Skip to content

README.md

Latest commit

 

History

History
55 lines (34 loc) · 1.53 KB

README.md

File metadata and controls

55 lines (34 loc) · 1.53 KB

Memoising Scope Graph Query Resolution

This is the repository containing the implementations of both A_mem and A_cur, alongside the tools to recreate all graphs in the thesis.

Organisation

  • scope-graph: contains Scope graph and query resolution algorithm definitions
  • data-parse: parses raw data obtained from artifact of Specializing Scope Graph Resolution Queries.
  • pattern-recog: crude pattern recognition for data obtained from data-parse.
  • graph_bench: graph and table generator for benchmark results.
  • graphing: utilities to create MMD and PlantUML diagrams, for debugging.

Requirements

It is recommended to use uv for the python scripts. To install, run:

curl -LsSf https://astral.sh/uv/install.sh | sh

If this doesn't work, consult the official documentation.

How to run

Steps to recreate the benchmarks:

  1. Run the benchmarks
cargo bench -p scope-graph --bench sg-patterns

This can take between 5-10 minutes, the results are in scope-graph/output/benches/results.json

  1. Create graphs

With uv:

cd graph_bench
uv run plot_bench.py

The figures and tables are saved in graph_bench/plots/

data-parse

data-parse parses the data obtained from the artifact and provides a cosmo.csv to view the resulting scope graph in cosmograph. Run it with:

cargo run -p data-parse --release

It is recommended to run in a release build, as deserialisation can take a while.