fuyou001
fuyou001
good improvement，looks to good to me.
in some scenes，reduce 80% disk bandwidth, also speed up broker startup.
虽然是sleep 0，但因操作系统时钟精度原因，实际会sleep 几毫秒，不同的系统，精度可能值会不一样； 删除之后，就没有sleep的效果了；
From the point of view of stability, I suggest keeping the logic as it is, in benchmark case, consumer is not one instance, if we change , the code will...
> > From the point of view of stability, I suggest keeping the logic as it is, in benchmark case, consumer is not one instance, if we change , the...
what's the plan to fix it ?
更标准做法，是要放到try外面。
1、Could you provide the performance benchmark data for the original Fastjson 1? 2、Could you provide compatibility test results with Fastjson 1, particularly when the object contains camel-case properties?
Could you develop a tool to validate whether objects with camelCase properties are compatible with Fastjson 2, so we can check compatibility? @yx9o