Hi, Am Freitag, den 23.05.2008, 11:35 +0200 schrieb Niklas Gustavsson: > On Fri, May 23, 2008 at 11:16 AM, Felix Meschberger <[EMAIL PROTECTED]> wrote: > > Looking at this report, there is one method - objectId(Object) - which > > is not clearly destined at creating a LogFactory implementation. > > > > So my question with respect to JCL is: Do these bundles really require > > 1.1.1 as per > > > > Import-Package: org.apache.commons.logging;version=1.1.1 > > > > Or do such bundles need the objectId method ? > > I'm pretty sure they import 1.1.1 only due to having a Maven > dependency on 1.1.1, not because the require anything from 1.1.1. So, > one option is of course to complain at the source (which I'll do as > well) and one is for Sling to support 1.1.1.
ok. > > I would also add that the static fields might actually be used by > clients who want to configure the JCL behavior. Well, as far as I can tell, jcl-over-slf4j ignores any JCL configuration, because it is hardly linked to SLF4J. So any users of the JCL LogFactory class out of the Sling log bundle, will always get SLF4J backed JCL loggers no matter what they configure. Of course, their applications would break if they would use the constants regardless of their use in here... > > > If it is for the first reason, we may "hack" this by just exporting the > > o.a.commons.logging package as being of version 1.1.1. If for the second > > reason, we can probably not do much about it. > > One possibility is to get SLF4J to support JCL 1.1.1, no idea if Ceki > thinks that a worthwhile thing to do. I assume technically, it is not much: just update the dependency and include the most recent LogFactory class. Regards Felix > > /niklas
