engelphi / algorithm Goto Github PK

During reorganization of algorithms, the documentation was dropped accidentally. Add documentation again.

Add CoverityScan for additional static analysis.

Move the library out of the include directory into base directory.

Reorganize algorithm from algorithms.hpp into separate files to avoid merge conflicts in case of parallel development.

Currently each push commit is build by travis. This cause travis to build the same commit twice if a pull request ist created which is a waste of resources.

Add an unzip algorithm that takes a range containing pairs of values and separates the elements into two ranges that contain elements of the pairs.
Example:
std::vector<std::pair<int, char>> input{{1, 'a'}, {2, 'b'}, {3, 'c'}, {4, 'd'}, {5, 'e'}};
std::vector first_range{};
std::vector second_range{};
algo::unzip(input.begin(), input.end(), std::back_inserter(first_range),
std::back_inserter(second_range));
Result:
first_range == {1,2,3,4,5};
second_range == {'a', 'b', 'c', 'd', 'e'};

When it comes to static analysis more tools are better. Therefore add cppcheck in addition to clang-tidy for additional analysis.

Currently the only compiler which is supported is clang 5.0.0. As the library is based on C++17 earlier Versions of clang are not compatible and do not need to be included in the continuous integration process. On the other hand g++ supports C++17 experimentally since version 7. The library should therefore be build and tested against g++.

Github provides the possibility to make releases available through the releases tab of the projects github page. The continuous integration process should be extended to deploy a release to github when the repository was tagged appropriately.

Switch to catch unit testing framework as it easier to integrate does not require to link against external dependencies.

The library should be documentated in a commonly used format which can be automatically build by the
continuous integration process. The generated documentation could then be deployed together with the library itself.

add zip_with and zip_with_n algorithms which implement the following behavior:

std::vector<int> first_range{1,2,3,4};
std::vector<int> second_range{1,2,3,4};
std::vector<int> output_range{};

zip_with(first_range.begin(), first_range.end(),
              second_range.begin(),
              std::back_inserter(output_range),
              [] (const auto& first, const auto& second) {
                 return first + second;
              });

output_range == std::vector<int>{2,4,6,8}