jiayuasu opened a new pull request, #1141:
URL: https://github.com/apache/sedona-db/pull/1141

   ## What changes are proposed in this pull request?
   
   Adds the missing `geography` entries to the `kernels:` frontmatter of the 
SQL function reference, so the docs match what the function registry actually 
accepts.
   
   The two pages reported in #1087:
   
   - `st_collect_agg.qmd` — `geography -> geography`
   - `st_makeline.qmd` — `(geography, geography) -> geography`
   
   Following the issue's suggestion, I also audited every other `.qmd` kernel 
list against the registry and found 14 more pages with working-but-undocumented 
geography signatures:
   
   - `ST_AsEWKB`, `ST_CRS`, `ST_Dump`, `ST_EndPoint`, `ST_StartPoint`, 
`ST_HasM`, `ST_HasZ`, `ST_M`, `ST_Z`, `ST_SRID`, `ST_Zmflag`, `ST_SetSRID`
   - `ST_Transform` — 2- and 3-arg geography kernels; the description notes the 
target CRS must be geographic
   - `ST_Envelope_Agg` — geography input, but the envelope is returned as 
**geometry** (planar edges), which the description now calls out
   
   ## How was the audit done?
   
   For every documented geometry kernel I built the geography twin and checked 
it against the registry at plan time, on two builds (the 0.4.0 wheel with 
`s2geography` and a dev build for the core kernels), then executed each 
candidate signature and inspected the output type's `edges`/`crs` metadata to 
get the correct `returns:` value. Each addition was also cross-checked against 
the `ArgMatcher` definitions in the Rust source. Probes that fail on argument 
*values* rather than types (CRS strings, SRIDs) can hide real kernels, which is 
how `ST_SetSRID` and `ST_Transform` were initially missed and then confirmed 
with valid values.
   
   Deliberately excluded: `ST_KNN` type-matches geography arguments but the KNN 
join rewrite has no geography path, so it always fails with "Can't execute 
ST_KNN() outside a spatial join" — that's tracked in #1086 (see also 
#783/#801), and documenting it here would repeat the inverse of this bug.
   
   All edited frontmatter renders cleanly through 
`docs/reference/sql/_render_meta.py`, showing both overloads in the Usage 
section.
   
   ## Are these changes tested?
   
   Docs-only. Every added kernel signature was verified by executing it against 
the registry as described above; the rendering pipeline was exercised on all 16 
edited pages.
   
   ## Are there any user-facing changes?
   
   Yes — the SQL reference now documents the geography overloads for 16 
functions.
   
   Closes #1087
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to