Michael Hunger wrote: > I looked at the implementation and would like suggest the following to reduce > the coupling to MD5 and to > "UnsupportedEncodingExceptions" (and allow different > SchemaVersionCalculations). > > 1) create an interface SchemaVersion > 2) rename the methods on ValueType etc. to versionize(SchemaVersion) > 3.1) add such methods on QualifiedName and TypeName (to be created class) > 3.2) or have SchemaVersion.update(QualifiedName) and updateFrom(TypeName) > 4) have methods like update(byte[]), byte[] calculage(), String base64() on > SchemaVersion > > have a subclass of the interface called MD5SchemaVersion which is used in > EntityType.
It is essential that all schema calculations be done the same way. If you plug them, then yes, you have to specify with the schema what algorithm was used to create them. I fail to see the point of it... why not just use the default one? What is the gain of pluggability, considering the increase in complexity? /Rickard _______________________________________________ qi4j-dev mailing list [email protected] http://lists.ops4j.org/mailman/listinfo/qi4j-dev
