paulk-asert opened a new pull request, #2866:
URL: https://github.com/apache/groovy/pull/2866
… text
Inside withBatch { ... } the wrapper's only SQL-accepting method takes a
String, so a GString handed to it is coerced and its values become part of the
statement. Every query method on Sql binds the same idiom, and the class
documentation said so without qualification, so the one surface where it does
not hold read exactly like the ones where it does.
It is documented rather than rejected. A JDBC Statement batch may hold
different statements, which is why it takes text and has nothing to bind
against, and there is no parameterized equivalent to send callers to: batching
heterogeneous statements, generated DDL, and identifiers that cannot be bound
are all legitimate here. The module's own SqlBatchTest uses this idiom.
Rejecting would remove a capability with no replacement, which is the same test
applied to the quoted-expression case that is rejected precisely because
removing the quotes costs nothing.
So: an addBatch(GString) overload that renders and adds the command as
before, and logs once per batch when a value was composed in that was not
marked with Sql.expand. That mirrors asSql, which warns whenever it inlines
rather than binds; the difference is that inlining is the only option here, so
there is no strict mode to fall back from. Deliberate interpolation marked with
Sql.expand passes quietly and now substitutes the expanded value, which plain
GString rendering did not do.
The documentation is the substance of the fix: the class-level guidance now
carries the exception, both withBatch forms that hand out this wrapper say the
commands are text, and addBatch says it on the method itself.
P1 gains the rule that settles the finding's disposition: an API qualifies
when it has a GString form that binds, so this surface is outside P1 rather
than violating it, and the obligation is to say so and report the coercion.
--
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]
