Tasks #6: Build systems

Other tasks: 1, 2, 3, 4, 5, 6, 7.

General notes

Use your GitLab repository to submit your solutions. Please, use exact filenames as specified in the task name.

Please, respect language requirements of individual tasks. For Python, we expect Python 3 as the default (if you really need to use Python 2, use a proper shebang). For shell scripts, we expect that you will use Bash although none of the tasks requires any use of Bashisms for an efficient solution.

Deadline for this set of tasks is 2020-07-26 AoE, it is required to get at least 5 points from this group.

Tasks description

06/Makefile (4 points)

Source code for this group of tasks is in thesis subdirectory of teaching/nswi177/2020-summer/upstream/examples repository.

The whole subdirectory contains LaTeX template for typesetting bachelor thesis at our faculty. The template comes originally from here. We have extended it a little bit to show you that you can use LaTeX even without learning it in great detail.

If you look into content/, you will see several Markdown files. Each represents one chapter. These files are processed by Pandoc that converts them to LaTeX and they are inserted into the main LaTeX document that is later converted into PDF.

All this is automated by the build.sh script that converts the individual chapters and generates the final PDF.

Your first task is to convert this script into Makefile to ensure that the final PDF is properly re-made but no unnecessary computation takes place.

Updates

You will need to install the following packages: pandoc, latexmk and TeX Live. Note that Tex Live packaging differs across distributions, for Fedora we recommend to install texlive-scheme-medium that covers most of the usually needed packages and also texlive-pdfx.

If the PDF generation stops in the middle with a ? prompt, you probably have not installed some package: its name is hidden in some of the preceding lines. You can also try to hit Enter only to continue the compilation (of the PDF) or X to terminate.

Note that we have also updated the upstream repository: we have moved thesis.xmpdata one folder up.

make clean (1 point)

Add target clean to the Makefile that removes all generated files.

Automated table (2 points)

The table src/tab_test_overview.tex is actually computed from data/tests.csv.

To ensure that the (fictional) thesis would always contain the most up-to-date data, implement a script that generates the table from the CSV source data.

Ensure that the Makefile rule captures properly the dependency (i.e. do not regenerate the table if the source CSV does not change but regenerate the PDF if the table changes).

The table is actually very simple – it lists tests, configurations and number of rows for this pair.

You can use Python or Shell for implementing this script. Note that the core of the script is just sort | uniq -c so do not look for extra complexities.

Do not forget to remove the original table from src/. There is no need to store the table in Git if we are able to recompute it. And recompute it quickly.

Also note that it would make more sense to store the table in out/ as other generated files.

Automated plots (3 points)

You probably noticed that the thesis contains one plot. Use pyplot to generate a bar-plot showing maximum for each test (i.e. ignore the configuration column altogether).

Note that the current plot was taken as a demonstration only and does not show the intended result of your script.

Look into content/80-evaluation.md to see how the image is inserted into the final PDF. You will probably want to change the path to the file, you will not need to modify the rest of the code.

Ensure that the Makefile rule captures properly the dependency (i.e. do not regenerate the plot if the source CSV does not change but regenerate the PDF if the plot changes).

Important: generate the image as PDF or other LaTeX-compatible vector format. We will not accept solution generating a bitmap.

Spell checker (2 points)

Add (phony) target spelling to your Makefile that runs Aspell on your input files to help you fix spelling typos.

We leave it to you what exactly the command will do. It can run the interactive check, it can highlight misspelled words (grep --color style) or …

GitLab job artifact (2 points)

Add .gitlab-ci.yml to your repository (if you do not have it there already) that builds the PDF.

Configure the job so that it stores as an artifact the generated PDF.

Please, ensure that the artefact expires in less than a week.

Update: we recommend to use aergus/latex as the image in your GitLab CI. This image also contains Pandoc and make.