hickey-ai opened a new issue, #16521: URL: https://github.com/apache/dubbo/issues/16521
### Pre-check - [x] I am sure that all the content I provide is in English. ### Search before asking - [x] I had searched in the [issues](https://github.com/apache/dubbo/issues?q=is%3Aissue) and found no similar issues. ### Apache Dubbo Component Java SDK (apache/dubbo) ### Dubbo Version Apache Dubbo 3.3.7-SNAPSHOT, branch 3.3 at b73b9d131b31ed43869778abbbe799b8cd0d1d1b; OpenJDK 21; Windows. The actual dubbo-common module was built and tested (release 8 compiler target). Released artifacts and full-repository or production-network behavior were not separately tested. ### Steps to reproduce this issue Reproduce against the unmodified source above with JUnit 5: ```java import java.math.BigDecimal; import java.math.BigInteger; import org.apache.dubbo.common.utils.CompatibleTypeUtils; import org.apache.dubbo.common.utils.PojoUtils; import org.junit.jupiter.api.Test; import static org.junit.jupiter.api.Assertions.assertEquals; class DecimalConversionRegression { @Test void integralValueMustNotWrapThroughLong() { BigDecimal value = new BigDecimal("9223372036854775808.75"); BigInteger expected = new BigInteger("9223372036854775808"); assertEquals(expected, CompatibleTypeUtils.compatibleTypeConvert(value, BigInteger.class)); assertEquals(expected, PojoUtils.realize(value, BigInteger.class)); } } ``` Both conversion entry points return -9223372036854775808 instead. The BigInteger branch currently constructs BigInteger.valueOf(number.longValue()), so narrowing to signed 64 bits happens before construction. Large negative values and larger positive/negative decimals reproduce the same loss. ### What you expected to happen Preserve the complete arbitrary-precision integral part when converting BigDecimal to BigInteger. Fractional values should continue truncating toward zero, matching BigDecimal.toBigInteger(). Existing Integer/Long conversion behavior and the existing BigInteger identity path should remain compatible. ### Anything else The public PojoUtils.realize entry point reproduces the corruption. The compatibility response filter also delegates to these helpers; this report does not assume a particular deployment or serializer returns this shape. Seven decimal cases cover the two long bounds, large positive/negative fractional values and small-fraction controls; additional tests cover the POJO entry point, legacy integral conversions and identity preservation. Five regression invocations fail before the fix. All 54 CompatibleTypeUtilsTest / PojoUtilsTest invocations pass after the targeted BigDecimal.toBigInteger change, along with lifecycle Spotless and git diff --check. Searched existing issues and PRs for CompatibleTypeUtils, compatibleTypeConvert, BigInteger and BigDecimal precision; no matching fix was found. The tested proposed patch and regression tests are on hickey-ai/dubbo branch codex/fix-bigdecimal-biginteger-value; I will link the corresponding PR. This is a numeric correctness report, not a security or measured production-impact claim. Investigation and patch preparation were AI-assisted; the reported tests were executed locally. ### Do you have a (mini) reproduction demo? - [x] Yes, I have a minimal reproduction demo to help resolve this issue more effectively! ### Are you willing to submit a pull request to fix on your own? - [x] Yes I am willing to submit a pull request on my own! ### Code of Conduct - [x] I agree to follow this project's [Code of Conduct](https://www.apache.org/foundation/policies/conduct) -- 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]
