Skip to content

LEGAL-REVIEW.md

Latest commit

 

History

History
102 lines (75 loc) · 4.46 KB

LEGAL-REVIEW.md

File metadata and controls

102 lines (75 loc) · 4.46 KB
Raw

Redistribution of EASA publications — findings and decision

Status: resolved 2026-08-10. Git history is not being rewritten.

This file existed because the repository used to contain complete EASA Easy Access Rules publications under an MIT licence covering the whole tree. Below is what was actually found, and what was decided.

Not legal advice. This is a reading of publicly available terms by the people building the tool. If this project is used commercially, get a qualified opinion.

1. What was committed

File Size Publication sha256
tests/real_samples/cs-vla.xml 6 402 678 B (~6.1 MiB) Easy Access Rules for Very Light Aeroplanes (CS-VLA), Amendment 1 b24a3ab4fe969d646fe3ffb217e693e596ad490e9bf1ae5907723ca754f05bee
tests/real_samples/cs-23.xml 4 184 733 B (~4.0 MiB) Easy Access Rules for Normal-Category Aeroplanes (CS-23) cbaf332c96b4c7b54e6776f8a74df36a8d411885b7d135fbab2f363e76504743

Both complete XML exports of the published regulation text, introduced by a single commit (59fcb91) and never modified afterwards.

2. Finding: reproduction is authorised

EASA's copyright page carries the standard EU formula: reproduction is authorised provided the source is acknowledged, save where otherwise stated. Where prior permission is required, that requirement cancels the general permission and must say so explicitly.

Two things were checked before relying on this:

  1. The "Official Publication of the Agency" carve-out does not apply here. An earlier draft of this document assumed it might. It does not: that clause (referencing Executive Director Decision 2012/163/E) sits in the Disclaimer section and concerns accuracy and completeness of website material. It is not a copyright exception.
  2. The publications state no reproduction restriction of their own. The front matter of CS-VLA is a liability disclaimer — it notes that the document is a consolidated, unofficial compilation and that EASA accepts no liability. There is no "otherwise stated" restriction to trigger.

Conclusion: redistributing EAR XML with attribution appears to be permitted. Committing those files was not a licensing violation, and rewriting history would solve a problem that does not exist.

3. The real issue: the MIT grant

The sharper problem the review raised remains valid, and is independent of whether redistribution is allowed. MIT grants any recipient the right to use, modify, sublicense and sell the covered work. Applied to a tree containing regulatory text, it purports to grant rights over that text which nobody here holds.

That is fixed by stating the boundary, not by deleting files. See the root NOTICE, which carves regulatory text out of the MIT grant and records the attribution EASA's policy requires.

4. Decision

Question Decision
Rewrite git history? No. Nothing was violated; a rewrite would invalidate every clone, fork and the v0.1.x tags for no benefit.
Keep publications out of the repository? Yes — but for engineering reasons, not legal ones (see below).
Address the MIT scope? Yes — root NOTICE, carving regulatory text out of the grant.

Publications stay out of the repository because:

  • 11 MiB of binary-ish XML in a repository of ~3.5k lines of code is a poor trade, and it bloats every clone forever;
  • the pins (*.meta.yaml) already guarantee reproducibility by sha256, so nothing is lost — fetch_samples.py reconstructs the exact bytes and refuses anything that does not match;
  • it keeps the sdist on PyPI lean.

The blobs remain reachable in history. That is now a deliberate choice, not an oversight.

5. If the decision is ever revisited

A rewrite was prepared and deliberately not executed:

# 1. Full backup first — this rewrite is not reversible in place.
git clone --mirror . ../easa-erules-backup.git

# 2. Requires git-filter-repo (pip install git-filter-repo).
git filter-repo --invert-paths \
    --path tests/real_samples/cs-vla.xml \
    --path tests/real_samples/cs-23.xml

# 3. Verify the blobs are gone.
git log --all --oneline -- tests/real_samples/cs-vla.xml   # expect no output

# 4. Force-push, then tell anyone with a clone to re-clone.
git push --force --all && git push --force --tags

It invalidates existing clones, forks and tags, and PyPI releases already published cannot be altered.