Repoint the hansen_singleton pair at QuantEcon/data-lectures (wave B2')

[mmcky]

Repoints both hansen_singleton lectures at QuantEcon/data-lectures, completing wave B2′ of the datasets migration. Mirrors the wave landed in QuantEcon/data-lectures#82. Work plan: QuantEcon/workspace-lectures#39.

Deletes nothing. The _static copies and their READMEs stay until this has published, per the repoint → publish → delete ordering rule.

No figure can change

The bytes in data-lectures are byte-identical to the copies these lectures read today — verified at the landing PR, and again against main just now:

URL the lecture now reads	HTTP	sha256	Bytes
data-lectures/raw/main/lectures/hansen_singleton_1982_data.csv	200	3e9d4f37c31dbcab…	11,662
data-lectures/raw/main/lectures/hansen_singleton_1983_data.csv	200	dc5c1f8dac4b50ab…	26,084

Both builder links resolve too (blob/main/builders/hansen_singleton_198{2,3}_data.py, 200 each), and a never-existed control path on the same host returns 404 — so those 200s are real hits, not a soft-404 page.

Six edits, and only two of them are code reads

Three per lecture. This is the part worth reviewing carefully, because two of the three break silently:

1. The data URL — collapsed, not patched. It was a three-line string concatenation:

DATA_URL = (
    "https://github.com/QuantEcon/lecture-python.myst/raw/refs/heads/main/"
    "lectures/_static/lecture_specific/hansen_singleton_1982/"
    "hansen_singleton_1982_data.csv"
)

The stem line could not be edited in place. The published tree in data-lectures is flat — it has no _static/lecture_specific/<lecture>/ segment — so patching the stem alone would have left the filename appended to a path that does not exist. Both are now single lines on the bare main form.

2. The prose link to the maintenance script. It pointed at _static/lecture_specific/hansen_singleton_1982/make_data.py in this repo. That builder moved repos and was renamed: both files were called make_data.py, which works beside their own output where the parent directory supplies the meaning, and collides outright in a flat builders/ tree. Each took its dataset's stem.

3. The same builder path, named again in a docstring inside the data cell. Easy to miss — it is not a link and not a read, just a path in prose that would have quietly become wrong.

Nothing in this repo or anywhere else would have caught edits 2 and 3. data-url-guard.yml greps only for the LFS media host. jupyter-cache hashes code cells only, so a stale prose link cache-hits and goes green — which is exactly the failure class that made the D2 decision necessary in wave B1′. Both were found by grepping the filename and the URL stem separately; a line-based grep for the whole old URL returns a confident zero against the wrapped form.

Verification

• Both directions. No surviving reference to lecture_specific/hansen_singleton or make_data.py in either lecture (grep exit 1, against a positive control proving the pattern matches when present); all six new references present and correct.
• Rest of the repo swept. The only other mentions of these filenames are inside the two _static/.../README.md files, which are scheduled for deletion in the follow-up PR along with the CSVs and builders.
• The data cell re-executes on this PR — the DATA_URL change alters the cell hash — so CI genuinely exercises the new URL rather than restoring it from cache.

What follows

1. this PR merges → the lecture-python.zh-cn sync PR generates;
2. migration.yml flips landed → repointed in data-lectures, recording consumers;
3. publish lecture-python.myst, then merge the sync PR, then publish lecture-python.zh-cn;
4. deletion of the _static copies, builders and READMEs — its own PR, after the publishes.

Note for step 4: deleting these files will not stop the site serving them. ci.yml and publish.yml restore the weekly build-cache artifact and build over it, and Sphinx never prunes removed _static assets, so a deletion keeps being served until cache.yml rebuilds clean (Mon 03:00 UTC) and a publish follows. Verify that deletion against the published URL, never against main. See QuantEcon/data-lectures#81.

🤖 Generated with Claude Code

[Copilot]

Pull request overview

Repoints the Hansen–Singleton (1982/1983) lectures to load their CSV datasets from the centralized QuantEcon/data-lectures repository as part of the datasets migration, while updating accompanying documentation references to the new builder scripts.

Changes:

• Updated each lecture’s DATA_URL to fetch hansen_singleton_198{2,3}_data.csv from QuantEcon/data-lectures.
• Updated the prose link and in-cell docstring references from the old in-repo make_data.py scripts to the renamed data-lectures/builders/hansen_singleton_198{2,3}_data.py.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

File	Description
lectures/hansen_singleton_1982.md	Switches dataset source + updates builder references for the 1982 lecture.
lectures/hansen_singleton_1983.md	Switches dataset source + updates builder references for the 1983 lecture.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[github-actions]

📖 Netlify Preview Ready!

Preview URL: https://pr-1036--sunny-cactus-210e3e.netlify.app

Commit: 1588de6

📚 Changed Lectures

• hansen_singleton_1982
• hansen_singleton_1983

---

Build Info

• Workflow: Build Project [using jupyter-book]

[mmcky]

✅ Translation sync completed (zh-cn)

Target repo: QuantEcon/lecture-python.zh-cn
Translation PR: QuantEcon/lecture-python.zh-cn#250
Files synced (2):

• lectures/hansen_singleton_1982.md
• lectures/hansen_singleton_1983.md