Skip to content

Fix file recipe: replace Markdown downloads that now 404 - #550

Open
claudespice wants to merge 1 commit into
spiceai:trunkfrom
claudespice:fix-file-recipe-dead-markdown-downloads
Open

Fix file recipe: replace Markdown downloads that now 404#550
claudespice wants to merge 1 commit into
spiceai:trunkfrom
claudespice:fix-file-recipe-dead-markdown-downloads

Conversation

@claudespice

Copy link
Copy Markdown
Contributor

What the recipe said

file/README.md Step 1 of the Query Markdown Documents walkthrough downloads four docs pages:

files=(
  "clickhouse.md"
  "databricks.md"
  "debezium.md"
  "delta-lake.md"
)
for file in "${files[@]}"; do
  curl -O "$base_url/$file"
done

What actually happens

databricks and delta-lake are now directories in spiceai/docs (website/docs/components/data-connectors/), not flat .md files, so those two URLs 404. Because the script uses curl -O without -f, the 404 body is written to disk instead of failing — the reader silently ends up with two junk files:

$ for f in clickhouse.md databricks.md debezium.md delta-lake.md; do curl -sO "$base_url/$f"; done
$ ls -l *.md | awk '{print $5, $9}'
8088 clickhouse.md
14 databricks.md
17917 debezium.md
14 delta-lake.md

$ cat databricks.md
404: Not Found

The dataset still loads with 4 rows, so Step 4's output looks right — but two of the four "documents" contain the string 404: Not Found, which quietly defeats the point of the recipe.

What changed

Swapped the two dead entries for ducklake.md and iceberg.md — both still flat .md files in the same directory, and both keep the lakehouse-connector flavour of the originals. Updated the Step 4 expected output to match.

Verified all four download real content:

clickhouse.md      8088 bytes  --- title: 'ClickHouse Data Connector' …
debezium.md       17917 bytes  --- title: 'Debezium Data Connector' …
ducklake.md        8161 bytes  --- title: 'DuckLake Data Connector' …
iceberg.md        19050 bytes  --- title: 'Iceberg Data Connector' …

Checked against current stable Spice v2.1.1. The download step is confirmed by running it; the spice run / spice sql steps were not re-run (no CLI in the audit environment), but nothing about them changes — only the four filenames do.

Note for reviewers: adding -f (or --fail) to the curl would make a future breakage loud instead of silent. Left out here to keep the diff to the actual drift, happy to add it if wanted.

databricks.md and delta-lake.md became directories in spiceai/docs, so
curl -O silently writes 14-byte '404: Not Found' files. Swap in
ducklake.md and iceberg.md and update the expected query output.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant