Relation literal rows are matched to columns by position, so rows with different field names or order produce wrong data

Aperta
  • Qualcosa non va?
#6,415 2 commenti 0 reazioni 0 assegnatari Vedi su GitHub

I maintainer di solito rispondono entro 3 giorni

Nessuno ha ancora preso questa issue.

Valutazione

Difficoltà
3/5
Tempo stimato
1-2 giorni
Idoneità per principianti
70/100
Tipo di issue
Bug
Chiarezza
Specificata chiaramente
Stato di attività
Attiva
Stack tecnologico
rust
Ambito
compilers

Direzione di ricerca

Start at prqlc/prqlc/src/semantic/lowering.rs L377-L414, where pl::ExprKind::Array is lowered into RelationLiteral.columns: columns come from the first row's lineage, and later rows' tuple values are copied in position, ignoring field names. First settle with the maintainers the open design question — reorder each row by name, or reject mismatched rows — plus the rule for unnamed fields like {1, 2}; coordinate with open PR #6413, which edits the same block. Done means from [{a=1, b=2}, {b=3, a=4}] compiles to SQL preserving names, and the other two example shapes either reorder or fail cleanly as decided.

Scritto dal modello di indicizzazione a partire dal testo della issue.

Descrizione

bug

A relation literal takes its column names from the first row only. Every later row is matched to those columns by position, ignoring its own field names. Queries whose rows disagree compile without error and produce wrong data.

from [{a=1, b=2}, {b=3, a=4}]

compiles (at c945ca81) to

WITH
table_0 AS (
  SELECT
    1 AS a,
    2 AS b
  UNION ALL
  SELECT
    3 AS a,
    4 AS b
)
SELECT
  a,
  b
FROM
  table_0

So the second row comes back as a=3, b=4 rather than a=4, b=3. Two related shapes go wrong the same way:

  • from [{a=1}, {b=2}] yields one column a with values 1, 2. The second row's b is renamed to a.
  • from [{a=1}, {a=2, b=3}] yields only column a. The second row's b=3 is dropped.

The lowering of pl::ExprKind::Array builds RelationLiteral.columns from the expression's lineage, then copies each row's tuple values in order without looking at their names (lowering.rs).

There are two plausible fixes: reorder each row's fields to match the first row's columns by name, or reject rows whose field names differ from the first row's with a compile error. Either way, unnamed fields ({1, 2}) need a rule too: whether they stay positional, and whether mixing named and unnamed fields is allowed. Open PR #6413 changes the same block (it reports malformed rows instead of panicking), so a fix here will probably build on it.

A review of #6413 first spotted this; it isn't caused by that PR.

Lingua principale
Rust
Stelle
10.9k
Fork
261
Merge medio
3g 21h
PR unite (30g)
84

Preparare l'ambiente

Apri in Codespaces

Avvia il container di sviluppo del progetto nel browser, con il tuo account GitHub.

Come iniziare

  1. Leggi tutta la issue e poi la guida ai contributi del progetto.
  2. Commenta sulla issue per dire che te ne occupi tu — evita che due persone facciano lo stesso lavoro.
  3. Fai un fork del repository e lavora su un branch.
  4. Apri una pull request che faccia riferimento al numero della issue.

Altre issue di PRQL/prql

Tutte le issue di PRQL/prql

Issue simili

Altre issue su Rust

Ricevi le nuove issue nella tua casella

Un breve riepilogo di issue GitHub adatte ai principianti.