Hi folks, I want to discuss more details of sharding algorithm API for 5.x.
I have 2 issues that need to be discussed. Issue 1: As you know, we added a new type of sharding table -- auto table, which can assign actual data nodes automatically, users do not need to know actual table information details anymore. The auto table sharding algorithm configuration is similar to the original standard sharding algorithm. The current algorithm API is adding a new interface `ShardingAutoTableAlgorithm`, which can be extended by the original `StandardShardingAlgorithm` interface. The current algorithm types are standard, complex, hint and none. Do we need to add a new type of sharding algorithm to separate them totally? If needed, what is the name? Issue 2: Apache ShardingSphere uses SPI to introduce all algorithms, which are different from the original way, users can not use class names to configurare sharding algorithms anymore. Do we need to add a built-in algorithm type to provide a way to let users configure class names, just like INLINE type? ------------------ Liang Zhang (John) Apache ShardingSphere & Dubbo
