WEBVTT 1 00:00:00.662 --> 00:00:04.307 Hello and welcome to the Haskell Weekly Podcast, a podcast about Haskell. 2 00:00:04.307 --> 00:00:07.961 Haskell is a functional programming language that we use here at ITProTV, 3 00:00:07.961 --> 00:00:10.459 and we're just super excited about this episode. 4 00:00:10.459 --> 00:00:14.295 I'm your host Cameron, and with me today we have Taylor Fausak, 5 00:00:14.295 --> 00:00:16.358 our lead engineer here at ITProTV. 6 00:00:16.358 --> 00:00:17.360 How are you doing Taylor? 7 00:00:17.360 --> 00:00:18.257 >> I'm doing well, thanks. 8 00:00:18.257 --> 00:00:19.437 How are you Cam? 9 00:00:19.437 --> 00:00:20.393 >> Can't complain. 10 00:00:20.393 --> 00:00:21.927 It's going well, I would say. 11 00:00:21.927 --> 00:00:22.619 >> Full of fries? 12 00:00:22.619 --> 00:00:24.751 >> I just ate so many fries, ugh. 13 00:00:24.751 --> 00:00:26.815 It was so good. 14 00:00:26.815 --> 00:00:31.975 But today, I just wanna kind of talk about an architecture 15 00:00:31.975 --> 00:00:35.674 that is a good functional architecture. 16 00:00:35.674 --> 00:00:37.310 It is Ports-And-Adapters. 17 00:00:37.310 --> 00:00:39.668 And what are Ports-And-Adapters? 18 00:00:39.668 --> 00:00:45.193 So I'll kind of talk more about Ports because that's really what I wanna do. 19 00:00:45.193 --> 00:00:50.233 So Ports are kind of responsible for interacting with the outside world and 20 00:00:50.233 --> 00:00:55.353 that could be a web app, it could be the database layer, a way to communicate 21 00:00:55.353 --> 00:00:59.315 with the outside world [CROSSTALK] >> It's the thing that's making a call to 22 00:00:59.315 --> 00:01:02.019 the database or to some external service that you're using or 23 00:01:02.019 --> 00:01:03.331 even writing to the console. 24 00:01:03.331 --> 00:01:04.263 >> Exactly, yeah. 25 00:01:04.263 --> 00:01:06.592 And adapters are kind of everything inside, 26 00:01:06.592 --> 00:01:08.745 kind of handling all the business logic. 27 00:01:08.745 --> 00:01:09.266 >> Okay. 28 00:01:09.266 --> 00:01:11.965 >> Being able to kind of going hey, this is what this is. 29 00:01:11.965 --> 00:01:16.359 >> So adapters take their input from the the ports and then do some other stuff. 30 00:01:16.359 --> 00:01:19.625 And then what talk to another port talk to another adapter. 31 00:01:19.625 --> 00:01:21.694 >> Just depends on what the flow you know of your- 32 00:01:21.694 --> 00:01:22.572 >> So it could be either one. 33 00:01:22.572 --> 00:01:26.372 >> Right if you have an internal structure that needs to talk to another internal 34 00:01:26.372 --> 00:01:28.249 structure you'd be using adapters. 35 00:01:28.249 --> 00:01:31.559 But if you needed to call out to the database, you'd be using a port. 36 00:01:31.559 --> 00:01:32.797 >> Okay. >> Kind of, 37 00:01:32.797 --> 00:01:38.300 it kind of keeps things together, not so much, not so jumbled. 38 00:01:38.300 --> 00:01:43.470 It kind of keeps it clear, the boundaries between pure logic and 39 00:01:43.470 --> 00:01:46.067 input-output logic, right? 40 00:01:46.067 --> 00:01:47.390 >> Right, and so you mean, 41 00:01:47.390 --> 00:01:51.058 the pure logic is what a lot of people sometimes call business logic. 42 00:01:51.058 --> 00:01:54.342 The interesting stuff that your business is doing, or your application is doing. 43 00:01:54.342 --> 00:01:56.750 >> Right. >> And the ports handle the, maybe not so 44 00:01:56.750 --> 00:02:01.213 interesting stuff have actually performing an HTTP call, or something like that. 45 00:02:01.213 --> 00:02:01.864 >> Exactly. 46 00:02:01.864 --> 00:02:05.447 >> Okay, so now that we have kinda the vocabulary of this architecture, 47 00:02:05.447 --> 00:02:09.023 could you explain in more detail what the architecture is? 48 00:02:09.023 --> 00:02:13.935 >> Kind of the Ports-And-Adapters architecture allows 49 00:02:13.935 --> 00:02:16.447 us some niceties around it. 50 00:02:16.447 --> 00:02:20.794 So it kind of allows us to separate that input-output stream from 51 00:02:20.794 --> 00:02:22.858 the internal business logic. 52 00:02:22.858 --> 00:02:27.353 And it keeps our the ability to test internal business logic easier, 53 00:02:27.353 --> 00:02:32.314 cuz you're not always interacting with some input-output that could mess 54 00:02:32.314 --> 00:02:33.491 with the result. 55 00:02:33.491 --> 00:02:37.400 >> Yeah, it's a lot easier to test a pure function because if you have to write 56 00:02:37.400 --> 00:02:40.270 a test that, let's say interacts with the database. 57 00:02:40.270 --> 00:02:43.609 Then that means to run your test suite, you have to have your database up and 58 00:02:43.609 --> 00:02:45.623 running and your schema has to be up to date and 59 00:02:45.623 --> 00:02:47.852 you may have to have fake data populated in there. 60 00:02:47.852 --> 00:02:49.438 But if you're testing pure code, 61 00:02:49.438 --> 00:02:53.007 you just have to have a bunch of values in memory which are a lot easier to make. 62 00:02:53.007 --> 00:02:56.526 Also, you can test things in parallel if they don't have to talk to a database. 63 00:02:56.526 --> 00:03:00.778 Cuz maybe you have two tests for the same piece of business logic in your code and 64 00:03:00.778 --> 00:03:03.756 they would both be writing to the same database table. 65 00:03:03.756 --> 00:03:07.160 You're gonna have a bad time unless you really do a good job of separating those 66 00:03:07.160 --> 00:03:07.735 things out. 67 00:03:07.735 --> 00:03:11.664 Whereas in pure code, you can just run two things at the same time, no problem. 68 00:03:11.664 --> 00:03:14.696 >> Right, and at some point, some level, you want to test the ports. 69 00:03:14.696 --> 00:03:17.755 You want to make sure that given a set of input, 70 00:03:17.755 --> 00:03:20.819 you're getting a reliable source of output. 71 00:03:20.819 --> 00:03:24.944 Because that's what you expect from an API. 72 00:03:24.944 --> 00:03:28.064 You're saying the same thing to an API you would expect, 73 00:03:28.064 --> 00:03:30.085 it's returning to the same values. 74 00:03:30.085 --> 00:03:33.166 >> At some point you want that integration test to prove that your API 75 00:03:33.166 --> 00:03:34.014 actually works. 76 00:03:34.014 --> 00:03:36.160 >> Right, but it does- >> But the unit tests on the inside, 77 00:03:36.160 --> 00:03:37.630 those are still really valuable. 78 00:03:37.630 --> 00:03:41.117 >> Right and the integrations are still a little more boilerplate kinda set up 79 00:03:41.117 --> 00:03:41.731 generally. 80 00:03:41.731 --> 00:03:45.478 And they're a little more, I won't say in bases, but 81 00:03:45.478 --> 00:03:48.831 they're definitely a little more meaty per se. 82 00:03:48.831 --> 00:03:51.676 So it's nice to be able to have that separation and 83 00:03:51.676 --> 00:03:56.352 have guarantee that the pure functions, they're gonna do what they need to do and 84 00:03:56.352 --> 00:03:59.564 those individual units are gonna perform as intended. 85 00:03:59.564 --> 00:04:02.539 But we also can create this test for the ports as well. 86 00:04:02.539 --> 00:04:06.371 >> Yeah. >> To allow that kinda testing. 87 00:04:06.371 --> 00:04:10.290 >> So Cam, we're talking about this architecture today because of this blog 88 00:04:10.290 --> 00:04:13.588 post that Mark Seaman wrote called Functional Architecture is 89 00:04:13.588 --> 00:04:14.972 Ports-And-Adapters. 90 00:04:14.972 --> 00:04:17.387 So pretty straightforward. 91 00:04:17.387 --> 00:04:22.545 But he makes a point in there that you can do this architecture in other languages. 92 00:04:22.545 --> 00:04:25.166 He uses F+ as an example. 93 00:04:25.166 --> 00:04:28.856 But in those languages, it can be really hard to have the discipline to force 94 00:04:28.856 --> 00:04:31.176 yourself into this architecture constantly. 95 00:04:31.176 --> 00:04:36.656 And if you ever mess up, then suddenly you have something that is both a port and 96 00:04:36.656 --> 00:04:40.490 an adapter, and it becomes hard to tease those apart. 97 00:04:40.490 --> 00:04:44.998 What is it about Haskell that helps us write programs that are in this 98 00:04:44.998 --> 00:04:48.254 architecture and get these benefits from them? 99 00:04:48.254 --> 00:04:53.465 Yeah, the ability of purity, in types. 100 00:04:53.465 --> 00:04:58.869 Types systems enforces a lot of need to have type A, 101 00:04:58.869 --> 00:05:02.649 cross the boundary is type A still. 102 00:05:02.649 --> 00:05:05.431 Like it shouldn't be able to change across the boundery. 103 00:05:05.431 --> 00:05:08.995 >> Yeah, and we've mentioned purity a couple of times, and I just wanna be clear 104 00:05:08.995 --> 00:05:12.188 about what we're talking about specifically with regards to ASCII. 105 00:05:12.188 --> 00:05:17.792 It's a function that doesn't operate in I/O for our purposes at least at ITPro. 106 00:05:17.792 --> 00:05:21.827 That's typically what we mean by pure or not in our like application level handler. 107 00:05:21.827 --> 00:05:23.449 >> Right. >> Something that you give it a bunch of 108 00:05:23.449 --> 00:05:24.640 inputs and it gives you an output. 109 00:05:24.640 --> 00:05:25.815 End of story. 110 00:05:25.815 --> 00:05:30.432 Whereas impure code is something that has to be executed in some context, 111 00:05:30.432 --> 00:05:34.239 either with a database connection or with I/O or whatever. 112 00:05:34.239 --> 00:05:36.880 So when we're talking about pure functions, that's what we mean. 113 00:05:36.880 --> 00:05:38.001 >> Right, right, right. 114 00:05:38.001 --> 00:05:41.679 You know, we read through this article I'm a little bit in. 115 00:05:41.679 --> 00:05:44.888 It's about Ports-And-Adapters in school and stuff like that. 116 00:05:44.888 --> 00:05:49.900 But kind of talking about is a lot different than implementing it because 117 00:05:49.900 --> 00:05:54.993 in reality, Haskell kind of forces out hand this like And we don't have to 118 00:05:54.993 --> 00:06:00.684 really think about what the architecture really is in the grand scheme of things. 119 00:06:00.684 --> 00:06:03.770 >> Right, it forces you to do this architecture without having to be 120 00:06:03.770 --> 00:06:07.712 consciously aware, thinking, I'm implementing Ports-And-Adapters here. 121 00:06:07.712 --> 00:06:09.230 >> Right. >> Instead, you say, ih, 122 00:06:09.230 --> 00:06:12.538 this function suddenly needs I/O, so I have to put it in the type. 123 00:06:12.538 --> 00:06:14.134 And it propagates out everywhere. 124 00:06:14.134 --> 00:06:18.470 And then that encourages you to try to constrain the places where that's used to 125 00:06:18.470 --> 00:06:21.327 say, no let's only I/O over here at the boundary. 126 00:06:21.327 --> 00:06:22.870 And keep everything else pure. 127 00:06:22.870 --> 00:06:27.252 >> Yeah, and on our code base here, we're very adamant about that. 128 00:06:27.252 --> 00:06:32.129 If we see that, we're passing the monastic context through all of these functions, 129 00:06:32.129 --> 00:06:36.238 we don't necessarily always need the database connection every single one 130 00:06:36.238 --> 00:06:37.415 of those functions. 131 00:06:37.415 --> 00:06:42.644 We should take a step back and say, why don't we just find the data we need and 132 00:06:42.644 --> 00:06:47.977 then pass it to a pure function that allows us to have a little more certainty? 133 00:06:47.977 --> 00:06:52.968 And not have to worry too much about passing this working 134 00:06:52.968 --> 00:06:55.796 in- [CROSSTALK] >> Instead of passing this 135 00:06:55.796 --> 00:07:00.119 giant implicit context around them, basically the whole real world. 136 00:07:00.119 --> 00:07:02.474 You identify the little pieces of that that you do need and 137 00:07:02.474 --> 00:07:04.351 turn them into regular function arguments. 138 00:07:04.351 --> 00:07:05.009 >> Right. 139 00:07:05.009 --> 00:07:08.853 >> And in the course of doing that refactoring, sometimes your function's 140 00:07:08.853 --> 00:07:13.013 signature can look more complicated, because it's getting more arguments. 141 00:07:13.013 --> 00:07:16.192 Instead of just being an I/O, or app, or whatever. 142 00:07:16.192 --> 00:07:18.696 But conceptually it becomes a lot easier to reason about. 143 00:07:18.696 --> 00:07:21.822 Because just looking at that type declaration, you know those are the only 144 00:07:21.822 --> 00:07:24.217 things that has access to use that's all it's gonna do. 145 00:07:24.217 --> 00:07:24.948 >> Right. 146 00:07:24.948 --> 00:07:27.261 >> This is especially nice for a long running, 147 00:07:27.261 --> 00:07:31.339 a long lived application where many different developers are working on it. 148 00:07:31.339 --> 00:07:33.715 And people are maintaining it, fixing bugs. 149 00:07:33.715 --> 00:07:37.659 And you get that confidence on that pure code when you repack or something, 150 00:07:37.659 --> 00:07:41.979 you haven't accidentally broken some other part of the system that relied on some 151 00:07:41.979 --> 00:07:44.442 internal state that that thing fiddled with. 152 00:07:44.442 --> 00:07:45.794 >> Right, right, right. 153 00:07:45.794 --> 00:07:49.299 Yeah, and I highly value that with Haskell. 154 00:07:49.299 --> 00:07:53.946 It really makes, and there's a few of our podcasts we talked refactoring in general. 155 00:07:53.946 --> 00:07:58.770 And how easy Haskell makes that. 156 00:07:58.770 --> 00:08:02.381 But the the type features just, it makes it easy to understand. 157 00:08:02.381 --> 00:08:05.208 Okay, we're going to pass all of these things in, and 158 00:08:05.208 --> 00:08:08.531 we're going to adapt them to the type we want them to be, right? 159 00:08:08.531 --> 00:08:10.982 Okay, adapter, right? 160 00:08:10.982 --> 00:08:11.540 Crazy, guys. 161 00:08:11.540 --> 00:08:13.853 >> That's where the name came from. 162 00:08:13.853 --> 00:08:14.932 [LAUGH] >> Wow, [SOUND]. 163 00:08:14.932 --> 00:08:19.528 But that that's the nice aspect of Haskell is being able to read and 164 00:08:19.528 --> 00:08:23.549 know, okay I take these arguments and I transform them and 165 00:08:23.549 --> 00:08:26.525 I adapt them to be the type I'm returning. 166 00:08:26.525 --> 00:08:30.362 Whereas a lot of other languages that's not always the case. 167 00:08:30.362 --> 00:08:34.220 And it's not easy to understand, wait, 168 00:08:34.220 --> 00:08:38.715 why am I trying to keep these boundaries separate? 169 00:08:38.715 --> 00:08:40.626 There's other languages, just not as clear. 170 00:08:40.626 --> 00:08:44.977 >> Right, it can be hard to tell looking at a function, if it's pure or not. 171 00:08:44.977 --> 00:08:48.045 In Haskell, it's very clear if that's the case. 172 00:08:48.045 --> 00:08:53.030 And it's interesting because this blog post by Mark reminded me of an earlier 173 00:08:53.030 --> 00:08:57.479 screen cast by Gary Bernhardt who did that famous JavaScript 174 00:08:57.479 --> 00:09:02.329 talk where he shows up all these kinda weird cases of JavaScript behavior. 175 00:09:02.329 --> 00:09:05.746 But he has a screencast where he talks about this 176 00:09:05.746 --> 00:09:09.172 concept of functional core imperative shell. 177 00:09:09.172 --> 00:09:13.218 Which is very similar where all of your business logic on the quote and 178 00:09:13.218 --> 00:09:15.751 quote inside of your application is pure. 179 00:09:15.751 --> 00:09:17.609 >> Mm-hm. >> And then you have a very small layer on 180 00:09:17.609 --> 00:09:20.918 the outside that's responsible for collecting stuff from the outside world, 181 00:09:20.918 --> 00:09:23.631 passing it off to your application and then taking that result and 182 00:09:23.631 --> 00:09:25.233 sending it back to the outside world. 183 00:09:25.233 --> 00:09:25.771 >> Right. >> So 184 00:09:25.771 --> 00:09:30.001 it's funny how this concept of either functional core imperative shell or 185 00:09:30.001 --> 00:09:34.764 Ports-And-Adapters keep showing up in a guy who works with Ruby full time or Mark, 186 00:09:34.764 --> 00:09:38.341 I think he does a lot of F sharp, C sharp kind of the .NET world. 187 00:09:38.341 --> 00:09:41.584 Or even like you mentioned in computer science curriculum, which is often Java or 188 00:09:41.584 --> 00:09:42.853 Python or something like that. 189 00:09:42.853 --> 00:09:44.121 >> Right, very object oriented. 190 00:09:44.121 --> 00:09:48.344 >> All these language ecosystems have recognized that this is a powerful, 191 00:09:48.344 --> 00:09:51.329 useful architecture to set up your applications. 192 00:09:51.329 --> 00:09:54.933 But in Haskell It's just the way that you do things. 193 00:09:54.933 --> 00:09:57.784 There's it's almost harder to not follow this architecture. 194 00:09:57.784 --> 00:09:59.279 >> Right. >> So I see a lot of benefit and 195 00:09:59.279 --> 00:10:02.132 using a language that pushes you in the right direction like that. 196 00:10:02.132 --> 00:10:06.168 >> Yeah, cuz obviously if all these people are using it, 197 00:10:06.168 --> 00:10:09.508 it's not wrong, like it's- >> Probably not a bad idea. 198 00:10:09.508 --> 00:10:11.296 >> Right, like there's probably other options out there. 199 00:10:11.296 --> 00:10:15.609 But the fact that it allows you to protect your internal business logic from 200 00:10:15.609 --> 00:10:18.012 the outside world, that's important. 201 00:10:18.012 --> 00:10:20.892 >> Yeah. >> You don't want to be able to just come 202 00:10:20.892 --> 00:10:26.316 in and immediately from the outside world somehow modify some of the internal 203 00:10:26.316 --> 00:10:30.834 business logic and Haskell doesn't really allow it to happen. 204 00:10:30.834 --> 00:10:33.543 >> Yeah and it can be even more benign than leaking 205 00:10:33.543 --> 00:10:35.699 details about your business logic. 206 00:10:35.699 --> 00:10:38.196 One example I've seen before is, in Ruby, 207 00:10:38.196 --> 00:10:42.391 there is a templating language which is basically a wrapper around Ruby. 208 00:10:42.391 --> 00:10:46.619 And there had been times where you can have a template, that inside the template, 209 00:10:46.619 --> 00:10:48.285 it makes a call to the database. 210 00:10:48.285 --> 00:10:51.861 And if you put that inside in your template cuz you're like listing all of 211 00:10:51.861 --> 00:10:54.981 your users and you wanna get some associated object with them. 212 00:10:54.981 --> 00:10:58.748 You could be making a thousand queries in your template of all places. 213 00:10:58.748 --> 00:11:01.261 And templates feel like they should be pure functions. 214 00:11:01.261 --> 00:11:02.750 >> Right. >> Like don't talk to the database, 215 00:11:02.750 --> 00:11:04.303 you should already have the information you need. 216 00:11:04.303 --> 00:11:05.325 >> Right. 217 00:11:05.325 --> 00:11:08.760 >> So, this architecture helps you avoid problems like that in addition to 218 00:11:08.760 --> 00:11:09.674 a bunch of others. 219 00:11:09.674 --> 00:11:12.768 >> Right, and it's not to say that like, like you said, 220 00:11:12.768 --> 00:11:16.491 it's just harder to make Haskell not be Ports-And-Adapters. 221 00:11:16.491 --> 00:11:21.348 I can go through code we've written when we were first starting Haskell and 222 00:11:21.348 --> 00:11:24.832 just, we would pass this giant contacts all around. 223 00:11:24.832 --> 00:11:27.161 And we're gonna find data here, here and here and 224 00:11:27.161 --> 00:11:29.054 always be talking to the outside world. 225 00:11:29.054 --> 00:11:34.157 And kind of over time you kind of realize, this just doesn't feel right. 226 00:11:34.157 --> 00:11:38.028 And so Haskell allows you to refactor that out too. 227 00:11:38.028 --> 00:11:41.347 Like we can make you make this one piece. 228 00:11:41.347 --> 00:11:44.258 >> Yeah, and that feeling of this isn't quite right. 229 00:11:44.258 --> 00:11:47.309 That's Haskell nudging you in the quote unquote right direction for 230 00:11:47.309 --> 00:11:48.309 this architecture and 231 00:11:48.309 --> 00:11:51.684 where a lot of other languages kind of push you away from this architecture. 232 00:11:51.684 --> 00:11:54.159 Haskell pulls you in, so it's, yes, keep doing that. 233 00:11:54.159 --> 00:11:55.109 That's the right thing. >> You can do it. 234 00:11:55.109 --> 00:11:59.661 >> [LAUGH] >> Right, I think there's, a nice, 235 00:11:59.661 --> 00:12:05.032 Ports-And-Adapters allows you to kind of see that outside, 236 00:12:05.032 --> 00:12:10.505 you've got various aspects that are that are calling your API, 237 00:12:10.505 --> 00:12:13.674 your code base whatever that may be. 238 00:12:13.674 --> 00:12:16.017 >> Yeah, we talked about API's a lot cuz we do web programming. 239 00:12:16.017 --> 00:12:17.267 So that's kinda our bread and butter. 240 00:12:17.267 --> 00:12:18.618 >> Right. >> But this applies to other 241 00:12:18.618 --> 00:12:19.606 applications as well. 242 00:12:19.606 --> 00:12:25.407 >> Right, and allows you to see okay, what has the possibility to talk to. 243 00:12:25.407 --> 00:12:27.247 What does it have the possibility to talk to? 244 00:12:27.247 --> 00:12:32.780 And then inside it allows you to kind of keep all of those functions 245 00:12:32.780 --> 00:12:39.840 that allow you to be agile and quick and reuse and have these validation functions. 246 00:12:39.840 --> 00:12:42.423 Or having like auth checks or stuff like that, 247 00:12:42.423 --> 00:12:44.620 depending on what your object case is. 248 00:12:44.620 --> 00:12:46.944 That's calling out and finding a user and 249 00:12:46.944 --> 00:12:49.676 understanding what app permissions they have. 250 00:12:49.676 --> 00:12:55.778 That could probably involve more of a port than you're wanting to use in that case. 251 00:12:55.778 --> 00:13:02.461 But it just kind of allows you to keep all that logic just so separate. 252 00:13:02.461 --> 00:13:08.005 And not feel like, I don't know, I feel like there's been times, 253 00:13:08.005 --> 00:13:13.356 and there's frameworks that do kind of force your hand in this and 254 00:13:13.356 --> 00:13:15.615 other languages, right. 255 00:13:15.615 --> 00:13:17.334 Cuz there's MVC, right. 256 00:13:17.334 --> 00:13:22.959 MVC kind of, in my mind can translate into Ports-And-Adapters in some regards. 257 00:13:22.959 --> 00:13:26.721 >> Yeah, cuz your controller ends up usually being a port where it's collecting 258 00:13:26.721 --> 00:13:28.547 information from the outside world. 259 00:13:28.547 --> 00:13:32.819 And then your view is kind of a port in that it's presenting that information 260 00:13:32.819 --> 00:13:33.431 back out. 261 00:13:33.431 --> 00:13:36.619 And then your model is typically an adaptor in that scenario. 262 00:13:36.619 --> 00:13:42.336 >> Right, right and we, using sales when we're in JavaScript land, 263 00:13:42.336 --> 00:13:45.662 that kind of kept that stuff separate. 264 00:13:45.662 --> 00:13:50.153 They get hairy at times, like, we can find ourselves like in a service pretty much 265 00:13:50.153 --> 00:13:52.317 doing all the time controller action. 266 00:13:52.317 --> 00:13:57.467 When the idea of a service is really, it should be more than an adapter. 267 00:13:57.467 --> 00:14:00.872 And say hey take these outputs and I give you this output. 268 00:14:00.872 --> 00:14:01.466 >> Right. 269 00:14:01.466 --> 00:14:05.626 >> Rather than like we're gonna call out to this service and that service and 270 00:14:05.626 --> 00:14:08.109 we're gonna build this giant thing for you. 271 00:14:08.109 --> 00:14:11.185 And the controller's just going to say, I'll listen here and 272 00:14:11.185 --> 00:14:12.355 I accept request here. 273 00:14:12.355 --> 00:14:14.255 >> Yeah. >> And then I enter to service, 274 00:14:14.255 --> 00:14:15.419 that's a big thing. 275 00:14:15.419 --> 00:14:20.522 And that's just a super hard to go back into and understand like it doesn't 276 00:14:20.522 --> 00:14:25.952 like yeah, MVC could be like Ports-And-Adapters but it's not, it's not. 277 00:14:25.952 --> 00:14:27.298 It's easy to not do that. 278 00:14:27.298 --> 00:14:29.317 >> And nothing is pushing you in the right direction. 279 00:14:29.317 --> 00:14:31.192 >> Right. >> You said that shoving all this stuff 280 00:14:31.192 --> 00:14:34.382 into the service didn't feel right but that was just kind of a gut feeling. 281 00:14:34.382 --> 00:14:37.493 There was nothing in the language or the framework telling you, 282 00:14:37.493 --> 00:14:38.595 maybe don't do that. 283 00:14:38.595 --> 00:14:39.139 >> Right. 284 00:14:39.139 --> 00:14:42.795 >> Whereas in Haskell, you do get, you know, I have to propagate I/O through all 285 00:14:42.795 --> 00:14:45.278 these functions or I'm passing in a hundred arguments. 286 00:14:45.278 --> 00:14:47.775 This is very clearly telling me something is wrong. 287 00:14:47.775 --> 00:14:49.458 >> Mm-hm, know for sure. 288 00:14:49.458 --> 00:14:52.838 >> So I think I've said just about everything I know about this functional 289 00:14:52.838 --> 00:14:55.043 architecture being Ports-And-Adapters. 290 00:14:55.043 --> 00:14:56.709 Do you have anything else to add? 291 00:14:56.709 --> 00:14:57.537 >> I don't think so. 292 00:14:57.537 --> 00:15:00.176 I'm a little all over the place today. 293 00:15:00.176 --> 00:15:05.140 Just kind of coming out of the food coma a little bit and yeah. 294 00:15:05.140 --> 00:15:08.326 Yeah it's a little warm in here but that is what it is. 295 00:15:08.326 --> 00:15:09.676 >> It's Florida I mean. 296 00:15:09.676 --> 00:15:14.360 >> It's Florida but I also have my fair issues of sweating. 297 00:15:14.360 --> 00:15:15.990 >> At least we're not recording outside. 298 00:15:15.990 --> 00:15:17.891 >> I'm always here for you guys. 299 00:15:17.891 --> 00:15:19.336 >> [LAUGH] >> Always representing. 300 00:15:19.336 --> 00:15:24.319 But anywho we, this was a great article it was cool kind of go blast from 301 00:15:24.319 --> 00:15:27.117 the past kind of was all over the place. 302 00:15:27.117 --> 00:15:31.695 But I think because this Haskell has this identity 303 00:15:31.695 --> 00:15:36.273 that pushes you towards Ports-And-Adapters. 304 00:15:36.273 --> 00:15:42.015 It makes it really easy to just keep that logic separate from input and 305 00:15:42.015 --> 00:15:46.973 outputs and all the internal business logic those things. 306 00:15:46.973 --> 00:15:49.813 It's very important to keep those separately. 307 00:15:49.813 --> 00:15:54.932 It allows employees who are coming in or when we revisit a code later, 308 00:15:54.932 --> 00:15:59.008 like understanding like this is just business logic. 309 00:15:59.008 --> 00:16:03.596 This is the inputs, the outputs, if you need to change something later, 310 00:16:03.596 --> 00:16:07.306 like Haskell is functional and it's compiler is amazing. 311 00:16:07.306 --> 00:16:07.913 >> Sure is. 312 00:16:07.913 --> 00:16:10.098 >> And the type system is incredible. 313 00:16:10.098 --> 00:16:14.214 So that really just kind of all meshes nicely together. 314 00:16:14.214 --> 00:16:16.540 >> Yeah, well said. 315 00:16:16.540 --> 00:16:19.905 So thanks for chatting with me about Ports-And-Adapters, Cameron. 316 00:16:19.905 --> 00:16:20.999 >> Of course, thanks for having me. 317 00:16:20.999 --> 00:16:22.481 >> It's always great to have you on the show. 318 00:16:22.481 --> 00:16:25.517 And thank you for listening to the Haskell Weekly podcast. 319 00:16:25.517 --> 00:16:28.746 If you liked what you heard today, and want to know more about us, 320 00:16:28.746 --> 00:16:31.534 please check out our website at haskellweekly.news. 321 00:16:31.534 --> 00:16:35.464 This has been episode ten, and we'll see you next time. 322 00:16:35.464 --> 00:16:36.805 >> Adios.