This is my Charles University CS bachelor thesis. Feel free to grab what you like, but don't sell it and keep attribution (legalese: http://creativecommons.org/licenses/by-nc-sa/2.0/legalcode, humanese: http://creativecommons.org/licenses/by-nc-sa/2.0/). The source code is hosted on GitHub.

text/ contains the text of the thesis. src/ contains associated source code in C. TODO.md contains some ideas for future work.

I am developing this project on Linux with GCC. It may work somewhere else, but only accidentally.

The code requires LibUCW 6.4 to build (I use its red-black trees). My code assumes that it's installed under the /usr prefix.

For benchmarking, you additionally need:

• jansson for JSON output
• matplotlib for plotting graphs

experiments/btree-dot needs:

• graphviz for rendering graphs

experiments/cloud needs:

• urllib3 for fetching the dataset

make test
# Or:
make
bin/test

You need GCOV and LCOV.

make test_coverage
# browse src/coverage-out/index.html

src/experiments/performance contains a "script" that runs several operation sequences against dictionary implementations. Use it like this:

make bin/experiments/performance

# To run all benchmarks:
bin/experiments/performance

Results are saved in src/experiments/performance/results.json. You can plot some interesting graphs by running:

cd experiments/performance
./plot_results.py