This is something I built a little while ago. It's a proof-of-concept and inspired by what I was doing in my day job.<p>To play with the demo spreadsheet you'll need to clone the repo and run the demo script, ie:<p><pre><code> git clone git@github.com:bsdz/calcengine.git
cd calcengine
python3 -mvenv --prompt calceng .venv
. ./.venv/bin/activate
poetry install -E demo
python demo/spreadsheet/main.py
</code></pre>
There's an inline gif showing the spreadsheet in use.<p>Hope it's interesting :)
This is really interesting because I was working recently on a way to have complete spreadsheets using only CSV files. The project is in my Github (<a href="https://github.com/yassirnajmaoui/runcsv" rel="nofollow">https://github.com/yassirnajmaoui/runcsv</a> and <a href="https://github.com/yassirnajmaoui/runcsv-gui" rel="nofollow">https://github.com/yassirnajmaoui/runcsv-gui</a>)
The only (major) thing is that the cells are run in top-bottom left-right instead of running a graph like this repo suggests.
This is cool! Also gives me a great excuse to post one of my favorite papers, "Build Systems à la Carte", about how make and Excel are solving the same problem in different ways: <a href="https://www.microsoft.com/en-us/research/uploads/prod/2018/03/build-systems.pdf" rel="nofollow">https://www.microsoft.com/en-us/research/uploads/prod/2018/0...</a>
This is pretty cool.<p>Spreadsheets are so powerful and really great and producing one-off solutions, but when you want to automate them you always need to redo most of the work in a "serious" programming language and wrap all of it in an API etc.<p>I always wondered how one could bridge the gap between creating a PoC computation in a spreadsheet and getting the same code to run in production without the rewrite. This could be a nice first step for a python server application designed in a spreadsheet.
Looks very interesting.<p>Can we plug numpy into this? Or is it not necessary with the assumptions that most steps will be cached and don't have to be recalculated.