we now have an object whose initialization code failed, so it may be an unusual state requiring a bool IsValid() state checking mechanism (or similar) which is easy to forget to call. • You cannot take have a FooBuilder class be a friend of Foo so that it can construct the inner state of Foo correctly, without exposing this state to the world. In some cases it may be useful to make a unittest class a friend with a factory function or an Init() method, but these require heap allocation or a new “invalid” state, respectively. • Exceptions are really handy in testing frameworks. • When you add a throw statement

break before the closing brace. • Line break after the closing brace, only if that brace terminates a state- ment or terminates the body of a method, constructor, or named class. For example, there is no line 4.8.4 Switch statements Terminology Note: Inside the braces of a switch block are one or more state- ment groups. Each statement group consists of one or more switch labels (either case FOO: or default:) names to separate logical compo- nents of the name. One typical pattern is test_<state>, for example testPop_emptyStack. There is no One Correct Way to name test methods. 5.2.4 Constant

statement. After it yields a value, the runtime state of the generator function is suspended until the next value is needed. Pros: Simpler code, because the state of local variables and control flow are preserved should follow the same import and naming guidelines as functions. Decora- tor pydoc should clearly state that the function is a decorator. Write unit tests for decorators. Avoid external dependencies in automatically closed when the file object is destructed, tying the life-time of the file object to the state of the file is poor practice, for several reasons: 25 • There are no guarantees as to when the runtime

 Problem of Weak Baselines  Authors want to show improvements  Successive improvements over state-of-the-art don’t add up Experiments and Result Presentation [Timothy G. Armstrong, Alistair Moffat

Introduction and Overview Matthias Boehm, Graz University of Technology, WS 2020/21 Abstract % 1. State the problem Large-scale machine learning (ML) algorithms are often iterative, using repeated read-only