C++20 is now a year old. Time to take a closer look at the benefits you get when using C++ for standards 11 to 20. You learn about Coroutines by building a coroutine-based parser. The new ranges and the spaceship operator help you write less code. You will learn how Concepts help you to express constraints better and improve error messages. Doing things at compile-time saves run-time. Let‘s see how C++20 improves your code with the new features consteval and constinit. After this talk, you learned some C++20 features and saw the improvements to C++17 code.

Target Audience: Developers
Prerequisites: Knowledge about at least C++11
Level: Advanced       

Extended Abstract
C++20 is now a year old. The major compilers already provide a solid support for the new standard. It is time to take a closer look and figure out the benefits you get when using C++20. Because there are so many improvements in C++20, we will focus on: Concepts, Coroutines, ranges, the spaceship operator, and compile-time evaluation.
Coroutines are the feature that likely will have a strong impact on future finite state machine code. In a brief tour, we will see what we need to do for building a coroutine-based byte-stream parser and how this code is so much more beautiful than without Coroutines.
Less code is always a good thing. The new ranges part of the STL helps us achieve this goal. We will take a brief look at how ranges simplify our code.
Speaking about less code, how about writing a class with all comparison operations without having to write an awful amount of boilerplate code? Sounds good? Then let's see how C++20 helps us there with the spaceship operator.
And while we are speaking about less code, generic programming comes to mind. Concepts give us a whole new way to express constraints and requirements in generic code. Concepts make our code better readable while, at the same time, give us way better error messages.
We don't stop there. Together we will look at the updates to the compile-time world. With consteval and constinit we have two new keywords allowing us new things to do. A lot of former restrictions of constexpr functions were dropped as well. The most popular is probably the ability to dynamically allocate memory at compile-time. We will explore these features using some practical examples.
After this talk, you will have seen the probably most impactful features of C++20. You learned by examples how to use them. The comparisons to pre-C++20 code give you a good understanding of what your benefits are by using C++20.

C++ class design has been error prone since the beginning, until Scott Meyers told us about the Rule-of-Three. But even that is not enforced by the language. In addition, C++11 changed the set of compiler-provided special member functions and the intrinsic rules have become more complicated. However, instead of starting from the rules for special member functions, we will look at what role a class type plays and how that maps on what special member functions to define and how, if any. Examples for class roles are Value, Relationship, Manager (scoped, unique, general), Mix-in.  

We will also look at the roles of member variable types that influence or imply the role of the class type. The role of a class will also take out many mysteries of move operations and will lead to clear guidance when and how to provide move operations: for suppressing copying, for managing unique ownership (Unique Manager) or for optimizing object copies (General Manager). We will rehearse the Rule of Zero and learn more about the Rule of Five/Six, the Rule of DesDeMovA, and the Rule of Unique Three. This all will enable you to much more consciously and safely design your class types and combine types of member variables. You will have then the ability to select from a few reasonable combinations from the plethora of possibilities of special member function combinations: {public:,protected:,private:} x {destructor, default constructor, copy constructor, copy assignment, move constructor, assignment} x {noexcept(true),noexcept(false)} x { =default, =delete, {/body/}, not declared } (math as homework)

Target Audience: Developers
Prerequisites: C++ class design
Level: Expert

For many embedded C++ applications, compliance with the AUTOSAR or Misra rules is required. Among them is AUTOSAR Rule A18-5-5 which does not allow memory allocations with new. Since new and delete violate A18-5-5, the default STL containers must not be used in applications requiring AUTOSAR compliance. This holds for many embedded applications. With the allocators available since C++17 in the namespace std::pmr (polymorphic memory resources) these requirements can often be satisfied.

Target Audience: Architects, Developers, Project Leader
Prerequisites: Good C++ knowledge
Level: Advanced

Extended Abstract
For many embedded C++ applications, compliance with the AUTOSAR or Misra rules is required. Among them is AUTOSAR Rule A18-5-5.
Memory management functions shall ensure the following:
(a) deterministic behavior resulting with the existence of worst-case execution time,
(b) avoiding memory fragmentation,
(c) avoid running out of memory,
(d) avoiding mismatched allocations or deallocations,
(e) no dependence on non-deterministic calls to kernel.

This rule has far-reaching consequences, because per default the C++ standard library containers allocate their memory with new and free it with delete. These calls
• do not have deterministic execution times.
• can cause memory fragmentation.

Since new and delete violate A18-5-5, the default STL containers must not be used in applications requiring AUTOSAR compliance. This holds for many embedded applications.
With the allocators available since C++17 in the namespace std::pmr (polymorphic memory resources) these requirements can often be satisfied. This means that, for the first time in the history of C++, the containers and algorithms of the C++ standard library can be used in applications that require AUTOSAR Rule A18-5-5.

New MISRA C++ 202x are coming. This release will address modern C++ and thus is relevant not only for safety critical code in the automotive sector, but also for day-to-day C++ development. Expect the new MISRA rules to be less "miserable" for your day-to-day coding and use static analysis tooling to enforce them. Understand what kind of C++ will be considered unsafe, get examples of guidelines and learn which rules better to suppress in static analysis tools in non-safety-critical software.

Target Audience: Developers
Prerequisites: C++
Level: Advanced 

Extended Abstract
New MISRA C++ guidelines are coming. In contrast to the previous release in 2008, MISRA C++ will address modern C++ as it is used in modern automotive systems. It is expected that other domains with safety critical software will incorporate MISRA C++ as well. In addition many of the rules, especially those supported by corresponding static analysis tools, can make your own C++ coding practices better. The talk briefly introduce the development of safety critical software, highlight some of the modern MISRA C++ rules applicable to all kinds of code and will also show corners of the rule set, better avoided for "normal" C++ code. The latter is important to know, when you are asked to retrofit your code to the output of a static analyzer tuned to MISRA-C++.
* History of MISRA C++ guidelines
* Peculiarities of software development for safety-critical systems
* Example guidelines and how they influence coding
* "strong stuff" that should not bother in normal C++
* Outlook on further revisions

The adoption of static analysis of C++ and Java requires that the findings and errors can be prioritised in an efficient way. Our work shows that Machine learning (ML) can support this presentation of static analysis results to end-users. The ML engine learns from the codebase itself, and also observes the violations that the user fixes and which he ignores. The ML uses this to suggest the next best violations to fix, relying on probability of violations to be harmful or most likely to be a noise.

Target Audience: Developer Managers, R&D Managers, Software Architects, Software Engineers
Prerequisites: English, Software development, Coding experience, C++, Java, C#.
Level: Expert

Extended Abstract
Static code analysis is often understood as a mandatory part for checking the source code compliance to government and industry regulations, company-wide guidelines and practices. It can play, however, a more fundamental role in estimating the quality of the code in general, understanding the amount of technical debt, creating the strategy to reduce the amount of technical debt, as well as a helper in making decisions on how to speed up the development by creating a more maintainable, understandable, sociable codebase.

However, by its nature, static code analysis is bound to produce a large amount of noise and false alerts that can distract the team from the actual bugs in the code and prevent them from working thoroughly with the findings. One of the reasons for that is the level of soundness of the static analysis tool. If we want to be sure that the analysis is bound to find all errors in the code, the static analysis tool has to report all possible candidates. The more sound the tool is configured to be, the larger the number of the possible errors is reported, which increases the number of false positives as well.

To improve the user experience of working with static analysis technology, we have developed a machine learning (ML) based approach to presenting the results of the static analysis to users. The ML engine can learn from the code base itself, from a user's preferences, as well as from the interaction within the team. At the code level, our engine learns from the syntactical and semantical structure of the analyzed code to understand which violations are more likely to cause more harm, which violations are more likely to be noise, what underlying problems can be fixed to drastically reduce the number of reported violations. At the user level, the ML engine observes which violations the user fixes and which violations the user ignores. Based on these observations, the ML engine builds a model and uses it to suggest the next best violations to fix.

Maintaining a database containing millions of products can be very challenging, especially when the information you require of these products is subject to changes over time.

We show how we used state of the art Deep Learning methods (such as Transformers, BERT) in connection with smart text matching in order to extract relevant information from free-form text.

We also explain how we leveraged the existing database to create an automatically labelled training dataset.

Our model enables us to continuously update idealos database automatically.

Target Audience: Decision Makers, Technical Project Leaders, Developers
Prerequisites: Basic knowledge of machine learning methodology
Level: Advanced

Extended Abstract

To maintain idealos product base, product information in the form of values of predefined product attributes needs to be extracted from free-form text product descriptions.

Before the use of a Machine Learning based solution, this process required a lot of manual work to define rules to extract this information. There is also a high effort connected to keeping these rules consistent across the whole database and different types of products, especially since the source of this information (the product descriptions) as well as the required information (the product properties) are subject to changes over time.

In this talk we present a machine learning solution, based on fine tuned state-of-the-art models such as BERT, which is able to extract product information automatically from product descriptions with production-ready performance.

Our solution contains two different models, each following one of the well-known problem settings in Natural Language Processing (NLP): Semantic Segmentation of text (also known as Token Classification) and Question Answering. We will present both models in detail, as well as discussing their advantages and disadvantages for solving the task at hand and how we measured its performance (metrics).

We will also emphasize the importance of identifying aspects of your data that ensure that the developed model can actually fulfill your business needs before curating your dataset.

This highlights another benefit of implementing a Machine Learning model for a huge database: You will get sanity checks of your existing data “for free”, as consistent data is a prerequisite for a successful Machine Learning project.

One problem that is very common in large organisations is that there is often no or only very little training data in the form of labels for specific text sections available. We show how we mitigated this problem by leveraging the existing database to generate a large artificial training dataset. This allowed us to only use a few thousand manually labelled examples for training and testing to reach sufficient performance.