Hi Eugen, On Mon, 2020-06-15 at 11:13 +0300, Eugen Stan wrote: > Hi, > > Removing the implementation leak a MUST. > > As I've mentioned in the issue/mail, logging implementation is a > deployment decision.
I agree > We should provide documentation and make it easy for people to > switch. > In a "product like" distribution we will deliver an implementation > suitable for that use case. I agree on all that, too. > Libraries should NOT depend on an logging implementation. > > Add the logging implementation to the test scope if needed. What is done in the source code for now. We use slf4j facade everywhere. > > == The rationale > > Some people log to memory > > Some people log to (rotating) files > > Some people log to local (on the same host) log aggregator > > Some people log to a central, over the network logging collector. > > > Every use case is valid - it depends on the decisions made when the > infrastructure was built. > > We can't know in advance what users will use. The only detail is that in that specific case, we are actually talking about the CLI tool: we don't really expect a use to choose the logging library for his CLI. And the error is about providing none. We can just include a Noop implementation to shut down the warning IMO. -- Matthieu Baechler --------------------------------------------------------------------- To unsubscribe, e-mail: server-dev-unsubscr...@james.apache.org For additional commands, e-mail: server-dev-h...@james.apache.org
