WEBVTT NOTE Created by CaptionSync from Automatic Sync Technologies www.automaticsync.com 00:00:00.236 --> 00:00:03.696 align:middle Everybody's got their caffeine? 00:00:04.966 --> 00:00:13.096 align:middle I know it's the last session of the first day, so, everybody needs it, right? 00:00:14.716 --> 00:00:19.356 align:middle No? Somebody can live without coffee. 00:00:19.526 --> 00:00:23.606 align:middle I need some help with how to do that. 00:00:25.436 --> 00:00:33.946 align:middle So, first off, I see a lot of young faces in the crowd, so, do you know what this thing is? 00:00:34.126 --> 00:00:40.106 align:middle That's how old when I started developing professionally. 00:00:40.566 --> 00:00:42.886 align:middle So, almost 20 years now. 00:00:43.656 --> 00:00:47.956 align:middle So, I started off doing small websites because at 00:00:47.956 --> 00:00:51.326 align:middle that time people just wanted to have some internet presence. 00:00:52.226 --> 00:00:59.056 align:middle And far away from any kind of rocket science, even by those days, and for most of you today, 00:00:59.346 --> 00:01:02.226 align:middle it would be like, yeah, really what's the problem there? 00:01:02.226 --> 00:01:09.796 align:middle But that was the time that we didn't have cool things like frameworks, templating engines, 00:01:10.436 --> 00:01:13.836 align:middle database abstraction layers or ORMs. 00:01:14.756 --> 00:01:20.176 align:middle We had to handle all of that ourselves, but this was actually the fun part of the job. 00:01:20.906 --> 00:01:25.426 align:middle The problems started when you had to make it work for things 00:01:25.426 --> 00:01:28.936 align:middle like Netscape or Internet explorer 5.5. 00:01:29.626 --> 00:01:35.846 align:middle And if any of you are doing front-end these days and think you're screwed, trust me. 00:01:37.636 --> 00:01:45.356 align:middle And, I got so uninterested in development that I actually went back 00:01:45.356 --> 00:01:50.336 align:middle to university doing some freelance on the side, earning some money. 00:01:50.336 --> 00:01:56.526 align:middle But about 10 years ago I found a team, I found a company, I found a really great job. 00:01:56.526 --> 00:01:58.586 align:middle It was a small product. 00:01:58.586 --> 00:02:02.826 align:middle It started off, let's say, like a dog house - really, really small. 00:02:04.006 --> 00:02:10.916 align:middle As we kept working on it, it kept growing in both features, customers, money. 00:02:12.206 --> 00:02:15.536 align:middle So, I was good with, let's say, carpentry. 00:02:16.166 --> 00:02:23.546 align:middle So, in few months we had a shed, which meant more and more people could fit in. 00:02:23.736 --> 00:02:26.316 align:middle They wanted more and more, more money was coming in, 00:02:26.316 --> 00:02:28.726 align:middle more customers, more content, more everything. 00:02:29.896 --> 00:02:31.006 align:middle So it's grew again. 00:02:32.276 --> 00:02:36.866 align:middle You understand that you can't use wood to build skyscrapers, right? 00:02:37.326 --> 00:02:38.556 align:middle You have to change your base. 00:02:39.076 --> 00:02:44.186 align:middle So yes, I had to learn how to use concrete to get a big, better stuff done. 00:02:45.646 --> 00:02:51.136 align:middle But what actually, my code looked like, something like this. 00:02:51.306 --> 00:02:55.796 align:middle You know, and to be honest, probably, in this case, 00:02:56.136 --> 00:02:58.806 align:middle this was done by somebody and then by somebody else. 00:02:58.976 --> 00:03:02.076 align:middle And for most of it was us, us. 00:03:04.736 --> 00:03:07.806 align:middle And, okay, it was ugly, but it worked. 00:03:08.786 --> 00:03:14.466 align:middle But what our customers wanted was bigger and bigger and nicer and not now concrete. 00:03:14.936 --> 00:03:17.606 align:middle It's aluminum, it's glass, it's shiny. 00:03:18.086 --> 00:03:24.166 align:middle But as we kept growing, we kept failing. 00:03:24.316 --> 00:03:28.556 align:middle Because, hey, I didn't have the experience needed for that project. 00:03:29.826 --> 00:03:37.066 align:middle I used to work in small websites, small apps, not 100 million user visits a month. 00:03:37.126 --> 00:03:46.146 align:middle And what customers actually paid and what I thought was, oh, this is what we're doing. 00:03:46.686 --> 00:03:55.166 align:middle But in honesty, most of the developers don't understand that businesses want a nice feature, 00:03:55.466 --> 00:03:59.566 align:middle another nice feature, they really really want something to just glue them together. 00:04:00.646 --> 00:04:04.356 align:middle By design, because businesses don't know always what they want. 00:04:05.346 --> 00:04:09.556 align:middle So, this project has took about five years of my life. 00:04:10.476 --> 00:04:16.456 align:middle And the next project that I came on, yeah, it's going to grow like that, right? 00:04:16.456 --> 00:04:21.066 align:middle It's going to be that small product is going to be the biggest thing ever. 00:04:21.856 --> 00:04:25.176 align:middle Unfortunately, what I was suffering from was Stockholm Syndrome. 00:04:26.126 --> 00:04:31.066 align:middle Because in reality, not every project or product becomes big. 00:04:32.396 --> 00:04:36.196 align:middle So, let's say that you want to buy a dog house, like a real dog house. 00:04:36.196 --> 00:04:38.606 align:middle You have a dog and you want to buy a dog house. 00:04:39.756 --> 00:04:43.976 align:middle How would you feel if I came to you and told you that the price 00:04:44.016 --> 00:04:49.216 align:middle for this nice small dog house includes something like this? 00:04:50.546 --> 00:04:56.946 align:middle Because it's going to be a skyscraper so I want to do it before and avoid all of the problems. 00:04:57.676 --> 00:05:03.416 align:middle Right? Found yourself any time in that situation? 00:05:05.736 --> 00:05:11.526 align:middle So, Donald Knuth said: Premature optimization is the root of all evil. 00:05:12.786 --> 00:05:17.936 align:middle And, that's why I want to talk to you about when testing makes no sense. 00:05:18.546 --> 00:05:19.566 align:middle My name is Miro Svrtan. 00:05:19.626 --> 00:05:21.686 align:middle You can find me on Twitter using this handle. 00:05:21.686 --> 00:05:27.606 align:middle So why am I talking about buildings and testing? 00:05:28.586 --> 00:05:29.486 align:middle What's the connection? 00:05:30.616 --> 00:05:34.286 align:middle So, who here is testing? 00:05:34.826 --> 00:05:36.806 align:middle It's an easy question, right? 00:05:37.606 --> 00:05:41.436 align:middle Okay. Who here isn't testing? 00:05:41.566 --> 00:05:46.486 align:middle Again, so, this was a partially loaded question. 00:05:46.826 --> 00:05:47.516 align:middle Sorry for that. 00:05:48.036 --> 00:05:50.186 align:middle But actually all of us test. 00:05:50.656 --> 00:05:57.336 align:middle You write some code, you open your browser, you check that it works - it's also a testing. 00:05:57.336 --> 00:06:02.286 align:middle So, I hope nobody just writes the code and pushes it to production. 00:06:04.316 --> 00:06:06.656 align:middle To be honest, I did it few times. 00:06:06.656 --> 00:06:10.946 align:middle But not as a normal way of working. 00:06:12.426 --> 00:06:16.076 align:middle So, let's say that we have a small website. 00:06:17.166 --> 00:06:20.166 align:middle No Mars rover thing, just a small website. 00:06:21.076 --> 00:06:21.966 align:middle And we have a bug. 00:06:22.966 --> 00:06:28.306 align:middle How much time do you think after somebody reports it will take us to fix it? 00:06:29.656 --> 00:06:30.496 align:middle Five minutes? 00:06:30.836 --> 00:06:33.876 align:middle An hour? Half a day? 00:06:35.816 --> 00:06:38.006 align:middle How much time does it take you to deploy? 00:06:39.716 --> 00:06:42.236 align:middle Probably 5 to 10, maybe 15 minutes. 00:06:42.756 --> 00:06:49.446 align:middle Right? So, why would we test and invest in testing 00:06:49.446 --> 00:06:52.026 align:middle when we can fix something quickly and deploy it everywhere? 00:06:53.086 --> 00:07:00.526 align:middle But the problem was that about 7-8 years ago, a lot of web developers became phone developers. 00:07:02.116 --> 00:07:03.396 align:middle Been there, done that. 00:07:03.396 --> 00:07:08.146 align:middle A few of my colleagues went that way, but they didn't have the mindset of testing. 00:07:09.296 --> 00:07:14.886 align:middle Which is pretty nice for Google store because it takes about 3 days 00:07:14.886 --> 00:07:16.946 align:middle after you fix it for people to get it. 00:07:17.846 --> 00:07:20.636 align:middle But what if it's an iPhone? 00:07:20.636 --> 00:07:25.766 align:middle iOS app? 6-7 years ago it would take from 2 weeks to 3 weeks, 00:07:25.766 --> 00:07:29.036 align:middle sometimes even a month to fix something. 00:07:29.526 --> 00:07:33.446 align:middle What if you're doing desktop applications? 00:07:33.816 --> 00:07:39.206 align:middle You have a bug and now all of your customers have to be somehow contacted 00:07:39.356 --> 00:07:42.276 align:middle to download the new version because, okay, 00:07:42.636 --> 00:07:47.386 align:middle maybe Mac users do have App store, but Windows users not so much. 00:07:48.736 --> 00:07:51.056 align:middle How hard does it then to fix a bug? 00:07:51.636 --> 00:07:54.966 align:middle It might not be a critical bug, but it might be something 00:07:55.166 --> 00:07:57.336 align:middle that bothers your customers, your users. 00:07:58.856 --> 00:08:00.836 align:middle Hey, we have hardware. 00:08:01.836 --> 00:08:03.566 align:middle But that's at least easy, right? 00:08:03.616 --> 00:08:07.356 align:middle Because, this is a router, so it's connected to internet, right? 00:08:07.356 --> 00:08:08.736 align:middle So it's easy to fix it. 00:08:09.146 --> 00:08:13.036 align:middle If we have a bug, we can just ship it to all the routers. 00:08:13.506 --> 00:08:18.906 align:middle What if the bug is in the piece of code getting the update? 00:08:21.396 --> 00:08:26.216 align:middle Are you going to send all of your staff to the whole continent, 00:08:26.216 --> 00:08:29.346 align:middle to the whole world to fix people's routers? 00:08:31.356 --> 00:08:32.526 align:middle What if it's... 00:08:32.526 --> 00:08:34.266 align:middle what if it's a washing machine? 00:08:35.056 --> 00:08:42.156 align:middle Hopefully, it's not connected to internet, but you find out that you shipped some of it 00:08:43.066 --> 00:08:46.306 align:middle and you can't wash cotton on 60 degrees. 00:08:47.776 --> 00:08:52.666 align:middle So, you now have to call all of your mechanics, all of your service departments, 00:08:53.266 --> 00:08:56.296 align:middle to contact the customers to flash the washing machine. 00:08:57.456 --> 00:09:03.806 align:middle So, when should we write tests? 00:09:03.956 --> 00:09:05.946 align:middle When it's a high cost to fix? 00:09:06.476 --> 00:09:16.736 align:middle Because, maybe, sometimes it's cheaper to test than to try to fix it once it happens. 00:09:16.736 --> 00:09:19.366 align:middle Maybe when the cost is larger. 00:09:19.746 --> 00:09:23.386 align:middle Let's go back to the website. 00:09:24.546 --> 00:09:25.906 align:middle So, should we ever test websites? 00:09:25.906 --> 00:09:27.976 align:middle We can fix it easily, right? 00:09:29.076 --> 00:09:29.886 align:middle Yep. Right? 00:09:30.286 --> 00:09:31.556 align:middle Anybody? Nobody? 00:09:32.006 --> 00:09:33.386 align:middle Everything test? 00:09:35.296 --> 00:09:37.966 align:middle But banks do use websites as well. 00:09:39.326 --> 00:09:49.496 align:middle I mean I hope you don't experience a bug where you send the money instead of receiving it. 00:09:50.446 --> 00:09:53.886 align:middle The other way around, probably, you'd like it. 00:09:55.696 --> 00:09:59.396 align:middle Right? So, let's maybe look at it as when it brings value. 00:10:00.076 --> 00:10:07.816 align:middle Because value is not just the money, it's how much your consumers love the product, 00:10:08.076 --> 00:10:11.576 align:middle how much your clients are happy with your work. 00:10:12.516 --> 00:10:19.316 align:middle Maybe for you, if you're a development company, maybe because of, maybe, it would bring value 00:10:19.316 --> 00:10:23.676 align:middle in the sense of people not leaving the company often because they would be more happy? 00:10:25.336 --> 00:10:30.306 align:middle There's a lot of cost when a disgruntled employee decides to leave 00:10:30.976 --> 00:10:33.706 align:middle because then you have to get somebody in. 00:10:34.296 --> 00:10:39.266 align:middle It might take about 6 months to get the new person in. 00:10:42.076 --> 00:10:47.116 align:middle So, let's define the value as benefit minus the cost. 00:10:47.226 --> 00:10:55.066 align:middle So, what I'm trying to say that if, let's say, you have a tomato farm that produces a lot 00:10:55.066 --> 00:10:59.596 align:middle of tomatoes, you probably shouldn't use the same tactic - 00:11:00.506 --> 00:11:10.366 align:middle you might use on your own tomato garden. 00:11:10.526 --> 00:11:16.746 align:middle You maybe like your tomatoes a lot, but would you, um, would you create a whole lab? 00:11:16.746 --> 00:11:18.996 align:middle Would you test it every day? 00:11:20.346 --> 00:11:27.136 align:middle Probably for tomato farm you would because if one goes down - you might lose everything. 00:11:27.816 --> 00:11:29.456 align:middle Everything, if you get the disease. 00:11:30.496 --> 00:11:34.906 align:middle In case of your tomatoes in the garden, you can just go to a farmer's market 00:11:34.906 --> 00:11:37.096 align:middle or a supermarket and get some others. 00:11:37.666 --> 00:11:43.736 align:middle Okay? So, value is benefit minus the cost. 00:11:44.126 --> 00:11:48.396 align:middle Of course, I do understand that some people might invest a lot of money 00:11:48.506 --> 00:11:53.926 align:middle into their tomato gardens because it's a hobby and we tend to spend a lot of money 00:11:53.956 --> 00:11:56.796 align:middle on our hobbies, but let's look at it as a business. 00:11:57.326 --> 00:11:59.296 align:middle The benefit minus the cost. 00:12:00.036 --> 00:12:07.726 align:middle So there is some, before I go further, there are lots of different things about testing. 00:12:08.156 --> 00:12:14.806 align:middle So what I'm going to try to focus on is writing test, automating the tests, 00:12:15.306 --> 00:12:21.236 align:middle the ones that help you ensure that the app does what was expected? 00:12:22.026 --> 00:12:27.756 align:middle Okay? So, who here writes tests in that way? 00:12:27.926 --> 00:12:30.936 align:middle Okay. Who tests everything? 00:12:35.646 --> 00:12:37.886 align:middle Okay. 6 to 7 hands. 00:12:38.616 --> 00:12:41.566 align:middle So, actually, can you raise your hands up? 00:12:42.616 --> 00:12:43.736 align:middle Who tests everything? 00:12:44.576 --> 00:12:46.176 align:middle How much security testing do we do? 00:12:47.536 --> 00:12:50.566 align:middle Performance testing? 00:12:50.676 --> 00:12:52.176 align:middle Visual regression testing? 00:12:53.016 --> 00:12:54.076 align:middle That all is testing, right? 00:12:55.326 --> 00:13:02.136 align:middle So, why I decided to do this talk was people came to a stage 00:13:02.136 --> 00:13:04.266 align:middle like this and said we test everything. 00:13:04.676 --> 00:13:11.906 align:middle And a lot of people understood that, yeah, to be a good developer you have to test everything. 00:13:11.906 --> 00:13:18.476 align:middle And they started testing, but they started testing the wrong things. 00:13:18.656 --> 00:13:19.696 align:middle I've done it that way. 00:13:20.206 --> 00:13:29.966 align:middle And, 6 months ago I did this talk in Amsterdam and I spoke to another speaker afterwards. 00:13:29.966 --> 00:13:34.556 align:middle He wasn't at my talk, but he was interested in what I was talking about and he told me: Hey, 00:13:34.556 --> 00:13:39.816 align:middle yeah, well no, you shouldn't be talking about when not to test because we test everything. 00:13:41.136 --> 00:13:43.606 align:middle I said: Okay, so performance? 00:13:43.856 --> 00:13:45.766 align:middle Oh, no. Security? 00:13:46.426 --> 00:13:48.146 align:middle Some. Visual regression? 00:13:48.486 --> 00:13:50.746 align:middle Nah, we actually TDD everything. 00:13:51.226 --> 00:13:53.836 align:middle Wait, you're using Symfony and you TDD everything? 00:13:54.146 --> 00:14:02.376 align:middle Yes. I was really interested because I found TDD to be very wrong for Symfony, the Symfony part, 00:14:02.736 --> 00:14:05.456 align:middle like using the Forms, using the Doctrine and stuff like that. 00:14:06.346 --> 00:14:09.096 align:middle And I was wondering, okay, how do you do that? 00:14:09.306 --> 00:14:10.496 align:middle Oh, we don't. 00:14:12.616 --> 00:14:21.296 align:middle So, wait, you just told me you test everything and you do it TDD way, but you don't do the UI, 00:14:21.366 --> 00:14:24.616 align:middle the infrastructure and stuff like that? 00:14:24.616 --> 00:14:27.196 align:middle Nope, that's the UI. 00:14:27.286 --> 00:14:32.546 align:middle So if I wasn't really interested in that part, like how do you do that? 00:14:33.456 --> 00:14:38.286 align:middle I would think, yeah, I'm the stupid one because I don't test everything. 00:14:39.146 --> 00:14:40.826 align:middle Please, don't be. 00:14:41.446 --> 00:14:43.266 align:middle There are a lot of fallacies with testing. 00:14:47.056 --> 00:14:54.286 align:middle 6 or 7 years ago, when I wanted to start writing tests, one of the biggest problems I had, 00:14:54.286 --> 00:14:58.216 align:middle I was a team lead at the time, was... 00:14:58.436 --> 00:15:03.076 align:middle So similar feature was about 100 hours. 00:15:03.796 --> 00:15:10.916 align:middle Now I need to set an estimate of about 150 hours because, well, 00:15:12.186 --> 00:15:15.496 align:middle writing the code, writing the tests, right? 00:15:15.746 --> 00:15:19.546 align:middle So it's gonna take more money from the clients. 00:15:19.596 --> 00:15:20.836 align:middle They're not going to be happy. 00:15:22.136 --> 00:15:28.906 align:middle What I didn't think of - do you charge for manual testing? 00:15:29.006 --> 00:15:36.656 align:middle So, do we say: Hey, it's gonna take me 7 hours to write the code and X amount of hours 00:15:36.656 --> 00:15:40.476 align:middle to run it in the browser as I click save on the code every time. 00:15:41.056 --> 00:15:46.406 align:middle We think of it as the same thing, it's part of development. 00:15:48.156 --> 00:15:56.126 align:middle And, if you have to explain to anybody, try to explain to them the bugs cost too. 00:15:56.926 --> 00:16:05.086 align:middle So most of the companies should have, should solve the bugs for free. 00:16:06.016 --> 00:16:11.716 align:middle Right? I mean I know of companies whose whole business plan is 00:16:12.266 --> 00:16:15.596 align:middle to charge for bugs, for bug fixing. 00:16:15.596 --> 00:16:21.686 align:middle But you might notice that that's a pretty bad business model because somehow they want 00:16:21.686 --> 00:16:25.606 align:middle to produce more bugs because that means more work for them. 00:16:26.146 --> 00:16:30.746 align:middle Right? Anybody here has a car? 00:16:32.056 --> 00:16:33.416 align:middle A new car maybe? 00:16:33.546 --> 00:16:34.906 align:middle Or somebody's buying a new car? 00:16:35.626 --> 00:16:41.156 align:middle Recently? So, let's say you want to buy this car. 00:16:41.726 --> 00:16:46.356 align:middle You go to, I think it's an Opel dealership, and you say: Hey, I want this one. 00:16:46.626 --> 00:16:47.366 align:middle How much does it cost? 00:16:47.366 --> 00:16:54.246 align:middle Oh, 20,000 euros, let's say, I have no idea, please, don't get me on the numbers. 00:16:55.216 --> 00:16:57.026 align:middle Okay. So, it's tested, right? 00:16:57.636 --> 00:16:59.536 align:middle Yeah. So can I get the version that's not? 00:16:59.686 --> 00:17:02.586 align:middle It's going to be cheaper, right? 00:17:03.476 --> 00:17:05.706 align:middle Because why would I pay for testing? 00:17:06.876 --> 00:17:11.226 align:middle So products don't come out without those things. 00:17:11.836 --> 00:17:16.316 align:middle I mean, when you buy food, you expect it to be tested, right? 00:17:17.506 --> 00:17:18.856 align:middle You don't want to be poisoned. 00:17:23.036 --> 00:17:25.406 align:middle Try to ask for forgiveness, not permission. 00:17:26.296 --> 00:17:28.096 align:middle Just try to do it. 00:17:28.636 --> 00:17:30.966 align:middle Try to do it in a small way. 00:17:31.176 --> 00:17:32.746 align:middle Start building up your tests. 00:17:33.886 --> 00:17:37.366 align:middle Every new feature, add more tests. 00:17:37.986 --> 00:17:43.186 align:middle You don't have to do everything the first time. 00:17:43.406 --> 00:17:50.146 align:middle Another fallacy is that it takes a lot of time. 00:17:50.316 --> 00:17:56.856 align:middle That might be true for a first timer. 00:17:57.036 --> 00:17:58.816 align:middle Because there is a cost of learning. 00:18:00.286 --> 00:18:05.026 align:middle I mean, I guess, most of you that had a car that are driving had 00:18:05.026 --> 00:18:07.286 align:middle to go to school for driving, right? 00:18:07.286 --> 00:18:09.846 align:middle It took like 20 to 40 hours. 00:18:10.226 --> 00:18:19.446 align:middle Right? So, you don't take that into the cost of every time you go to the shop. 00:18:19.446 --> 00:18:21.946 align:middle Right? It's learning. 00:18:22.926 --> 00:18:28.896 align:middle Yes when your first project, it's going to cost you, your company, your product team, whatever. 00:18:29.276 --> 00:18:30.216 align:middle But it's learning. 00:18:30.816 --> 00:18:31.656 align:middle You wish don't learn. 00:18:36.516 --> 00:18:42.206 align:middle It takes extra time writing the code, writing the test. 00:18:42.676 --> 00:18:46.856 align:middle Right? So, if I rock, if I write 10 lines of code, 00:18:47.226 --> 00:18:49.996 align:middle I need to write some lines of tests as well. 00:18:49.996 --> 00:18:52.156 align:middle So, that takes time, right? 00:18:52.986 --> 00:18:56.866 align:middle Manual testing takes time as well because you have to Alt+Tab 00:18:56.866 --> 00:18:59.846 align:middle or Ctrl+Tab something, check your browser. 00:19:00.446 --> 00:19:05.386 align:middle So, if you write your tests as you're writing your code, 00:19:05.946 --> 00:19:08.566 align:middle you don't have to do that manual check. 00:19:09.076 --> 00:19:16.396 align:middle You can just run the test and it will tell you that, what you expected, works. 00:19:19.506 --> 00:19:27.186 align:middle So, for some years I thought that if I implement something that's ready for production, 00:19:27.466 --> 00:19:34.006 align:middle let's say, you want to implement a new social login or a new payment gateway 00:19:34.976 --> 00:19:42.386 align:middle and you get one PHP file where you see that it works, then that's it, right? 00:19:42.656 --> 00:19:48.276 align:middle But actual feature lifecycle is the part where you explore. 00:19:48.796 --> 00:19:50.766 align:middle You look how that system works. 00:19:51.296 --> 00:19:53.376 align:middle It could be a library, it could be a new database, 00:19:53.376 --> 00:19:56.446 align:middle it can be a third party integration, it could be whatever. 00:19:56.596 --> 00:19:59.426 align:middle That is the part where you don't test. 00:20:00.606 --> 00:20:05.966 align:middle You should explore, you should enjoy, you should understand what you're doing. 00:20:07.176 --> 00:20:08.596 align:middle Then you start modeling. 00:20:09.776 --> 00:20:16.656 align:middle Again, we're not testing, you're using tests but you're not testing - you're modeling. 00:20:17.776 --> 00:20:21.836 align:middle And this is a very confusing thing, when I try to explain it to people. 00:20:22.636 --> 00:20:24.926 align:middle If I'm writing tests, right? 00:20:25.726 --> 00:20:27.026 align:middle How is that not testing? 00:20:28.256 --> 00:20:33.276 align:middle So, if you are in a car, it doesn't mean that you are the driver. 00:20:35.016 --> 00:20:40.886 align:middle Okay? Uber and Taxi doesn't work that way. 00:20:41.046 --> 00:20:46.176 align:middle Okay? So, yes you could be using tests just like you use a car, 00:20:46.176 --> 00:20:49.126 align:middle but that doesn't mean you're driving it. 00:20:51.476 --> 00:20:57.396 align:middle The third part is development and that's where testing needs to be done. 00:20:57.696 --> 00:20:58.476 align:middle A lot of it. 00:20:58.966 --> 00:21:03.376 align:middle Because you're not sure about all of the edge cases that could happen. 00:21:04.076 --> 00:21:06.366 align:middle Here tests have a great value. 00:21:07.566 --> 00:21:13.946 align:middle And usually, when I go to production, I delete some of the tests. 00:21:14.086 --> 00:21:17.906 align:middle Because when I was writing them in the previous phase, 00:21:18.496 --> 00:21:26.516 align:middle I wasn't sure which of the tests will have value later and I don't want my tests to run 00:21:26.586 --> 00:21:32.616 align:middle for hours just because during the development phase I wasn't sure if that can break or not. 00:21:33.506 --> 00:21:39.666 align:middle I'm not saying delete all of the tests, but be... 00:21:40.106 --> 00:21:44.796 align:middle feel free to delete some of those that you do not need. 00:21:45.796 --> 00:21:51.656 align:middle 100% code coverage is something that I had a lot of problems with. 00:21:51.976 --> 00:21:54.136 align:middle And I hear from the others. 00:21:58.896 --> 00:22:05.726 align:middle If I don't check the code coverage, how would I know if something is tested or not? 00:22:07.416 --> 00:22:12.766 align:middle So you want to go and have 100 percent because in that case you know your app is tested. 00:22:14.066 --> 00:22:21.196 align:middle So, what usually people start is, well, let's get the code coverage as high as we can, 00:22:21.436 --> 00:22:26.526 align:middle as cheap as we can, which means testing getters and setters. 00:22:31.496 --> 00:22:39.526 align:middle It can be fun for like one or two entities, five-ish is Ballmer's peak. 00:22:40.756 --> 00:22:42.876 align:middle Oh, need of Ballmer's peak. 00:22:43.856 --> 00:22:48.446 align:middle About 10 - I'm happy if everybody doesn't quit their company. 00:22:50.456 --> 00:22:58.326 align:middle So, should we test something like this? 00:22:58.526 --> 00:23:02.726 align:middle I really don't see the value. 00:23:02.726 --> 00:23:06.456 align:middle When I moved to using commands and events, command and handler events, 00:23:06.456 --> 00:23:11.526 align:middle command handler patterns, I found that I spent a lot of time 00:23:11.526 --> 00:23:14.516 align:middle on testing those setters and getters, once again. 00:23:15.716 --> 00:23:20.886 align:middle What now I usually do is if I do use code coverage, I just put all of the commands 00:23:20.886 --> 00:23:23.096 align:middle and event namespaces into ignore. 00:23:23.926 --> 00:23:24.926 align:middle And I just don't care. 00:23:25.716 --> 00:23:32.456 align:middle I want to achieve the higher percentage, but I don't want to know about those DTO classes. 00:23:32.756 --> 00:23:39.076 align:middle We have tests. 00:23:39.236 --> 00:23:43.306 align:middle I've heard some people complaining a few years ago about they wanted 00:23:43.306 --> 00:23:46.686 align:middle to participate in open source project. 00:23:49.176 --> 00:23:54.786 align:middle And they implemented the changes because implementing those changes was rather easy, 00:23:55.456 --> 00:23:59.436 align:middle but making all of the tests work was the hard part. 00:24:01.136 --> 00:24:07.146 align:middle You don't want you and your fellow developers to be in the position that, 00:24:07.516 --> 00:24:11.056 align:middle to change the code, is so hard because of the tests. 00:24:11.596 --> 00:24:16.896 align:middle Because people in that case will not test. 00:24:16.896 --> 00:24:21.946 align:middle I mean if you have a car on the bottom of a lake, what's the value of it? 00:24:23.876 --> 00:24:29.326 align:middle I mean, and that most often comes from: Let's write tests later, 00:24:31.816 --> 00:24:34.656 align:middle which is most often coming from management. 00:24:35.136 --> 00:24:43.236 align:middle And as a disgruntled developer, I might come, I might build the whole feature. 00:24:43.296 --> 00:24:50.286 align:middle I know everything works and before merging it, my manager comes and says: where are the tests? 00:24:50.436 --> 00:24:52.016 align:middle We're not merging that without tests? 00:24:52.786 --> 00:24:54.266 align:middle No. Everything works. 00:24:54.926 --> 00:24:55.536 align:middle I don't care. 00:24:55.536 --> 00:24:56.226 align:middle Everything works. 00:24:56.556 --> 00:25:01.086 align:middle No. You have to write those tests. 00:25:01.236 --> 00:25:02.136 align:middle What do you think? 00:25:02.136 --> 00:25:04.846 align:middle What kind of tests am I going to write? 00:25:04.846 --> 00:25:08.216 align:middle I'm just going to look. 00:25:08.526 --> 00:25:09.806 align:middle I'm going to split my screen. 00:25:10.406 --> 00:25:14.656 align:middle I'm just going to look at the code on the left side and replicate it on the right side. 00:25:15.656 --> 00:25:17.346 align:middle Because that's going to get shit done. 00:25:19.536 --> 00:25:28.346 align:middle So, yea, I'm just gonna add a simple case for this, maybe. 00:25:30.976 --> 00:25:36.566 align:middle Who here knows what TDD is? 00:25:37.746 --> 00:25:39.526 align:middle Anybody practicing it? 00:25:40.516 --> 00:25:52.356 align:middle Okay. If I say, if I say test driven development - is that TDD? 00:25:52.446 --> 00:25:54.136 align:middle No, sorry, another loaded question. 00:25:55.396 --> 00:25:56.966 align:middle Test driven design. 00:25:57.526 --> 00:26:01.796 align:middle Not development, design. 00:26:02.506 --> 00:26:08.446 align:middle You're supposed to use TDD to build your architecture, to model it. 00:26:09.996 --> 00:26:15.546 align:middle You can write your tests first and then your code - that's not TDD. 00:26:16.726 --> 00:26:23.476 align:middle TDD is that small red, green, refactor, red, green, refactor, red, green, refactor. 00:26:26.556 --> 00:26:31.886 align:middle Anybody's remembering this tweet? 00:26:32.066 --> 00:26:34.906 align:middle Anybody's remembering the shit storm coming after it? 00:26:36.176 --> 00:26:39.796 align:middle Where everybody came with: Haha, you see, testing makes no sense. 00:26:40.356 --> 00:26:41.946 align:middle You shouldn't test anytime. 00:26:42.446 --> 00:26:45.366 align:middle Why were you forcing me to test? 00:26:45.366 --> 00:26:50.456 align:middle People don't understand the TDD and testing are completely different things. 00:26:50.826 --> 00:26:55.846 align:middle Uh, do you all know who DHH is? 00:26:57.136 --> 00:26:59.336 align:middle So, he's an author of Ruby on Rails framework. 00:27:00.256 --> 00:27:05.786 align:middle He works for a small company called Basecamp, used to be called 37 signals 00:27:05.786 --> 00:27:10.936 align:middle and they have a few small products, really, really small products, really cool products, 00:27:11.006 --> 00:27:15.796 align:middle but really, really small products, that they have been building for about 10 or 15 years. 00:27:16.976 --> 00:27:21.866 align:middle What kind of modeling would they need when they know what they're doing? 00:27:22.356 --> 00:27:26.146 align:middle There's no modeling in it. 00:27:27.076 --> 00:27:30.406 align:middle They just need to test that what they want works. 00:27:31.186 --> 00:27:32.606 align:middle It's not really modeling. 00:27:33.206 --> 00:27:38.436 align:middle And, just to try to explain to you, that's one of the rare companies that would come 00:27:38.436 --> 00:27:44.116 align:middle to their clients, when they would ask for features, they told them: No. 00:27:45.176 --> 00:27:49.346 align:middle We have our product, we love it, as simple as it is. 00:27:49.856 --> 00:27:52.466 align:middle If you want more features - use Jira. 00:27:52.466 --> 00:27:53.896 align:middle Use this, use that. 00:27:54.386 --> 00:27:54.826 align:middle Thank you. 00:27:54.936 --> 00:27:55.626 align:middle Here's the door. 00:27:58.156 --> 00:28:03.306 align:middle So, don't forget the environment that this comes from. 00:28:03.856 --> 00:28:06.596 align:middle They don't build that many features. 00:28:11.496 --> 00:28:18.996 align:middle If you ever looked into TDD, one of the things that even the people who invented it said, 00:28:18.996 --> 00:28:21.046 align:middle you have to do everything the TDD way. 00:28:21.046 --> 00:28:28.966 align:middle Only in the last few years they started saying: Well, we were wrong. 00:28:29.176 --> 00:28:35.466 align:middle Because trying to build everything TDD way takes a lot of time. 00:28:35.976 --> 00:28:42.426 align:middle And if you're yet another payment gateway, yet another CMS, yet another E-commerce 00:28:42.686 --> 00:28:47.496 align:middle and you know how everything has to look, what part of modeling do you have to do? 00:28:48.426 --> 00:28:50.236 align:middle It's yet another house. 00:28:51.916 --> 00:28:55.456 align:middle Maybe has another color, but that's about it. 00:29:00.056 --> 00:29:04.956 align:middle I'm going to guess that some of you changed the company a few times. 00:29:05.386 --> 00:29:07.646 align:middle Have you ever come... 00:29:07.906 --> 00:29:15.886 align:middle have you ever changed it to company that has good tests? 00:29:16.046 --> 00:29:17.126 align:middle Whoa. Really? 00:29:17.626 --> 00:29:19.446 align:middle Wow. Three hands. 00:29:21.436 --> 00:29:33.766 align:middle Wow. So, my previous employer had such a great coverage that I started shipping the next day. 00:29:34.256 --> 00:29:37.086 align:middle I came on Monday and started working on Tuesday. 00:29:37.236 --> 00:29:45.626 align:middle Because it was easy to onboard me because I could have gone through the tests 00:29:45.956 --> 00:29:51.216 align:middle to understand what and why are things happening and even if I screw it up, 00:29:51.926 --> 00:29:58.806 align:middle because it's my second day, people, the CI says: Hey, this doesn't work anymore. 00:30:01.846 --> 00:30:07.476 align:middle It's much easier to add new features because you know that you didn't break older ones. 00:30:08.706 --> 00:30:11.306 align:middle You know that the older ones still work. 00:30:11.826 --> 00:30:17.436 align:middle So, what do we had? 00:30:17.736 --> 00:30:19.776 align:middle This, the previous exam? 00:30:20.416 --> 00:30:21.166 align:middle We know this suck. 00:30:21.166 --> 00:30:22.936 align:middle We have to add a feature now. 00:30:25.236 --> 00:30:31.716 align:middle The feature is that if you're not from, um, from that European Union country, I'm using Portugal 00:30:31.716 --> 00:30:38.316 align:middle since we're here, you shouldn't charge a VAT if it's a company. 00:30:38.386 --> 00:30:43.266 align:middle So, maybe this is the point where you should test. 00:30:43.456 --> 00:30:49.266 align:middle Switching from rather simple if, to another. 00:30:53.736 --> 00:30:55.916 align:middle What if you have to refactor something? 00:30:56.726 --> 00:30:58.136 align:middle So you have all of your tests. 00:30:58.376 --> 00:31:03.216 align:middle They run, you just change the code and all of your tests say: Hey, it still works. 00:31:03.376 --> 00:31:03.926 align:middle Still works. 00:31:04.216 --> 00:31:06.406 align:middle It's okay, everything's good. 00:31:08.676 --> 00:31:17.606 align:middle So let's make this bit more readable to avoid nested ifs. 00:31:18.186 --> 00:31:20.686 align:middle So this is what happened. 00:31:21.116 --> 00:31:22.506 align:middle This is what I have. 00:31:23.316 --> 00:31:25.466 align:middle Can anybody spot the bug? 00:31:27.276 --> 00:31:27.946 align:middle Where? Here? 00:31:27.946 --> 00:31:45.986 align:middle Nope. You always have to charge VAT for that country 00:31:46.256 --> 00:31:49.716 align:middle if you're selling something in your country. 00:31:50.216 --> 00:31:53.186 align:middle Any others? 00:31:54.386 --> 00:32:12.186 align:middle So, I just wasted two minutes of your life trying to understand if this was okay or not. 00:32:13.796 --> 00:32:22.336 align:middle So if we had tests for this, you don't have to look at this. 00:32:22.816 --> 00:32:25.376 align:middle You know it works. 00:32:25.726 --> 00:32:28.946 align:middle It's easier to rewrite things. 00:32:28.946 --> 00:32:50.486 align:middle And running tests means that we can do some fun while they're running. 00:32:50.486 --> 00:33:00.176 align:middle I mean at least for us in PHP world because we don't have compile times, 00:33:01.176 --> 00:33:04.306 align:middle which I'm something that I'm actually missing. 00:33:04.386 --> 00:33:05.926 align:middle But that's a long story. 00:33:07.016 --> 00:33:08.386 align:middle Cleaner code. 00:33:09.596 --> 00:33:15.556 align:middle Trust me, it's really hard to test shitty code. 00:33:16.516 --> 00:33:22.586 align:middle If you write your tests as you're writing your code, you'll will write cleaner code. 00:33:23.746 --> 00:33:32.166 align:middle Because for a 15 line code, 15 lines of code, you might have 300 lines of test. 00:33:32.816 --> 00:33:36.376 align:middle And then you add an if, which means 600. 00:33:40.236 --> 00:33:43.066 align:middle But maybe if you're building a small app, 00:33:43.486 --> 00:33:48.096 align:middle something that you have built often and that you understand. 00:33:49.446 --> 00:33:52.876 align:middle I don't think that testing brings any value. 00:33:53.386 --> 00:34:01.486 align:middle Let's say you are bringing, you're building a one-off app. 00:34:01.756 --> 00:34:04.716 align:middle Again, you can easily manually test it. 00:34:05.446 --> 00:34:06.046 align:middle That's okay. 00:34:06.876 --> 00:34:17.386 align:middle You know it works, but what happens a lot is, that simple app becomes a bigger one 00:34:17.746 --> 00:34:19.256 align:middle and a bigger one and a bigger one. 00:34:20.286 --> 00:34:26.516 align:middle When that one-off app for one concert in Lisbon becomes an app for all the concerts 00:34:26.516 --> 00:34:30.336 align:middle of that group in the world or all the concerts in the world 00:34:30.336 --> 00:34:34.706 align:middle for every group, you can't use the same thinking. 00:34:35.936 --> 00:34:40.446 align:middle When you're exploring something, if you're trying to learn a new language, a new library, 00:34:40.446 --> 00:34:46.106 align:middle a new framework, a new something - don't waste time testing. 00:34:46.726 --> 00:34:50.376 align:middle Enjoy it, learn how to use it. 00:34:50.616 --> 00:34:57.906 align:middle Last but not least: job security. 00:34:59.616 --> 00:35:06.246 align:middle If you're a key person in your company and you have no tests, you cannot be fired. 00:35:08.196 --> 00:35:13.366 align:middle You can ask for whatever you want - you'll get it. 00:35:14.156 --> 00:35:18.886 align:middle I want a new car, I want the best Mac, I want 5 month vacation - you'll get it. 00:35:20.206 --> 00:35:28.786 align:middle What people keep forgetting is that at some point they might look for another job 00:35:29.706 --> 00:35:33.666 align:middle because they're not happy where they are or they have to move to another city 00:35:34.356 --> 00:35:40.276 align:middle or they get the shitty manager because their previous one decided to go somewhere else. 00:35:41.936 --> 00:35:46.606 align:middle You will not find a senior position that says 00:35:47.616 --> 00:35:52.826 align:middle that doesn't have listed testing experience, testing knowledge. 00:35:53.206 --> 00:36:00.526 align:middle I didn't trust people when they started telling me that 6-7 years ago, but for last 3 years, 00:36:00.786 --> 00:36:06.866 align:middle no senior position ad that I saw, came without testing experience. 00:36:07.346 --> 00:36:13.226 align:middle So, it's gonna be really, really hard to move to find a new job 00:36:13.226 --> 00:36:17.916 align:middle and it's just getting worse and worse and worse. 00:36:19.716 --> 00:36:24.646 align:middle So if you're like, my grandma had a few chickens. 00:36:24.996 --> 00:36:29.096 align:middle No fancy lab to test them. 00:36:29.346 --> 00:36:38.126 align:middle Please, don't think that you can use the same principles to have a chicken farm. 00:36:50.706 --> 00:36:54.156 align:middle It's different mindset. 00:36:54.156 --> 00:36:54.576 align:middle Thank you.