On Wed, 6 Feb 2002, Craig R. McClanahan wrote: > Configuration is a feature of a particular *implementation* of logging. > The implementations we wrap all have their own configuration mechanism. > So does the simple logger implementation that writes to System.out (which > uses system properties).
Same is true for XML parsers - SAX is just an API to abstract the parser implementation, and each parser has its own config mechanisms ( as a feature of the implementation of parsing ). Yet SAX provides features, attributes that allow an application to tune the parsers without hardcoding calls to the actual implementation. ( like using internal xerces, crimson, etc ) What's different here ? The claim that commons-logging should be used only by commons components, and we want to force each application to still hardcode calls to the loggers it wants to use. Again, all that's needed in the interfaces is a simple get/setAttribute, with the mention that an logger adapter is free to support or not any attributes it wants. At least this would allow someone to read the adapter documentation, see what it supports, and pass info without using the actual class of the logger. Setting system properties doesn't work in a sandbox, storing config files in classes/ is ugly - and if we require Class.forName() and introspection or to hardcode a real logger - why bother with commons-logging. Costin -- To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>