones.

0:09:24.660,0:09:28.300
I'll accept that as big. And I hope you

0:09:28.300,0:09:32.910
will too. Ever wonder what happens to the[br]complexity

0:09:32.910,0:09:37.110
within a system as it grows? Something like[br]this

0:09:37.110,0:09:40.390
happens. And maybe this is bad as that, and

0:09:40.390,0:09:43.649
you just can't take it anymore.

0:09:43.649,0:09:46.700
When you introduce structure, you fundamentally[br]change the game.

0:09:46.700,0:09:49.899
Oh. I'm supposed to say, first, that this[br]is

0:09:49.899,0:09:52.860
because the number of interactions between[br]the elements inside

0:09:52.860,0:09:56.370
the system grows somewhat like an exponential,[br]in an

0:09:56.370,0:10:00.089
exponential function when you add new, new[br]parts to

0:10:00.089,0:10:01.830
this system.

0:10:01.830,0:10:06.230
So it explodes in complexity. Well, what if[br]you

0:10:06.230,0:10:11.290
were able to split a system and use a

0:10:11.290,0:10:14.600
complex web of service objects, maybe, to[br]connect the

0:10:14.600,0:10:17.010
two. But you have two smaller parts that can

0:10:17.010,0:10:20.200
now, that are now isolated, in some way, what

0:10:20.200,0:10:21.510
happens to the complexity?

0:10:21.510,0:10:26.440
Well, turns out, it's still this, and probably[br]still

0:10:26.440,0:10:33.440
that. But only half as fast. And that's good.

0:10:35.760,0:10:37.430
If you. I wish I could run up to

0:10:37.430,0:10:39.310
the slide right now. But, you see that green

0:10:39.310,0:10:42.160
line at the bottom? It's almost flat. Now[br]imagine

0:10:42.160,0:10:44.680
you could split it again, and again, and again,

0:10:44.680,0:10:46.700
and you always stay in this flatter part,[br]so

0:10:46.700,0:10:50.040
your complexity never really explodes. You[br]get those crazy

0:10:50.040,0:10:52.320
diagrams that I was showing earlier, but your[br]complexity

0:10:52.320,0:10:55.060
doesn't explode quite as badly. You're still[br]writing a

0:10:55.060,0:10:57.200
huge app, and that's something that I'm not[br]gonna

0:10:57.200,0:11:00.010
discuss away. But you might be able to manage

0:11:00.010,0:11:06.480
it better.

0:11:06.480,0:11:11.260
The rich get rich and the poor get. Anyone?

0:11:11.260,0:11:17.420
Children. I. This is from a song from the

0:11:17.420,0:11:20.100
1920s, and while it's funny, the, the first[br]part

0:11:20.100,0:11:25.660
of this alludes to a thing called preferential[br]attachment.

0:11:25.660,0:11:29.200
Preferential attachment is a set of processes[br]where, when,

0:11:29.200,0:11:33.350
where, god. This is so hard to say. Where,

0:11:33.350,0:11:36.630
when you, where the amount of things you have,

0:11:36.630,0:11:38.230
whatever that might be - the thing, or you

0:11:38.230,0:11:42.899
- defines how much you will get.

0:11:42.899,0:11:47.830
So, 2006, Chris Anderson I think, wrote about[br]the

0:11:47.830,0:11:51.420
long tail in, in publishing and books. And[br]he,

0:11:51.420,0:11:53.870
this, this curve became very popular. He called,[br]I

0:11:53.870,0:11:55.600
think he even called the book The Long Tail.

0:11:55.600,0:11:57.160
And he was concentrated on the right side[br]of

0:11:57.160,0:11:58.930
this tail.

0:11:58.930,0:12:00.790
So you have many, many small things that don't

0:12:00.790,0:12:04.540
get any traction. But, you know, Netflix allows[br]us

0:12:04.540,0:12:06.870
to view all those movies that no one wants,

0:12:06.870,0:12:08.930
that no one else wants to see. But there's

0:12:08.930,0:12:11.370
just a very few movies at the front that

0:12:11.370,0:12:14.410
everyone wants to see. The Blockbusters. And,[br]while the

0:12:14.410,0:12:17.440
long tail is very interesting, it's the ones[br]in

0:12:17.440,0:12:22.760
the front that harvest most of the benefits,[br]well

0:12:22.760,0:12:24.700
I guess in the case of movies.

0:12:24.700,0:12:27.519
So, the green to the yellow relationship is[br]about,

0:12:27.519,0:12:32.829
80% of the, the tail versus 20% is inverse

0:12:32.829,0:12:34.769
in the relationship to how much they have.[br]So

0:12:34.769,0:12:38.500
it's 20/80, 80/20.

0:12:38.500,0:12:43.209
And, preferential attachment, as I said, happens[br]when you

0:12:43.209,0:12:45.910
already have a lot, so you get more. And

0:12:45.910,0:12:48.350
I have seen so many code bases that I

0:12:48.350,0:12:50.140
have the feeling there's something like that[br]going on

0:12:50.140,0:12:52.750
in code. So I looked at these code bases,

0:12:52.750,0:12:56.180
and again, I typed in this bash script. First

0:12:56.180,0:13:00.670
try of course.

0:13:00.670,0:13:02.230
And I urge you to do the same. Now,

0:13:02.230,0:13:04.029
you will not be able to Tweet this, because

0:13:04.029,0:13:06.579
this lists all the files, all the Ruby files,

0:13:06.579,0:13:10.790
in your app by size. And when you do

0:13:10.790,0:13:14.730
that, that looks about like this for these[br]applications.

0:13:14.730,0:13:19.860
Now, does that not look similar?

0:13:19.860,0:13:22.040
There's another way of saying this, you know,[br]the

0:13:22.040,0:13:24.700
rich get richer. There's a German proverb.[br]I'm from

0:13:24.700,0:13:29.860
Germany. Hi. That proverb is that Der Teufel[br]scheißt

0:13:29.860,0:13:32.310
immer auf den größten Haufen. Or, the Devil[br]always

0:13:32.310,0:13:35.560
shits on the biggest pile.

0:13:35.560,0:13:39.089
And I think something like that is going on

0:13:39.089,0:13:46.089
here. And which file might that be? So who

0:13:46.440,0:13:48.010
thinks that might be a good guess as to

0:13:48.010,0:13:51.700
which file that is? And, lower your, don't[br]lower

0:13:51.700,0:13:54.290
your hand, because it's probably actually[br]yours. I said

0:13:54.290,0:13:55.839
this two days ago and then the guy next

0:13:55.839,0:13:57.870
to me shows me his Ruby, his user file

0:13:57.870,0:14:00.160
and it's literally this long and I'm like,[br]oh

0:14:00.160,0:14:02.170
god, don't do this to me.

0:14:02.170,0:14:04.550
But everyone else knows this problem. The[br]user is

0:14:04.550,0:14:07.420
around earlier. He is important to the system,[br]so

0:14:07.420,0:14:10.950
he attaches, he attracts functionality. But[br]to the point

0:14:10.950,0:14:13.100
where I would say that is a pile of

0:14:13.100,0:14:15.310
shit, man.

0:14:15.310,0:14:19.570
And I thought that would be something else[br]I

0:14:19.570,0:14:22.420
could find. So I looked at associations. Associations[br]are

0:14:22.420,0:14:28.510
another way of expressing dependencies. In[br]fact, so when,

0:14:28.510,0:14:30.250
I want to talk about has_manys, because I[br]find

0:14:30.250,0:14:31.730
them particularly interesting.

0:14:31.730,0:14:38.730
So, you have something that has_many. User,[br]maybe. So

0:14:40.360,0:14:42.130
since I assumed it was user, I want you

0:14:42.130,0:14:45.029
to read this code and, kind of reflecting[br]on

0:14:45.029,0:14:48.950
yesterday's keynote, I find this file very[br]readable. The

0:14:48.950,0:14:55.810
user has_many :cars, :flowers, :trees, :houses,[br]:moods, :checks. Hopefully.

0:14:55.810,0:14:57.360
I think we can agree that it's very readable.

0:14:57.360,0:15:00.040
It makes a lot of sense and we can,

0:15:00.040,0:15:01.850
we can empathize with this class.

0:15:01.850,0:15:04.389
Have you ever wondered what's funny about[br]this when

0:15:04.389,0:15:05.769
you look at the table and there is no

0:15:05.769,0:15:09.019
mention of :cars, :flowers, :trees, :houses,[br]:moods, and :checks

0:15:09.019,0:15:13.000
in that table at all? You don't find it

0:15:13.000,0:15:14.779
funny to write that user class, but would[br]you

0:15:14.779,0:15:16.570
find it funny if I wrote a color class

0:15:16.570,0:15:18.750
like that?

0:15:18.750,0:15:20.060
Why do we not write that color class? Why

0:15:20.060,0:15:22.329
do we always write that user class? I'll tell

0:15:22.329,0:15:24.950
you why. Because users are important to us,[br]and

0:15:24.950,0:15:26.970
we always write them first, and then we attach

0:15:26.970,0:15:30.290
all sorts of stuff to them that doesn't belong

0:15:30.290,0:15:34.579
to them. This table clearly says what belongs[br]to,

0:15:34.579,0:15:37.399
sorry, a user. Namely, in this case, a name

0:15:37.399,0:15:39.880
and a home.

0:15:39.880,0:15:42.500
So I've kind of gotten the opinion that we

0:15:42.500,0:15:47.329
shouldn't be using has_many too much. You[br]can check,

0:15:47.329,0:15:49.220
and I would be very interested in the results,

0:15:49.220,0:15:52.560
again, so there's another gist. Please, try[br]it out

0:15:52.560,0:15:55.880
if you can.

0:15:55.880,0:16:02.639
Whoops. And guess which file is on the left?

0:16:02.639,0:16:07.170
Yeah. He shat on that again. The, it's the

0:16:07.170,0:16:10.740
user. And it's ot always actually the user.[br]I

0:16:10.740,0:16:13.470
shouldn't be saying that. I should say, it's[br]probably

0:16:13.470,0:16:15.810
always, to some extent, the user. But what's[br]your

0:16:15.810,0:16:21.079
domain? If your domain is, I don't know, money?

0:16:21.079,0:16:25.730
Houses? Properties? I don't know. It's probably[br]that file.

0:16:25.730,0:16:28.010
Whatever I just said or whatever is in your

0:16:28.010,0:16:30.300
name. That's that file.

0:16:30.300,0:16:32.160
And that attracts all the functionality. But[br]the user

0:16:32.160,0:16:36.040
is typically not far behind. And, because[br]I want

0:16:36.040,0:16:39.070
to. Yeah, I just. I think this is very

0:16:39.070,0:16:41.079
enlightening. So, I just opened to the project[br]I

0:16:41.079,0:16:43.029
just showed you. I opened all the user files,

0:16:43.029,0:16:46.220
and I'm just gonna scroll through them, K.

0:16:46.220,0:16:53.220
This is not very intimidating. Eh. This is[br]a

0:16:54.579,0:17:01.510
bit better. Ooh. I'm trying to be non-discriminatory[br]here

0:17:01.510,0:17:08.510
and go the same speed

0:17:15.130,0:17:15.599
all the time. Ooh.

0:17:15.599,0:17:22.378
Oh. Oh you wait. Now, I have to speed

0:17:22.378,0:17:26.039
up.

0:17:26.039,0:17:29.509
And it's, you, this one file looks particularly[br]bad

0:17:29.509,0:17:32.789
maybe, but it's probably the case that these[br]user

0:17:32.789,0:17:36.119
files are big to the extent that their applications

0:17:36.119,0:17:40.999
are big. These files just tend to do that.

0:17:40.999,0:17:44.359
These models just attract that.

0:17:44.359,0:17:51.359
So, what does that mean?

0:17:51.549,0:17:53.850
To recap. We have things that grow, and when

0:17:53.850,0:17:56.619
they grow, they get exponentially more, they[br]get more

0:17:56.619,0:17:59.879
complex. They get exponentially more complex.[br]And then in

0:17:59.879,0:18:01.399
those things, there are certain things that[br]are, you

0:18:01.399,0:18:05.700
know, that are the top of the badness scale.

0:18:05.700,0:18:09.879
And, and those things that also depend on[br]everything

0:18:09.879,0:18:15.859
else in the system.

0:18:15.859,0:18:22.100
I don't know. I think we can do better.

0:18:22.100,0:18:24.479
We can try to reduce the size of the

0:18:24.479,0:18:28.210
parts. Maybe flatten out that curve and reduce[br]the

0:18:28.210,0:18:34.859
number of dependencies in between those parts.[br]Something that

0:18:34.859,0:18:38.429
has helped me in doing that has been Solid,

0:18:38.429,0:18:45.159
the five ideas, principles. Signs, truths.[br]I don't know.

0:18:45.159,0:18:45.840
No.

0:18:45.840,0:18:48.950
And especially for this problem, the Single[br]Responsibility Principle.

0:18:48.950,0:18:52.859
I think if you just apply that as a

0:18:52.859,0:18:56.409
rule of thumb, as a, as a starter for

0:18:56.409,0:19:00.739
a conversation, it will, over time, on average,[br]improve

0:19:00.739,0:19:04.119
the quality of your code. But I think there's

0:19:04.119,0:19:06.859
something funny going on when we discuss single[br]responsibility.

0:19:06.859,0:19:10.999
Oh. Yes. What's single responsibility for?[br]To find things

0:19:10.999,0:19:17.999
that don't belong. Clearly it's the cat with[br]the

0:19:18.239,0:19:21.179
green eyes.

0:19:21.179,0:19:27.889
Where, where to apply Single Responsibility[br]Principle? If you've

0:19:27.889,0:19:29.489
been to any of the refactoring talks - I

0:19:29.489,0:19:31.190
didn't have a chance this time - but I'm

0:19:31.190,0:19:35.509
pretty sure you refactored some methods. We[br]look at

0:19:35.509,0:19:37.679
methods, we want them to be better. We make

0:19:37.679,0:19:40.210
them smaller. We find what they are about[br]and

0:19:40.210,0:19:44.249
we give them one responsibility. Class. We[br]probably did

0:19:44.249,0:19:46.639
the same, right. We look at a class, we're

0:19:46.639,0:19:48.909
like, oh, this does a little bit much. But

0:19:48.909,0:19:51.159
then we're like, OK, ActiveRecord is cool.[br]I'm like,

0:19:51.159,0:19:53.739
allowing it to save itself and soforth. So[br]maybe

0:19:53.739,0:19:55.519
that's OK. But it still kind of should be

0:19:55.519,0:19:57.799
about this one thing.

0:19:57.799,0:19:59.629
Except for user.

0:19:59.629,0:20:01.309
So we're, on these two levels, I think we're

0:20:01.309,0:20:06.600
really good. Module, and by that I mean a

0:20:06.600,0:20:08.929
namespace. If you use module as mixin, I get

0:20:08.929,0:20:10.769
like that I don't want to talk about that,

0:20:10.769,0:20:13.970
and that's why I continue to say namespace.[br]Why

0:20:13.970,0:20:15.720
do I think this about mixins? I should maybe

0:20:15.720,0:20:18.559
explain in one sentence. If you mixin stuff[br]into

0:20:18.559,0:20:21.159
your objects, the only thing you're doing[br]is hiding

0:20:21.159,0:20:23.190
away the fact that you have more complexity[br]than

0:20:23.190,0:20:24.869
you can bear to look at on one screen,

0:20:24.869,0:20:26.129
so you put it on another so you don't

0:20:26.129,0:20:29.179
have to see it. I would not do that.

0:20:29.179,0:20:31.119
I mean, I have done it. I have written

0:20:31.119,0:20:32.929
apps like that, and now I hate myself for

0:20:32.929,0:20:35.210
it. But, I want to not do it anymore.

0:20:35.210,0:20:37.749
So I think we shouldn't. So I use modules

0:20:37.749,0:20:42.289
a lot, but as namespaces, to group things[br]together.

0:20:42.289,0:20:44.489
I think you should do that. You know, look

0:20:44.489,0:20:47.450
at modules or namespaces and, and analyze[br]whether they

0:20:47.450,0:20:51.970
have one responsibility. But have you done[br]that lately?

0:20:51.970,0:20:54.099
It's harder, right? It doesn't mean a lot.[br]You

0:20:54.099,0:20:55.919
can, it doesn't, you know, no one cares if

0:20:55.919,0:20:57.909
you go across these things. You just, you[br]can

0:20:57.909,0:21:01.789
load any class anyways.

0:21:01.789,0:21:05.909
So I think we could get better at that.

0:21:05.909,0:21:09.149
Let me jump a level. Application. And I think,

0:21:09.149,0:21:11.169
again, you should. And the fact that we have

0:21:11.169,0:21:13.220
this Big Rails track, and the fact that I'm

0:21:13.220,0:21:15.499
talking about components, but pretty much[br]everyone else is

0:21:15.499,0:21:19.200
talking about SOAs, shows that we are talking[br]about,

0:21:19.200,0:21:21.879
what do applications mean? What does it mean[br]for

0:21:21.879,0:21:25.419
an application to do something? Or, yeah.[br]To have

0:21:25.419,0:21:27.849
a function in, in something bigger.

0:21:27.849,0:21:30.519
So, I encourage you to go to all those

0:21:30.519,0:21:34.039
tracks and make your applications better on[br]that level

0:21:34.039,0:21:35.889
as well. But I just talked to you about

0:21:35.889,0:21:37.919
the component, and I think that sits in the

0:21:37.919,0:21:40.749
middle. And in, in a certain way, the component

0:21:40.749,0:21:44.039
is, is an improved namespace. Because of that[br]thing

0:21:44.039,0:21:46.299
I said earlier, namely, I can run tests and

0:21:46.299,0:21:48.739
I can prove that those tests run without ever

0:21:48.739,0:21:52.399
touching any code that is not in that namespace.

0:21:52.399,0:21:56.799
So, now you can. And I urge you to

0:21:56.799,0:22:02.019
do it. Because the cat picture, the discussion[br]of

0:22:02.019,0:22:04.950
what doesn't belong. If you are in your Big

0:22:04.950,0:22:07.879
Rails app and you're writ- painting all those[br]classes

0:22:07.879,0:22:10.789
as boxes and the dependencies they have as[br]arrows,

0:22:10.789,0:22:14.700
first off, it's really easy to screw that[br]up.

0:22:14.700,0:22:16.389
Because there's no, you can't really prove[br]that you're

0:22:16.389,0:22:18.309
not using another class. At least not in run

0:22:18.309,0:22:21.519
time with big applications.

0:22:21.519,0:22:25.789
So components give you an arbitrarily scalable[br]and provably

0:22:25.789,0:22:29.509
independent way to talk about parts of your[br]application.

0:22:29.509,0:22:32.690
So I use it. Now, I said you should

0:22:32.690,0:22:36.099
look at SOAs, and when, if you're considering[br]refactoring

0:22:36.099,0:22:38.389
your application to improve its quality and[br]to make

0:22:38.389,0:22:40.470
it more maintainable, I think there are a[br]couple

0:22:40.470,0:22:45.080
things you should know about or consider when[br]deciding

0:22:45.080,0:22:47.729
between the two. Right, you're kind of feeling[br]pains

0:22:47.729,0:22:50.559
right now, or you're not even yet feeling[br]pains.

0:22:50.559,0:22:53.249
Great place to be in, too. But there's a

0:22:53.249,0:22:55.529
couple things that are really cool about just[br]splitting

0:22:55.529,0:22:57.379
up your application into these components.

0:22:57.379,0:23:01.489
And that is, you stick with one repo. You

0:23:01.489,0:23:04.399
have one test suite. But you can split it

0:23:04.399,0:23:07.899
up. You have, still, one deployment. You can,[br]however,

0:23:07.899,0:23:11.279
use, like, a load balancer to push traffic[br]to

0:23:11.279,0:23:13.849
certain parts of the application to make the[br]one

0:23:13.849,0:23:16.340
faster, the other, you know, has less of a

0:23:16.340,0:23:19.590
demand or, you know, you can go crazy. But

0:23:19.590,0:23:21.929
you don't get these additional versioning[br]constraints of where,

0:23:21.929,0:23:23.799
like, if you, if you have this app and

0:23:23.799,0:23:25.519
it needs that app and that app doesn't have

0:23:25.519,0:23:27.619
the API yet, then I need to add this

0:23:27.619,0:23:30.450
stuff here at the same time, and I need

0:23:30.450,0:23:31.690
to deploy them at the same time. And if

0:23:31.690,0:23:33.799
I didn't do that then everything goes down.

0:23:33.799,0:23:35.210
You don't have that because you're still talking[br]about

0:23:35.210,0:23:40.099
one application. And also, one other thing[br]that's very

0:23:40.099,0:23:41.830
interesting to me, at least, is that, I never

0:23:41.830,0:23:45.099
know which parts I need. On this crazy graph

0:23:45.099,0:23:47.379
project that I just showed you, it went on

0:23:47.379,0:23:50.450
for about six months. And, and there were[br]three

0:23:50.450,0:23:53.539
pairs on that. And at some point. Four pairs.

0:23:53.539,0:23:55.090
And at some point, there was always one pair

0:23:55.090,0:23:57.940
refactoring something. And they created, I[br]think there were

0:23:57.940,0:24:00.179
a couple weeks where they created basically[br]a component

0:24:00.179,0:24:02.519
a day. But they also removed a component.

0:24:02.519,0:24:03.929
Try to do that with a SOA. Try to

0:24:03.929,0:24:06.349
move that stuff around quickly. You can do[br]a

0:24:06.349,0:24:09.669
SOA, I think, when you, when you're sure you

0:24:09.669,0:24:12.409
need pieces. But when you're playing around[br]with parts

0:24:12.409,0:24:14.529
of your application and you, you want to just

0:24:14.529,0:24:16.009
structure it, and you want to still be able

0:24:16.009,0:24:18.840
to move a part from one to the other,

0:24:18.840,0:24:21.369
SOA is a bit heavy.

0:24:21.369,0:24:26.690
Within a SOA, you can still think about your

0:24:26.690,0:24:29.929
applications as being built of these components.[br]There's, you

0:24:29.929,0:24:34.129
know, nothing is exclusive there. Just do[br]both.

0:24:34.129,0:24:41.129
So, I've now spent about, most of my time

0:24:41.499,0:24:44.299
telling you why you should refactor towards[br]this. But

0:24:44.299,0:24:45.840
I haven't done what I said in my title

0:24:45.840,0:24:47.309
I would do, namely tell you how to refactor

0:24:47.309,0:24:51.469
towards it. I'm sorry about that. But I gave

0:24:51.469,0:24:55.429
you this idea, right. And you all believed[br]me.

0:24:55.429,0:24:58.659
This is easy.

0:24:58.659,0:25:01.769
Today's day of the, International Day of the[br]Book.

0:25:01.769,0:25:06.869
I looked in these books. Refactoring to Patterns.[br]Refactoring.

0:25:06.869,0:25:10.849
Awesome books. But they don't talk about this[br]either.

0:25:10.849,0:25:12.379
And I think the reason is that this stuff

0:25:12.379,0:25:15.429
is not very important in other languages.[br]In Java,

0:25:15.429,0:25:17.759
you do packages every day and you have implicit

0:25:17.759,0:25:20.070
imports. In Go, you do that. Other languages[br]do

0:25:20.070,0:25:22.859
not have this problem, because other languages[br]do not

0:25:22.859,0:25:25.580
allow for this kind of freedom and openness[br]that

0:25:25.580,0:25:31.210
we've gotten because we do Ruby.

0:25:31.210,0:25:35.259
So, I call these two refactorings, teasing[br]out an

0:25:35.259,0:25:39.090
app component and extracting a functional[br]component. And the

0:25:39.090,0:25:40.259
reason I don't want to spend much time on

0:25:40.259,0:25:42.049
it is because it's actually very simple to[br]do

0:25:42.049,0:25:46.429
it. The devil's just in the details.

0:25:46.429,0:25:48.969
And there is no book about this. But I'm

0:25:48.969,0:25:52.219
trying to write one. And if you, I don't

0:25:52.219,0:25:54.469
know, show, if you're interested in this,[br]send a

0:25:54.469,0:25:56.440
message there and I might be more inclined[br]to

0:25:56.440,0:25:58.059
actually sit down.

0:25:58.059,0:26:01.789
All right. But to the first one - teasing

0:26:01.789,0:26:08.729
out an app component. So, step zero. Got tests?

0:26:08.729,0:26:11.099
If, if you don't have tests, I wouldn't move

0:26:11.099,0:26:13.590
as much code as we're about to move. I

0:26:13.590,0:26:16.710
mean, virtually. But, you should add some[br]tests around

0:26:16.710,0:26:19.190
the stuff that you want to extract.

0:26:19.190,0:26:21.479
And then you, for the app component. So that's

0:26:21.479,0:26:24.539
this teaser thing we looked at. So we're,[br]a

0:26:24.539,0:26:27.119
component of your application, where you almost[br]thing, this

0:26:27.119,0:26:30.070
could be its own application, maybe. So you[br]look

0:26:30.070,0:26:35.139
for this vertical. Views, controllers, models.[br]You find that

0:26:35.139,0:26:38.149
vertical that makes sense on its own and you

0:26:38.149,0:26:39.479
namespace it.

0:26:39.479,0:26:42.059
A funny thing happens when you just namespace[br]it.

0:26:42.059,0:26:44.609
Suddenly, the things in there. Well, first[br]off, what's

0:26:44.609,0:26:46.369
awful about it is you have to rename all

0:26:46.369,0:26:49.479
your tables. OK. But that aside, you suddenly[br]start

0:26:49.479,0:26:53.499
seeing what's actually in your vertical and[br]what's not.

0:26:53.499,0:26:55.379
You could actually put colon colon in front[br]of

0:26:55.379,0:26:56.849
everything else that's not, and you would[br]get a

0:26:56.849,0:26:59.719
really easy view parsing or, or way of seeing

0:26:59.719,0:27:02.269
in your files what's not in your vertical.

0:27:02.269,0:27:06.200
Those are your external dependencies to, an[br]external to

0:27:06.200,0:27:08.259
the app. So when you do this at first,

0:27:08.259,0:27:09.929
you're apps gonna be unhappy. So please make[br]your

0:27:09.929,0:27:12.889
tests green. And to do that, you will need

0:27:12.889,0:27:15.340
to find other parts that you didn't know you

0:27:15.340,0:27:19.899
depended upon, and you move them into what's[br]gonna

0:27:19.899,0:27:25.379
be the component. Into this vertical.

0:27:25.379,0:27:28.599
And when you got this vertical green, you[br]extract

0:27:28.599,0:27:34.139
it into an engine. Extracting into an engine[br]is

0:27:34.139,0:27:37.519
rails plugin new component dash dash full,[br]dash dash

0:27:37.519,0:27:39.659
mountable. In most cases.

0:27:39.659,0:27:41.239
It doesn't matter.

0:27:41.239,0:27:48.239
And step three, profit. Now, most of the time,

0:27:49.460,0:27:52.200
this vertical will not be as easy to find

0:27:52.200,0:27:55.330
as I just pretended it would be. And that's

0:27:55.330,0:28:01.519
because of, you guessed it, user. Right? The[br]user

0:28:01.519,0:28:05.419
is probably somewhere, and other things as[br]well, are

0:28:05.419,0:28:08.589
used somewhere, are needed somewhere, but[br]don't really belong

0:28:08.589,0:28:09.940
to this verticle.

0:28:09.940,0:28:13.119
And what I've found is when you refactor towards

0:28:13.119,0:28:14.979
this, if you abstract. If you, if you're coming

0:28:14.979,0:28:17.649
from one big piece, if you want to abstract

0:28:17.649,0:28:20.570
one thing, you're gonna abstract two. The[br]one thing

0:28:20.570,0:28:24.200
you wanted and the common stuff. But please[br]do

0:28:24.200,0:28:26.339
not call that stuff common, because that's[br]just gonna

0:28:26.339,0:28:28.219
be another pile of shit that you can attach

0:28:28.219,0:28:30.169
stuff to. Call it something else. Call it[br]user

0:28:30.169,0:28:34.210
if you have to.

0:28:34.210,0:28:36.049
So this special case is not a special case.

0:28:36.049,0:28:38.460
I think this is, of course there is lots

0:28:38.460,0:28:40.529
of meat in how to do this and where

0:28:40.529,0:28:43.029
you get hung up on certain gem dependencies[br]and

0:28:43.029,0:28:45.210
engines not doing it exactly the way that[br]Rails

0:28:45.210,0:28:48.019
does it and blah, blah, blah. But essentially[br]this

0:28:48.019,0:28:52.139
is it.

0:28:52.139,0:28:55.109
The other side to this is, is, something even

0:28:55.109,0:29:01.839
more interesting. It's abstracting functional[br]components. Now your app

0:29:01.839,0:29:05.469
is round, suddenly. I don't know why. Again,[br]got

0:29:05.469,0:29:08.759
tests? Let's start there.

0:29:08.759,0:29:11.200
So you find a functional component, and yes,[br]I

0:29:11.200,0:29:14.029
made that much smaller than that verticle.[br]Because typically

0:29:14.029,0:29:16.239
you find a piece that is really, really separate.

0:29:16.239,0:29:20.369
A very special functionality. Maybe even something[br]mathematically a

0:29:20.369,0:29:23.839
function. In and out. Some processing. You[br]find that

0:29:23.839,0:29:27.909
and you create a gem.

0:29:27.909,0:29:30.109
And you move the tests in there as well,

0:29:30.109,0:29:31.960
and your gem is, when you run those tests,

0:29:31.960,0:29:34.440
the gem's gonna be unhappy, because of all[br]those

0:29:34.440,0:29:38.219
dependencies that you didn't see coming. So,[br]you take

0:29:38.219,0:29:42.419
all those and you move them in. And you

0:29:42.419,0:29:44.539
make your tests happy.

0:29:44.539,0:29:47.489
Well, and now you require that gem in the

0:29:47.489,0:29:50.349
main app, and your main app is totally unhappy.

0:29:50.349,0:29:51.719
Because it doesn't know how to talk to that

0:29:51.719,0:29:57.950
gem anymore. And if there were any ActiveRecord[br]dependencies,

0:29:57.950,0:30:00.690
or other dependencies on, on stuff that came[br]from

0:30:00.690,0:30:03.659
Rails, I would urge you to try and keep

0:30:03.659,0:30:06.139
that out of, of these kinds of gems, because

0:30:06.139,0:30:08.200
it just adds such a huge surface area and

0:30:08.200,0:30:11.339
stuff that could happen, could be done and[br]could

0:30:11.339,0:30:12.139
go wrong.

0:30:12.139,0:30:15.299
I, I would urge you to find these connections,

0:30:15.299,0:30:20.629
these, the shims, maybe. Call them ports and[br]adapters.

0:30:20.629,0:30:23.419
And move them in the app closer to the

0:30:23.419,0:30:25.799
surface area of this gem, so that you can

0:30:25.799,0:30:28.309
do the processing of whatever that is there,[br]and

0:30:28.309,0:30:30.969
that talks to the gem and makes it work

0:30:30.969,0:30:33.580
again. So now the app is happy. And you

0:30:33.580,0:30:37.080
got this weird kind of lump on the side.

0:30:37.080,0:30:38.349
But that's a good thing.

0:30:38.349,0:30:40.739
In, in one of the books, actually I think

0:30:40.739,0:30:42.909
it's Eric Evans, it's called Bound to Context.[br]Now

0:30:42.909,0:30:45.440
you have a thing that is only relevant for

0:30:45.440,0:30:49.639
a context. But that's how you take from that,

0:30:49.639,0:30:52.450
from that curve that has the huge left side.

0:30:52.450,0:30:54.149
That's how you chip away at the top and

0:30:54.149,0:30:56.529
put it somewhere at the end. Because this[br]thing,

0:30:56.529,0:30:59.149
in a bound context, it only has one meaning.

0:30:59.149,0:31:00.599
And that's why it's gonna be smaller.

0:31:00.599,0:31:07.599
So, step three. Profit.

0:31:07.820,0:31:14.299
I hope that made sense. And I know. It

0:31:14.299,0:31:16.649
doesn't work that easily. It's not five steps.[br]It

0:31:16.649,0:31:19.609
can take weeks. The first time we tried this,

0:31:19.609,0:31:22.089
it took three attempts and four weeks in one

0:31:22.089,0:31:24.759
app that we wanted to split into two. And

0:31:24.759,0:31:26.889
we ended up with three, and now I think

0:31:26.889,0:31:30.169
it's about ten. But, trust me, you can get

0:31:30.169,0:31:30.820
there.

0:31:30.820,0:31:35.570
And I kind of felt challenged by Far- by

0:31:35.570,0:31:40.489
Farrah's keynote yesterday. So, I will help.[br]If you

0:31:40.489,0:31:42.330
want to attempt this, and you don't find the

0:31:42.330,0:31:44.279
resources and any of those links that I gave

0:31:44.279,0:31:47.089
you, Tweet at me, talk to me, send me

0:31:47.089,0:31:49.669
an email. I'll respond to all of them. Maybe

0:31:49.669,0:31:53.389
publicly so it helps others, too. I just think

0:31:53.389,0:31:55.769
this is a great way to structure applications[br]and

0:31:55.769,0:31:58.269
to make them better, and that's why I just

0:31:58.269,0:32:00.769
will help.

0:32:00.769,0:32:04.330
So don't let these guys ruin your day. Make

0:32:04.330,0:32:06.609
friends with the #cbra, and thanks for your[br]attention.