SAP HANA Database Standards
Priority: P0 (CRITICAL)
Rules
- Strict Parameterization: Always use
? parameter placeholders for SAP HANA queries. Never use raw string concatenation, fmt.Sprintf, or untrusted interpolation for values.
- Batching & Chunking Limits: Chunk large
IN (...) parameter lists into batches of at most 1,000 items to avoid HANA driver and engine parameter limits.
- Explicit Column Aliasing: Always explicitly alias columns in multi-table queries (e.g.
m.code AS code, m.name AS name). SAP HANA and Go database drivers map scan columns by name; duplicate column names across joins silently collide or overwrite values. Never use SELECT *.
- HANA Datatype Casting Gotchas: SAP HANA functions (such as
COALESCE, SUBSTR, CONCAT) require strictly compatible parameter types. Explicitly cast values when comparing or falling back (e.g. COALESCE(col, 'default'), TO_VARCHAR(...), CAST(? AS NVARCHAR)).
- Identifier Case-Sensitivity: Unquoted SQL identifiers in SAP HANA default to UPPERCASE. Quoted identifiers (
"columnName") preserve case. Keep identifiers consistent with schema conventions.
- Keyset & Streaming Pagination: Avoid deep
LIMIT/OFFSET on large columnar tables. Prefer keyset/seek pagination (WHERE id > ? ORDER BY id ASC LIMIT ?) and stream large result sets using rows.Next() directly into models to prevent memory spikes.
- Transaction Scope: Keep transactions short and focused on atomic write operations. Never perform slow external I/O or unbounded operations inside an open transaction. Always pair with
defer tx.Rollback().
Verify
- Query uses
? placeholders for all user-supplied or dynamic values.
- Large
IN (...) queries chunk parameters into batches of ≤ 1,000 items.
- All joined queries use explicit, distinct column aliases without
SELECT *.
- Built-in functions (
COALESCE, SUBSTR) use matching datatypes or explicit CAST/TO_VARCHAR.
- Pagination enforces strict
LIMIT caps and stable ORDER BY.
- Transactions are scoped to write paths with
defer tx.Rollback().
Anti-Patterns
- No SQL String Concatenation: Never build dynamic
WHERE clauses by concatenating unescaped strings. Build parameterized placeholder slices.
- No Unbounded IN Clauses: Passing thousands of IDs in a single
IN (?, ?, ...) clause triggers HANA driver buffer overflow or query parsing degradation.
- No SELECT * on Joins: Scanning duplicate column names across joined tables leads to silent bugs where one table's ID overwrites another.
- No Incompatible COALESCE Arguments: Providing mismatched types (e.g. integer column with string fallback) causes runtime HANA
inconsistent datatype errors.
- No Heavy Reads Inside Transactions: Do not perform long calculation view aggregations or external network calls while holding an open transaction lock.
References
