zrlw
zrlw
如果org.bson.types.ObjectId直接有readResolve,那么应该是支持的,但是它是内部类里面实现的,hessian为了执行高效没有去查内部类,还有你说的jdk9之类的List.of这些可能都不是hessian的菜吧,搞这些东西之后可能会把hessian的执行效率降低到java序列化一个档次。 要完整支持序列化,可能要用java序列化而不是hessian了。
close it as the author hasn't responded for a long time.
> https://oss.sonatype.org/content/repositories/snapshots/com/alibaba/fastjson2/fastjson2/2.0.58-SNAPSHOT/ 问题已修复,请帮忙用2.0.58-SNAPSHOT版本验证,2.0.58版本预计在5月初发布。 > > 能介绍为什么有这个问题么? 因为dubbo的fastjson2单元测试会偶发断言失败,断言反序列化Date序列化结果到Byte类型异常,偶然发现有时断言失败,并没有抛异常,所以debug看了一下,发现new Date取到的时间恰好毫秒数为0时,序列化的结果都能反序列化为byte类型
我看了一下 support readMillis as int32 (https://github.com/alibaba/fastjson2/commit/f5c0b6287e2c27c3e185f49fe959403a76bc7f45) 那么是否可以确认fastjson2允许将Date类型的序列化串反序列化为Byte?我看ObjectReaderImplByte#readJSONBObject是先获取int32,再将int32的byteValue值作为返回值(取int32最后一个字节)?
> 是测试遇到这个问题么?还是业务场景有这个问题? 只是在编写转换结果类型与预期类型不一致的异常测试案例用到了,以为date类型肯定转不了byte类型,预期是抛异常,测试结果却是偶尔可以转成功。 日期类型可以转byte(丢失超过byte范围的数据)是主流json工具的做法么?
for example, memory should not be allocated for projectExecutions when running ```mvn source:jar``` because it's forkedExecution is true.
@olamy any plan about maven-build-cache-extension release? we want restore the deleted cache plugin configuration with the new version of maven-build-cache-extension: https://github.com/apache/dubbo/issues/15334
> > There are many Chinese comments in the change commit. > > Done.Translation completed. no, there are still chinese comments. i.e. McpConstant.java
check the reviews. @RainYuY
@wcy666103 spotless:apply to fix code formatting issue.