Moscow Python Conf ++ in three days: what will Bobuk talk about and who will perform in parallel halls at this time?

    Report preparation experiment at Moscow Python Conf ++ from scratch on the finish line. The slides are ready, the runs were carried out, it remains only to wait for the premiere - this Friday, April 5th. The  schedule includes 24 reports on the use of Python in the production of large companies. Under the cut - a look at the organization through the eyes of the program committee and ridiculous problems like “who should be put in parallel with the headliner so that there are no empty rooms and offended speakers”.

    You ask, why do you need this experience, if you have another “one hundred thousand million” users, you can do without a zoo in microservices, and from ML a fairly simple linear regression is enough. Answer: mainly to reflect. And then, without leaving the place, ask cool developers, talk with other participants about similar problems and how to solve them, get to know interesting ideas and make useful contacts. That’s how I imagine it.

    I managed to find like-minded people - another 8 members of the Program Committee together with me formed important topics for discussion. In total, there were more than 100 specially found reports and submitted through the traditional Call for Papers. We all carefully studied them, asked additional questions and put maximum usefulness into the one-day program. We’ll find out exactly how it happened on Friday, but for now I’ll tell you about some of the reports that I personally supervised and in which I’m already oriented, almost as a co-author.

    Python Governance

    Of course, it would be unrealistic to get Guido van Rossum to visit. I tried, and I hope that one day it will work out, but this year we will have Core developer Python and author of the popular six library Benjamin Peterson . He is directly involved in the development of the language and  will tell us first hand about the new development management model. The very one that the community chose after Guido Van Rossum’s refusal to act as a “dictator” and rule everything individually.
    Let's talk about what this means for us, as users of the language, how releases, support, standard libraries will change, and what the future holds for Python.

    OpenStack - 12 million line Python project

    For many years, Vadim Ponomarev talks about certain aspects of working with OpenStack. But at our conference, for the first time, he will talk about the largest Python project from an unexpected angle: how is it actually being developed? Indeed, during the life of the project, it has grown with many internal "bicycles", development techniques and solutions for "problems of a huge amount of code." Vadim will give us a brief excursion into the history of the evolution of the project and tell you how the developers manage to maintain, improve, develop the project, which is not only that under no circumstances can fit in the head of one person - life is not enough to read.

    Predictably Good Microservices

    At CIAN they love, are able and practice microservices. And the business was organized in such a way that you need a lot of microservices and often: in C #, JS and Python. Over the years, the team singled out everything in common that is created in the microservices, and learned how to create universal “workpieces” with the minimum necessary set of everything. Oleg Ermakov will tell uswhat and how was included in this “gentleman's set”: what does the “blank” of microservice consist of, how microservices are deployed and automatically configured, which offers developers an internal cian-core library. Most of the talk is about working with the API. CIAN made their own DSL to describe the interfaces between microservices, for which there are clients, generators and validators. All this allows you to build APIs very quickly and automatically check that microservices speak the same language without errors.

    Async and await in production

    There are more than 50 Python developers at HomeClick and a modern pipeline with Python 3.6 and asyncio. Asynchronous programming has been used from the very beginning of the projects, so the teams were able to evaluate not only the benefits, but also the complexity. Unexpectedly, developers are running out of memory, no exceptions are caught, and the available "asynchronous" libraries for typical tasks are often very raw. Add to this not the easiest debugging and hiring situation. Sergey Borisov , technical manager of the Content team at DomKlik, will tell us about all this .

    How Python developed in Yandex

    About 12 years ago, Ivan Sagalaev came to Yandex and started doing services on Django. In those days, the company already used Python, but there were no major projects on it. When the first service, “Where Everybody Goes” (part of Yandex.Afisha) was done, the experiment was considered successful, Python and Django were suitable, and the stack was used in other projects. The number of services grew, standard libraries and approaches began to appear, some of which are still in use. About the decade of evolution of Python will tell Alexander Koshelev , head of the Python-development service of the department of services for organizations.

    How to create ten applications from one

    The IPONWEB developers have almost realized the programmer’s dream: over and over again they make very similar admins for managing their advertising companies for their clients. The nuance is that the "old" admins do not disappear anywhere, therefore all bug fixes and improvements of the "new" ones should be applied to them. Anna Moshkina will tell how they in the company made themselves the engine of “dynamic applications”. The client admin is described in JSON, which is then used by the Django backend and Angular frontend to create the application on the fly. In addition to the obvious advantages, this approach has difficulties, especially with testing and the database: ad arbitration brings its own specifics, and the company's clients manipulate through the admin area a large number of complex entities that are called "creatives" inside.

    Making a Dashboard for an airline: Dash and more

    The airline, in this case, S 7 , which has many developers, including Python, is already interesting in itself. But it’s even more interesting that the airline has a lot of different data, it’s complicated and it needs to be shown to different people with different goals. For this task, the Python ecosystem has, for example, Grafana and Dash, but are these really the best solutions? Nikolai Fominykh in his report will show the results of a large-scale study of possible solutions and talk about the practice of their use.

    Go vs python

    Yes, you read it right. One of the most experienced Go-developers and organizer of Go-meetings in St. Petersburg Vitaly Levchenko will come to us and talk about the strengths and weaknesses of Python and Go! This will be a hardcore comparison from the person who led the development of the MegaFon.TV backend. And so that Python doesn’t disappoint part of the report, Vitaly is actively helped by the program committee, which consists entirely of Python developers.

    Avito Prices: How We Manage a Billion Values

    It turns out that all prices for Avito are selected programmatically. Python has built an entire infrastructure that combines a backend, a team of analysts, Machine Learning and a huge data matrix with billions of prices. Dmitry Kliminsky will tell you how this infrastructure works , how useful Python is in this case, what difficulties you have to deal with, and how to automatically test that prices are all right.

    ML in the search and recommendations

    Traditionally, Machine Learning is used to train any models on historical data, but when you enter a search query on, the recommendation system has to rely on these dynamic data. For this to work, it took several unobvious tricks, which Igor Kitsenko will tell us about . The second interesting moment of this story is that ML is made in Python, but search in Java (Lucene). We learn from the report how HH managed to make these two systems friends.

    Secret report of Grigory Bakunov

    Bobuk has been programming in Python for much longer than I have since the late 90s. And yes, he found the transition from the 1st to the 2nd version, no matter how crazy it sounded. I'm not very good at listing regalia, everything can be found in Yandex. I can only say that it was Grigory who brought the first Python programmers to the company 12 years ago. The report will be about a mega popular myth: "Python is slow, write in Go or Java." I will not spoil the content, because I myself do not know, but as the head of the PC I will say: the report is an occasion to talk. We are going to the conference not to learn something in 30 minutes of the report or to learn something new not in the news feed. We come to talk with developers who have been using technologies close to us behind closed doors of large companies for years. Discuss over a cup of tea and a glass of beer that they will not write on Habré and will not answer on stackoverflow.

    Where is lunch going?

    New trends in the manufacture of conferences - "floating" lunch. It starts at one o’clock in the afternoon: the first room closes for an hour, at this time someone can have lunch, someone can go to the second and third rooms for presentations. And after an hour we close the 2nd and 3rd halls, continue lunch and open the 1st hall. And yes, it was in this slot that we placed the report of Gregory. He will have a very difficult task to compete with the second lunch: but something tells me that he will manage :)

    Bonus: On the pages of the conference in  social networks I talk about these and several other reports on camera, and speaker Vladimir Puzakov from Rambler & Co invites you to meet Mypy closer.

    I invite everyone for whom Python is one of the main development tools , it will be interesting to visit  Moscow Python Conf ++ on April 5 in Infospace. If you missed all the announcements and no longer have time to participate in person, there is a paid video broadcast or the option to wait six months until we post the video for free access. Or sign up for our  newsletter so you don’t miss an ad the next time.

    Also popular now: