DotNext 2016 Moscow: Calm after the storm



    In June, when the DotNext conference was held in St. Petersburg, the .NET world was experiencing a difficult moment: Microsoft suddenly changed a lot in .NET Core already at the release candidate stage. Now that the Moscow version of the conference has passed, you can compare them and understand: how have the past six months affected the situation? And what Moscow reports were devoted not to the latest shocks, but to “eternal problems”? A review has already appeared on Habré through the eyes of the viewer, and now we are publishing our text.



    The top 5 reports from the previous DotNext included a performance by Dino Espositojust about the advent (ASP) of .NET Core. There, he talked loudly and sarcastically about how, in the pursuit of revolutionary changes at Microsoft, they forgot about “that crazy little thing called legacy code”. Now, in the opening keynote, Dino talked about the same topic again, but his tone was calmer, and there were more diagrams on the slides: several months have already passed since the release 1.0, and it's time to seriously understand what the migration to Core means. However, he could not completely avoid the fuss: “Of course, I love system.web - I earned a living by writing about it!” So this is an old friend, but now he is leaving us. ”



    Then the theme of .NET Core (as well as .NET Standard) was picked up by Alex Thyssen(Xpirit). He made the reservation “Core has time to change in the time that you look at it” - but this did not stop him from telling a lot about what you can see now, starting with leaving project.json and ending with what to expect from .NET Standard 2.0.



    At first, the conference could be confused in the names of the speakers: this time, in addition to the long-loved DotNext viewers Dino Esposito and Sasha Goldstein, Dina Goldstein was also present(Aternity). She talked about Event Tracing for Windows, starting with the difficulty with traditional profilers: “They require recompilation or restart - just try restarting software at the airport just like that! And there are also difficulties with overhead and licensing. ” In general, the audience liked the report, but several of them immediately indicated in the reviews that something was missing: they wanted to see not only screenshots and code, but also a live demo.



    Dmitry Soshnikov (Microsoft) already briefly touched on the topic of Microsoft Cognitive Services in his keynote in St. Petersburg DotNext, and now has devoted his entire report to her and the Microsoft Bot Framework. Although the company is now known for non-AI technologies, it makes a big bet on them.. So even for those who are not connected with this right now, it’s useful to keep track of which “cognitive APIs” it provides . Of course, if you want to follow this topic, you can also follow the official news - but many viewers in their reviews noted the bright presentation of Dmitry, and this is not found in the news.



    In a recent interview with Andrei DreamWalker, Akinshin (JetBrains) can easily see how excited he is about the Rider project he is working on. And at the opening of the conference, he also said this: “I wake up in the morning, immediately open the Rider and start to do something.” But his report was devoted to a completely different one: after the Petersburg presentation on the intricacies of arithmetic operations was very warmly received, he developed the theme in Moscow.

    Do all these subtleties have practical significance in the life of the developer, or are these nuances that are interesting to know about, but which really never affect? Andrey began the report with several examples of how they affected very significantly. Probably the most impressive example is the story of how, in 1991, a “small and nothing affecting” error in the anti-aircraft missile system software accumulated up to a third of a second error in 100 hours of operation, and as a result 28 soldiers died.



    At Marco Chekkoni (Stack Overflow) were two reports per day. And if everything is clear with the theme of the first “Stack Overflow - It's all about performance!” (He also told us about it before the conference), then “Performance tuning Stack Overflow tags” could surprise: what, tags on the site is such a significant and self-sufficient topic from the point of view of performance that a separate report should be devoted to it? But Chekkoni quickly made it clear that there were enough pitfalls: “For example, in our search, the tag can be set not only as windows, but also as windows *, so that everything like windows-10 can be found with it. And we have such an audience that they really often use this opportunity! ”It’s interesting, you won’t guess in advance that the“ technical advancement ”of the site’s audience can affect the work with its performance.



    Sasha Goldstein(Sela Group) also performed twice - with "WinDbg Superpowers for .NET Developers" and "Squeezing the Hardware to Make Performance Juice". As you know, “software is hard, but hardware is harder”, so Sasha’s ability to explain everything with simple words was especially useful in the second case.

    Developers may feel “if I’m not launching rockets, then why should I bother with iron questions”, and the report had a very clear illustration of why: the question with Stack Overflow “Why is it faster to process a sorted array than an unsorted array? " . Due to how branch prediction works (which the author did not know about), the difference in processing time for him was six times, and when the differences are so, it’s clearly helpful to understand how things work. From this example, Sasha moved on to more complex topics like SIMD, while maintaining the availability of presentation.



    Finally, Jesse Liberty (Liberty Associates), who also spoke twice, talked about C # 7 innovations in the final keynote. It was mostly straightforward, informative, but at times Liberty sidestepped (“how do you pronounce tuple -“ tuple ”or“ tupl ” ? "), And sometimes he allowed himself liberties (" out variables - they ... can you say "suck" here? "). Among what he talked about was pattern matching (“a switch expression can have a pattern as a case!”), And here it is interesting to compare the C # world with Java developing in parallel. They also started talking recently.about the possible addition of pattern matching - but so far with the proviso “it is written with a pitchfork on water”.


    And now, when enough time has passed from the conference, you can calculate the top 5 reports according to estimates sent by the audience. And he turned out like this:

    1. Sasha Goldstein (Sela Group) - WinDbg Superpowers for .NET Developers
    2. Andrey Akinshin (JetBrains) - We continue to talk about arithmetic
    3. Sasha Goldstein (Sela Group) - Squeezing the Hardware to Make Performance Juice
    4. Roman Nevolin (EPAM) - F # to the Glory of Data Science
    5. Dmitry Soshnikov (Microsoft) - Intelligent Chatbots and Cognitive Services

    It is curious to compare it with the previous Petersburg top . First, of the four speakers on the list, three were leaders in the past DotNext. And secondly, Sasha Goldstein again took first and third places at the same time!

    It seems that after all the twists and turns with the release of Core, now the .NET world wants stability.


    Also popular now: