Design Testing: Top 10 Heisenbug 2018 Piter Reports

    Hello! We opened the video of the reports Heisenbug 2018 Piter. And especially for Habr they made a selection of ten best reports according to the visitors of the conference - specialists in the field of testing. The most liked report suddenly turned out to be the most "offtopic"!

    Reports in the collection are arranged by ascending rating. But this does not mean that the “younger ones” are much worse: everyone, with the exception of the leaders, has approximately the same score from 4.27 to 4.52. Therefore, as usual, you need to watch everything. Meet under the cut!

    Enterprise Automation with Selenium and Selenium

    Speaker: Michael Palotas
    Location: 10
    Rating: 4.3 ± 0.1
    Presentation of the report

    The top opens with a talk by the creator of Selenium Grid, Michael Palotas. Michael was responsible for testing in eBay, came up with new engineering practices and managed to work in Intel, Ericsson and other companies.

    Michael is not only talking about the Selenium automation tool itself. He rightly notes that Selenium is “less painful” in automation and testing, and gives many practical examples of how the introduction of the tool turned into work on a full-fledged large-scale project that needs to be accompanied and managed.

    Michael reveals the main problems that prevent development teams from creating scalable and reliable solutions using Selenium, and shows neat and cost-effective ways to achieve complete test automation.

    Is there auto test in mobile video games?

    Speaker: Dmitry Alekseev / Evgeny Shumakov
    Location: 9
    Rating: 4.3 ± 0.1
    Presentation of the report

    In the past Heisenbug, Phillip Keks has already touched on the topic of auto-testing in mobile games, but in his case the game was with a very simple gameplay. In the report of Dmitry and Eugene from Zeptolab, it's not at all easy: played Cut the Rope or King of Thieves? How to add autotests to them, if all players have different devices, there are no frameworks, and how to track bugs?

    The report of Dmitry and Eugene demonstrates that in the development and testing nothing is impossible. Testers from Zeptolab came up with a rather tricky way to abstract coordinates and dump graphic scenes by clicking on Appium in tests. The report is easy to understand, even for people from another area, and it shows well at what stages it is possible to save time and efforts of developers in mobile game development.

    JUnit, give me five! Porting code to JUnit 5 Extensions

    Speaker: Dmitry Tuchs
    Place: 8
    Rating: 4.3 ± 0.1
    Presentation of the report

    Dmitry Tuchs confidently declares: JUnit is designed for any tests. Just appeared JUnit 5, which received a new code base, architecture and API, and the simplicity and expressiveness of the framework were not affected.

    In the report, Dmitry clearly demonstrates not only the migration process from the previous version of JUnit (just replacing annotations!), But also the different testing styles that JUnit 5 supports, and answers the question - what is the point of switching to a new framework?

    The report is useful to all Java testers who are engaged in tests on large-scale web projects, write an AAA-style functionary (Arrange - Act - Assert, one of the A’s at the end of the report is no longer needed), and want to create simple APIs for beginners to work with test coatings.

    Testing based on Petri nets

    Speaker: Alexey Rodionov
    Location: 7
    Rating: 4.35 ± 0.05
    Presentation of the report

    Imagine that your tests do not know how to find errors that occur in unusual states, and it is already impossible to create more and more tests, because the runtime exceeds all possible limits.

    What to do? Go to the mathematical apparatus in search of alternative ways to develop tests using graphs. The program committee called it "Testing 2.0".

    A hardcore, full Ruby-code report from Alexey Rodionov about how Toptal passed about conventional tests to testing based on mathematical models, what is good and bad can be found on the way, and why you should pay attention to the Petri net to optimize testing.

    When speed and scaling are needed: a server of distributed iOS devices

    Speaker: Nikolay Abalov
    Location: 6
    Rating: 4.4 ± 0.2
    Presentation of the report

    UI test developers may be familiar with the problem of test runs on iOS. Nicholas cites the example of Badoo - when I started to prepare the report there were 1,200 end-to-end tests. When completed - 1300. At the conference, the number of tests increased to 1,400. This is 35-40 hours of machine time in the simulator, or 1.5 hours of real time.

    In the report, Nikolay tells how to reduce the time for passing the tests to 30 minutes by going to the device server and how this made the infrastructure and tests easier to scale and maintain. Nikolay talks about how to “unravel” a node from tests and infrastructure, and learn how to run processes in parallel using a new paralleling model. For this report we made a text version on Habré , so that it can be not only viewed, but also read.

    Finally - half-comic advice from Nikolai: if you need to reduce the time for passing the tests, simply remove the part. And the time will be reduced, and there will be no more unreliable tests, and it's easy to scale! If you need more seriously - see the report itself.

    Testing configuration for Java developers: practical experience

    Speaker: Ruslan Cheremin
    Location: 5
    Rating: 4.4 ± 0.1
    Presentation of the report

    At one of the past Heisenbug conferences, Andrey Satarin told how to cover not only the code, but also the configuration with tests. Ruslan Cheremin worked as a team with Andrey and was inspired to use this approach for his own purposes.

    Ruslan tells in an accessible form what can be considered a configuration (everything!), How to get rid of embarrassment from writing a configuration test, and why it is important, useful and quite simple. Excellent report with simple examples, multiple code inserts and an accessible explanation of what is happening.

    Testers as their own worst enemies

    Speaker: Michael Bolton
    Location: 4
    Rating: 4.46 ± 0.07
    Presentation of the report

    Legendary Michael Bolton with the final keyout, which every tester should look at.

    He will not talk about methods, tools, frameworks and much more. Michael talks about the essence of the tester, his role in the IT world, the importance of the profession and interaction with people, not applications. Testing is not about tests. Testing - about people.

    Michael reveals the problems of the profession of testers, suggests how to develop professional, social and mental skills that will not only increase the effectiveness of a specialist, but also respect among colleagues. Very cheerful, sincere and important report.

    Are you still sawing your report? Then we go to you!

    Speaker: Artem Eroshenko
    Location: 3
    Rating: 4.52 ± 0.06
    Presentation of the report

    The main slogan of the report is “What problem are we solving?”. Artem clearly and on the shelves describes the changes in the long-awaited version of Allure 3 and explains why we need new features - visualization, new tools, a single config and much more.

    The report is simple, interesting and visual and will be useful both to those who have long been "sitting" on Allure, and are not familiar with reports of this type.

    Fuzzing testing: looking for bugs in the JIT compiler and not only

    Speaker: Maxim Kazantsev
    Location: 2
    Rating: 4.6 ± 0.1
    Presentation of the report

    There is a problem. People may not realize that the bug in the application may be related to the compiler, and it is difficult to predict the probability of bugs appearing in the compiler itself, and finding the bug and fixing it is even more difficult.

    If in some reports the collections agitated for reducing the number of tests, then in testing compilers its own atmosphere and its own rules. Maxim Kazantsev from Azul Systems tells in the second place report how to simplify life both for those who work with compilers and in completely different directions using Fuzzing testing.

    It's simple: if one “good” test finds a bug with a probability of 10−6, then it does not find a bug with a probability of 0.999999. And five million tests will NOT find a bug with a probability of 0.9999995000000 ≈ 0.007. So, there is a bug with a probability of more than 99%!

    This is a kind of testing in which millions of random tests are generated that run through the entire project, mindlessly checking everything that comes up. And, oddly enough, this method perfectly helps to find problems where you need both speed and a high degree of reliability.

    Hardcore, quality report with code examples. Be sure to look at least because of a non-standard and interesting way to look for (and find!) Problems in the code.

    We test to the last: smart responsive interface design patterns

    Speaker: Vitaly Friedman
    Location: 1
    Rating: 4.72 ± 0.06
    Presentation of the report

    And here is the leader of our short-list, which, oddly enough, is not about testing. Vitaly Friedman, who is well known among web designers and web developers, then spoke to an atypical audience for himself - and won her over!

    Vitaly systematically goes through all the stages of UX and tells in detail about the interface components and the problems that are associated with them and can be used in testing. These include the implementation of “carousels” in various countries, and tips for creating really convenient comparisons of the characteristics of goods (and not as always), and a checklist useful in creating an “accordion”. Many viewers said this: “Yes, this is not about testing, but it was amazing.” Enjoy watching!

    And for those who are few dozens, we give a link to the playlistwhere there are other performances with Heisenbug 2018 Piter.

    If you are interested in the reports, please note: On December 6-7, Heisenbug 2018 Moscow will take place, where Aleksey Barantsev, the developer of Selenium WebDriver, who has been involved in testing since 1994, will arrive.

    The latest information about the program can always be seen on the conference website , and there you can buy tickets (the price of which is gradually increasing).

    Also popular now: