shtripat opened a new pull request, #758:
URL: https://github.com/apache/iceberg-go/pull/758
This change replaces panics with errors in schema visitors and add JSON
field validation
This PR fixes two classes of bug that could cause runtime panics in the
schema layer:
1. **`panic()` used as an error-signalling mechanism** in schema visitor
implementations (`indexByName`, `pruneColVisitor`, `sanitizeColumnNameVisitor`,
`validAvroName`, `sanitizeName`). While `Visit`'s `defer recover()` masked
these as errors at the surface, relying on panic-as-control-flow is
non-idiomatic, loses stack context, and is fragile.
2. **Silent JSON key mismatches producing zero-value fields**. Go's
`json.Unmarshal` ignores unknown keys, so a payload using `"field_id"` instead
of `"id"` (or `"field_name"` instead of `"name"`) silently produces `ID=0` /
`Name=""`. These zero values then propagate to the visitor code, triggering the
panics described above.
The `SchemaVisitor[T]` generic interface returns only `T` per method, so
error returns cannot be added without a breaking API change. Instead, **error
accumulation** is used: visitors store a first-error in an `err` field, visitor
methods short-circuit with `if x.err != nil { return <zero> }` after an error
is set, and the calling function (`PruneColumns`, `IndexByName`, etc.) checks
the accumulated error after `Visit` returns.
Fixes: https://github.com/apache/iceberg-go/issues/757
--
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]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
