Hi, and thanks for assembling SWARM-CG — having the PyCG benchmark, its extended cases, and the HeaderGen flow-sensitive variants unified in one place (together with the EMSE 2025 paper) is genuinely useful.
The repository currently has no LICENSE file. Under GitHub's default terms that legally means "all rights reserved", even though the paper describes the benchmark suite as published open-source. Without an explicit license, downstream users cannot legally vendor the ground-truth files into their own test suites — which seems to be exactly the kind of reuse a "Swiss Army Knife of Call Graph Benchmarks" is intended for.
Would you consider adding an explicit license? For example:
- Apache-2.0, which would match the upstream PyCG evaluation benchmark (vitsalis/pycg-evaluation), or
- CC-BY-4.0, if you prefer a data-oriented license.
If some subdirectories inherit terms from their origins (e.g., the pycg/ cases are Apache-2.0 upstream), a short per-directory provenance/licensing note would also be very helpful.
Thanks!
Hi, and thanks for assembling SWARM-CG — having the PyCG benchmark, its extended cases, and the HeaderGen flow-sensitive variants unified in one place (together with the EMSE 2025 paper) is genuinely useful.
The repository currently has no
LICENSEfile. Under GitHub's default terms that legally means "all rights reserved", even though the paper describes the benchmark suite as published open-source. Without an explicit license, downstream users cannot legally vendor the ground-truth files into their own test suites — which seems to be exactly the kind of reuse a "Swiss Army Knife of Call Graph Benchmarks" is intended for.Would you consider adding an explicit license? For example:
If some subdirectories inherit terms from their origins (e.g., the
pycg/cases are Apache-2.0 upstream), a short per-directory provenance/licensing note would also be very helpful.Thanks!