10 Burst results for "Arnon Axelrod"

"arnon axelrod" Discussed on .NET Rocks!

.NET Rocks!

02:40 min | 1 year ago

"arnon axelrod" Discussed on .NET Rocks!

"Early enough than by definition, you have full coverage of destination on every feature you ever developed. Yeah. And this coverage is not only just a metric of code coverage. But it's also the coverage of the user stories of the real solution to the real problems of the customers. I get it. I'm wondering how often those user stories end up being expressed this sort of accepted driven test model comes into really describing a business case anyway, and a business value proposition is the metric of a pass yet. Exactly. So if you'll have full coverage and full coverage of very valuable tests, not just technical tests, then you can easily reflect or your coat, and you can get rid more easily of technical debt. Or let's say if you do it from the beginning you can avoid having any technical debt. Yep. And that's what you want all the time. You can continuously reflector your car to make it the best possible design. Every every given moment, and this also prevents bugs when the design is good many bugs caused by design, and it's very hard to keep the design Goodwin. The when the requirements change and requirements always change. So you need to refractory in order to keep your design and your coat Clinton and easy to maintain. And if you don't do it the mo- boxes, but if you do it, and if you have full coverage of of acceptance tests than you can do it very easily and very safely and continuously refectory or called to keep it clean. And with good design. Sounds like a worthy goal tough to get to though. Yeah. Very tough Warren. And we wish you a lot of luck with your book and in the future, and and thanks for spending this hour with us. Thank you before we wrap up I want to to let you know that listeners of the show or religious to discount of my. Book if you go to a percent and the discount code destinations, you get twenty percent discount. Oh, thank you. That's very generous. I'm sure you'll get get some sales off of that. Thank you. And thanks again, aren't an and we'll see you your listener next time on dot net. Ross. Dot net. Rocks is brought to you by Franklin's net and produced by plop studios. A full-service audio video

Warren Ross Clinton Franklin twenty percent
"arnon axelrod" Discussed on .NET Rocks!

.NET Rocks!

05:30 min | 1 year ago

"arnon axelrod" Discussed on .NET Rocks!

"This is almost as far as destination. Maturity can go though, still typically bogs, you know, are just a way of life. But here the teams starting to think in strive towards getting into zero bogs at least zero bug as mindset DC that is actually achievable. It depends. How you how you define it. I think that it's more of a mindset, but I think that it's a Chipper Bill that if you prioritize fixing anybody everybody as quickly as possible over adding new functionality, right? And you also do root cause analysis for everybody and try to make things to prevent similar bugs from occurring. It can be not only from by changing the code. It can be by changing the way, we work some process. Gers or things and the key part. There is causing Alice's not just fixing the bugs. But trying to figure out how we got them in the first place. Yeah. Exactly. So I think it's very challenging in fact, the fifth level, which I'm going to talk about now, I don't know any team that really works that way. Exactly. But I saw teams that they used to work in a team that were was very close to this. So I called the fifth level the encompassing level because here the test automation is not only for testing not only for verifying functionality. But it encompasses the entire development life cycle. And this is where the bitty d we mentioned before are I like to call it. ADD acceptance test driven development, which is very similar methodology. This is where it kicks in in this methodology and mine. Set the team collaborates, let's say typically most developers think that the product owner should describe the specification of the new feature that they won't this is what most developers think of when they think about user story. But here, I think about user story only as the finishing of a problem, you know, the three sentences as a role in order to something I won't LA LA, LA, right? This formats of user stories or do. So I say that the product owner should only ride the to the first two sentences as they're all in order to. That's it what they want is already the solution to the problem. And I think that the team should collaborate on the product owner should come with a problem and discuss with an tire team. What should be the? The best solution to the problem. Sure. Now, I gotta think the guy on the news stories knows what he wants the answers to be. But he's leaving room for other people to present other ideas there so he can suggest his own solution. But her but making it a suggestion is way better than dictating. What should be done right and builds up the team's confidence and groggy of the problem. Exactly. So the team decides core decides collaboratively on the solution. And then they defined acceptance criteria for this solution, right? And when they defined this exception criteria, they defined it in a manner of an example, user example. So the user does this and this and that should be the result. So right. This is already finishing of test. And when it's done collaboratively, it's easier to find gaps in the understanding of different peoples in in the team because when we talk, you know, in general wards, sometimes I can I can use one word to mean one thing that you can use the same or to another thing. And we both show that we both understand each other. But when I go and implemented, the one thing deals think that it's completely different from what you. You you sought. So when you talk about Zampa lls concrete examples, it's easier to to spot this this gaps while I'm getting back to your whole point about the root cause analysis. How often do we find our real root causes? They didn't fully understand the problem. Yeah. Exactly. Yeah. And so going through this process to help them really understand. The problem is is key part of it. You're you're you're stopping bugs with that activity. It'll they'll never get written. Exactly. So now, you'll have the definition of the of the tests you can automate the tests even if the -plication is not written yet. Now, of course, you can decide collaboratively also which which kind of test should I write in order to test. It should it be an end to end test or unit test, and then you implement the test and implement the code that makes this test pass, you know, like. Thity? But the great thing is when you do if you do it from the beginning of a project or.

LA LA Alice LA
"arnon axelrod" Discussed on .NET Rocks!

.NET Rocks!

05:01 min | 1 year ago

"arnon axelrod" Discussed on .NET Rocks!

"And that's what they tend to work on. I who is the person leading your automation effort at this point. Is it a dedicated QA person? Or is it a developer? So it's typically a developer that belongs to the team. It's decimation developer. Sometimes it can be someone who was a manual tester, but has some more programming knowledge typically in this level, you don't use very let's say expensive or a junior developers companies try to hire people more juniors or testers that have some programming background, right? As I know a lot of developers like to automate everything, you know on a. Automated deployments, and and deliveries and automated testing just should be part of that. Yes. Of course, is part of the test of the commission of the of the deployment of the end of the integration and everything, but it requires different skills different mindsets. It's not just automating the process because test are a little bit different than just making a process. So the problems in this QA Centric level is that the maintainability instability are handled, but they're still struggling still a challenge and the collaboration with the developers is is not very good at this point. And in fact, the feedback loop is very long, right? I still sounds. You're repressing the edges of a DevOps conversation. Again, they I want those sorts of cycles. I want people to see things quickly. And I just really am enjoying your approaching this from a test automation perspective, so keep moving next level. Okay. So the next level never is collaboration and this level the decimation developers and the application developers work in the same team or in the same scrums team that list, and they work more closely together. Here a test is failing in the nightly build it should be fixed the next day. But it's typically in this level. These are still different trolls, they're all of the just automating, developer and obligation developers. And they're starting to cooperate. But it's still not full cooperation. They don't often do code reviews to one another in the commission breaks and tested commission. Developer is on vacation. Then maybe they're late for him to return. In order to check what happened and so on and also the test automation that that is developed here is more the end to end tests and less unit or integrations. Sometimes there are unit tests in this level too. But the developers right the unit test and test automation developers. Right. The system test. But they hardly talk to each other about what do you test here? And what do we test here? So they stopped to collaborate. But it takes time until this collaboration really really matures are there particular practices to get them to collaborate better as it's just come with experience. I think that first of all it's an organizational decision to. Working in the same teams that the test automation, developers the developers of applications start working in the same team or at least use the same scrum teams. But also in this level oftentimes the test automation developers. Right. The test automation only for the Fisher that were developed in the previous print, right? So a little bit of like, a mini waterfall. So so regarding bogs, the feedback loop is much smaller. But regarding the development of the test automation itself, they develop it once pre after the developers, and it's still hard to get them to fix or change something in order to make it more testable. Right. And I think that's a very interesting feedback that in having the devs involved in the testing is important because and they realize how hard they're code is the test. Yeah. So the first level is what I called fusion and here the collaboration is really is really tight here. You'll typically see either one team that everyone is developer an engineer, as you mentioned, Microsoft, these days work in this fashion. Yeah. So everyone lens the test right implemented test execute attest and..

developer Microsoft Fisher engineer
"arnon axelrod" Discussed on .NET Rocks!

.NET Rocks!

01:51 min | 1 year ago

"arnon axelrod" Discussed on .NET Rocks!

"These are the stabilisation springs and only then you fix the bugs. Right, right. I get it. The I think we're just sort of walking around this test automation is part of a DevOps practice of doing continuous delivery. That sort of thing I it's very related. I think the DevOps is morally is more applicable to SaaS products, you know, web applications or web service, but not locations in the world where you're not doing installations on client machines. Yes, medical applications that you know, they have to go through a day verifications. And right. So not everything is just a web application that you saw. Out and immediately. Everyone says it then everybody's always on the latest version like you just don't we live in the web world. You just lose track of the fact that that in the deployed world people are running old versions of your software. And you can't stop them, really. Exactly. Yeah. It's an inch. It's an interesting problem just dealing with that. But you know, maybe we should have addressed this right off the bat. What is automation getting me if not going down, this continuous delivery pipeline? Like, why do I want to automate? Okay. So it is about continuous delivery. Though. In those cases, what were you release the product only once every few months, then it's more as an internal process. Like, let's say continuous integration rather than continuous deployment. But it's also very beneficial there because you can work internally where every sprint you add a little bit of value to the product and you'll prioritize to start.

"arnon axelrod" Discussed on .NET Rocks!

.NET Rocks!

05:09 min | 1 year ago

"arnon axelrod" Discussed on .NET Rocks!

"Write my tesla donation sometimes it goes from it comes from a manual tester that takes his own initiatives and decides to start writing test automation, and again his thinks how hard can it be. And he starts and in the beginning. He is that it works and is very happy. And they shows it to his manager and manager is may be very happy. But it doesn't hold for very long because it doesn't have the skills to write maintainable stable test and every little every other day that break and takes longer to stabilize them and to fix them than it takes to test tested manually. So. This is the first level, and I've certainly dealt with organizations were in that place where I literally when v two of the product came out we had to throw away all the tests we written and write him again. Are we writing tests the wrong way, probably? It's point is tests shouldn't be that brittle. Yeah. It shouldn't be that real. It depends on how if the Desta written only in technical way, they test technical aspects of the -plication that if this technology changes, for example than this. This should be re re re written, by the way, one this advantage of smaller tests like unit and integration tests. Where system test are less brittle to this kinds of problems though. Of course, the you I took -nology can change too. Right. But so it's not only it's not only that it's also what the test verify. So if the tests verify the behavior of the -plication, let's say in business terms like eh the product to the inventory and sell the product. And if you if you're using this kind of terms than you can. Change the implementation of some metals in the maybe a lot of message, but eight you change only implemetation of some methods in the destination. But the test themselves can remain more or less the same. Now, we're describing tests that are much more about the intent of software rather than the implementation of it. Yeah. Exactly. Okay. By the way, tools, like cucumber inspect flow which fall under the category. People like to call them biddy details. Right. Oh bit. It is more muscle Aji rather than the tools, but this tools guide you to to ride the tests in that's tool men are that describes the business functionality of depletion rather than the technical details. Of course, the tool itself. Cannot force you to do the right thing. Right. So many people use these tools, but the sentence has the right are very. Technical. So y'all so need to use it correctly, Dan. But I appreciate a tool. It sort of helps you fall into the pit of success here rather than you go off on your own and just go the wrong way. We we've we've talked about cucumber inspect flow on the show before. But I think it's literally been years. Yep. I this is brought it up. So I appreciate you mentioning that. But that moving to the next stage of maturity is writing better tests part of that equation. Okay. So let me go through the other levels. So that level is what we talked about before. Which is the QA Centric where the commission is perceived to be a responsibility of the QA team, which is a separate team in the stage. But in this stage, the test automation itself is more mature, there's at least one dedicated automation developer sometimes team of commission developers. But the perception is that the goal is to find bogs. And when the bug is found than they filed a new bug. And if the bug is critical than any other critical bog, it's fixed immediately. But if the bug is not very critical than you file, a bargain, it can be fixed somewhere in in the future whenever the developers will have time two months from now or right or whatever after this this current sprint like if it's a rendering problem on a minor device. You just not going to prioritize it. Yeah. But even it's in the next print. It's it's better. But in some cases, you know, many many companies still work in water scrum way. Scrummed fall. But you release the software to the customer only once every six months, for example, scrummage so so the only the few sprints towards the release..

Dan developer six months two months
"arnon axelrod" Discussed on .NET Rocks!

.NET Rocks!

03:52 min | 1 year ago

"arnon axelrod" Discussed on .NET Rocks!

"And let alone start writing the code. You'll have. To look at the what up the -cation doesn't what's important to test than look the architecture. You also have to look at who's going to write and maintain the tests, this they are I think most of the more important things, then you need to decide about the architecture of the tests and out of the tests interact with application, you can have one solution for this. But sometimes you'll have different kinds of tests for different problems. So as I said, you can have both unit tests integration tests and system tests, for example, the this is the first things that you have to consider before you even start considering the the tools and. And how to write the code, right? Do you find there's architectures software architectures that are resistant testing are very hard to test yet. If the -plication is written in. I want to say monolithic, though, today would most say monolithic. They the think about something that is not micro services. But right as I said as I said before it can be a monolith. In terms of micro services. But it can be very modal are in the way that it's built so application that are really written as one, you know, spaghetti of dependencies forms and. Code behind. Something that you know, you it's hard to break to to smaller pieces. And also if the U I took -nology is not not all your technologies are very friendly to you, I automated so in this case as it can be very hard Yana. And and and you see folks using that as an excuse to not get into testing because it's so brutal and so difficult, and it can't get to the depth of the action design was going on there. Does that put the onus for writing tests on the devs or diesel see separate QA departments? These days. It seems like Microsoft's gone away from there. Yeah. I still I still see see those teams the less. I gave a talk both here and Israel and in Lithuania about destination. Maturity, oh, I got five levels. And one of these levers is separate QA teams from. The development, but when I started to work as a consultant for years ago, it was much more prevalent, and this days it's much less. So I see more and more companies going towards the unified teams. And sometimes into teams were everyone does everything the developers participate in the specification phase. And of course, the the design architectures right called they also test their on code, and they also monitor what happens in production and see the actual results. Let's dig into these five maturity levels of organizations that are adopting test automation. So the first level I call it the nave level because the main question here is how hard can it be some teams that before the the start using decimation? They either think sometimes it's manager thinks I'll just send my QA team to learn a little bit of programming and will teach them to let's say selenium, and they.

Microsoft consultant Lithuania Israel
"arnon axelrod" Discussed on .NET Rocks!

.NET Rocks!

01:54 min | 1 year ago

"arnon axelrod" Discussed on .NET Rocks!

"You you can query the layouts and ask. Were the position of things are, but you you normally don't do that. It's not very effective to to do that. So if about in exist, but it's not very it's not big enough. Or it's not the text is not clear, or even if the text is truncated because of the size of the button up Hume and selenium one tell you this things. Okay. There are other tools of visual testing that do that. But there are more oriented to our towards comparing images against a previous images, which has to some extent. But it also has drawback that in order to maintain the images of the all versions. It can be a lot of work. If something changes in the especially if something changes through the UI across the board. All those images have to be reshot. Then you know, you change the phone. In all of the screens. Then you'll have to re record again, the take the snapshots again are there any considerations for testing micro services that might not seem obvious. Typically, you test it by sending rest calls and get the responses and asserts that you received what you want in the response, but it depends on -plication the architectural delegation, and again what you want to test. I have also one of the chapters in my book, one of the bigger chapters is talking about the architecture the relationships between test automation and architecture of the nation. So you'll have to consider your.

"arnon axelrod" Discussed on .NET Rocks!

.NET Rocks!

04:46 min | 1 year ago

"arnon axelrod" Discussed on .NET Rocks!

"Enjoy your new music to code by for writing that already have. And they used it when I wrote the wrote the book. Oh, no kidding nice. Wow. That's so Hello to even mentioned it in the colledge meant, I mentioned you and the on the musical by the college, man. Cool. Thank you. So I guess the whole idea of test automation is just to make it invisible. To the developer as invisible as possible is it or are we still requiring developers to write their own tests? I think that developers should participate in writing the tests in sometimes developers. Right. The test themselves, and I think it's excellent. In some other teams are different people that write the test from the people that write the code are different butter organizational patterns regarding the who writes the test automation. I also have the chapter about it on my buco. So which I think we didn't introduce my book formerly now, let's do that. Okay. So I just throw the book called complete guide to test automation published by April, so very comprehensive book five hundred pages. Everything almost everything that I know about sorta commission in fact for everyone who's interested. It has to mention in anyway, which should be anyone in the development team. This includes developers the testers just donation, developers, of course, managers architects name it. The first part of the book is not very technical. It's more, you know, high level, I it focused more on the law, and the why questions rather than the how and the second part of the book is more technical where I show step by step like tutorial for how to go about designing and writing new testing solution. How right the call and outdoor refractory it because I think that these are very important things when you when you rights destination, it's important to write it in a way that is very easy to maintain and is very reliable. I'm glad that you brought up happy which is a framework for testing mobile apps. There are ways to automate that stuff in the cloud and automate things like happy, and do you think it's necessarily use these cloud based testing farms are can we get by with running local tests on mobile devices. Okay. So. It depends. What you what you want to test? And why I think that the first cycle of testing should be more functional tests and most of what I write in the book is about functional tests in order to write and to run functional debts. You don't need to run it on every device. Most functionality is should be identical. And also, you don't have to test everything through the UI you can test most of the things sway PI or through component test or even unit tests. So most of the functionality don't have to be tested on all devices. But after you have your functional tests bus sometimes you want to test that it works correctly on many other devices than you than it's beneficial though. It's important to remember that most iota mation tools up human being. One of them don't actually test that the UI looks good looks right. And is usable. It's not a replacement for usability testing. Right. So you'll have to keep that in mind. But I guess that there are things that can work work good in one device and don't work well on another device in the functional level either. Yeah. Is that what happens going to show up for you is that this is not rendering correctly on this device? No, so appeal which is very much an extension to selenium which behaves more or less the same only that selenium works on the browser and up him works on a mobile on mobile devices. They reflect what's the elements that are shown on the screen, but not their layout..

developer development team
"arnon axelrod" Discussed on .NET Rocks!

.NET Rocks!

04:12 min | 1 year ago

"arnon axelrod" Discussed on .NET Rocks!

"Ed, you or just Google or Bing web. Gazer Dutch AS you can actually do a real time demo, and I would make sure that it's the lights are on. That's all going to say. I got. Yeah. And got to think wearing glasses must complicate things. Like, it's not a trace trivial thing to to track is indeed. But cool. Yeah, it's very cool. So, you know, real time gays prediction on most major browsers. It does self calibration from clicks and cursor movements basically doing that presuming that you're looking where your mouth is point. Yeah. You're is gonna fall your mouse. Exactly. That makes us coupla is a Java script, you got it that whole I tracking thing was really about. What are the hot spots on your site? We'll seeing your ads now that's a great idea. If you can use it in correlation with your pages to sort of map where people's eyes are going. That's a great idea. Yeah. Our friend Kent all stat has a new companies been working on for a while. Now cloud army, which is specifically about applying Niro science to understanding people's behavior. And he and he uses it for market analysis for doing things like understanding whether people care about a brand or recognized packaging things like that. So in rather than serving directly this is actually studying people's. Savior rounded. It's it's complicated. Maybe we should have him on to talk about how we should. But he's big on the eye gazing thing, you think it's where people look it says a lot about what they're actually thinking. Oh, very good. Talk to him about that. All right. So that's what I got. What do you got grabbed come off the show? Fifteen eighty six from October two thousand eighteen we were talking to Patricia asked about reading other people's code, right? I think absolutely correlates with testing and this great comment, which a few months ago comes from one Arnon Axelrod never heard of him. Never heard me. This is great show is always Patricia gives some great tips and techniques for reading someone else's code. However, there's one key thing that I felt was missing for the conversation when you read someone else's code. There should be a reason and go forward, and that this should be kept in mind focusing your efforts and the overall shape of how you go about reading that code, for example, reading old coat of someone who left the company in order to make small change to is radically different from performing code review to a colleague on a new feature. Hey signs that I totally agree and often claim that myself that writing code is primarily a means for communicating between people sometimes between you and your self at some time in the future. And only then a means to program the computer, otherwise, we would just write our programs ones zeros. Yep. Which which true? Yeah. You know, these languages that we use? They're just giving ourselves different ways to express our intent, which certainly the computer correspond to. But also other people reading the cone. Yeah. The thanks Arnon copies. Dako- buys on its way to you. And if you'd like a copy music, oh by write a comment on the website at Donna Iraq's dot com or on Facebook. We publish every show there twice a week. And if you come in there, and we read in the show, we'll send you a copy music. Oh, by and definitely follow us on Twitter. He's at rich Campbell. I'm at Carl Franklin, send us a tweet because we we look at them. So. Yep. Read a comment from our guest. That's right here. He is on an Axelrod. He's a senior consultant and leads the test automated division at Salem group aren't in started coding in the eighties on a spectrum when he was ten and hasn't stopped ever. Since in the last decade specializes in everything related to test automation. And for him everything in the development life cycle is related to test automation. So this includes unit testing integration tests, new I automate automation, tools, including selenium happy spec. Flow and many more the best practices for developing test automation, TD clean, code factoring. Devops agile methodologies and processes.

Arnon Axelrod senior consultant Patricia Ed Google Bing Carl Franklin Twitter Kent rich Campbell Facebook Dako Donna Iraq Salem group
"arnon axelrod" Discussed on .NET Rocks!

.NET Rocks!

02:46 min | 1 year ago

"arnon axelrod" Discussed on .NET Rocks!

"Is Carl Franklin, and this is Richard Campbell. And hey, buddy, guess what? I gave up today would did you give up today friend? I'll maybe not today maybe last week, but coughing caffeinated coffee really I have to drink decaf now. Okay. I found it was spiking my blood sugar. Interesting the caffeine could spike your blood sugar. So I don't know if it's just caffeine or if it's the amount of caffeine, or if it was her titular brand of coffee, but all I know is I stopped drinking caffeinated coffee and my blood sugar went down to normal. It was as you know, been type two diabetic for a while. But I don't have the symptoms of type two diabetes anymore. Right, right. I have this continuous glucose monitor that goes in your arm, and has a little device, and it takes a sample of your blood sugar every minute or so and then when you put the. Device in near it. It transfers all the data points to it. So that you can see it in a graph. Right. And I found that normally my blood sugar is well under a hundred eighties nineties, sometimes even seventies and milligrams per deciliter. And then all I fasted for a day. And then the next day. I drank coffee. That's all I did. I just drank some half decaf, maybe four or five cups by the middle of the day was one hundred and fifty eight. It's crazy. So I don't know if the caffeine was perhaps adding stress in that was spiking up. But all I know is that I switch to decaf and everything's fine. So I'm getting old, my friend. Yeah. It's interesting. Because of course, there's arguments that that black hoffy even pulls your blood sugar down to. Yeah. That's I've heard that argument to. Yeah. But you've got data for you. That's all a matter. It's all the matters. Right. Yeah. Yeah. I don't know fasting effects that fasting. If you're just fasting with water and salt your blood sugar should drop. Yes. Yeah. And it did. But then. Yeah. It's just one interesting thing whether fasting, but your body particular state where sugars reveal he's not like your sugars and coffee now. No it though, like what it wears those sugars coming from that your budget goes up, anyway, interesting stuff. Interesting stuff and data. All I've got is my dog turned fourteen. Do you have a party bear hunter man now? Well, I thought I would bring up that testing because we're talking test automation with aren't an Axelrod. But first, let's roll the crazy music for better. No framework. What do you got? This is a very cool project. That's called web gazer.

caffeine Carl Franklin diabetes Richard Campbell Axelrod five cups