Long overdue and exactly what is needed by all framework developers.
On Tue, Dec 3, 2013 at 12:20 PM, Christian Grobmeier <[email protected]> wrote: > On 3 Dec 2013, at 16:14, Mike Kienenberger wrote: > >> On Tue, Dec 3, 2013 at 10:01 AM, Andrus Adamchik <[email protected]> >> wrote: >>> >>> Ideally I’d love to have zero dependency on a third-party logging >>> framework. So that probably means using j.u.l and let the users bridge that >>> as they want. Wonder if we’ll make everyone’s life miserable as a result? >> >> >> No, we made that horrible mistake of going with j.u.l at MyFaces a >> while back based in part on my lobbying for it. It is a disaster to >> end-users to have j.u.l as the default logging system. There's no >> way to bridge from j.u.l back to something else without huge >> performance hits, so you lose any control over what logging system >> your end-users can use. It's very difficult to get it to read >> configuration information from a file. It has horrible two-line >> logging output by default, which is again difficult to change. At >> least under weblogic app servers, you cannot identify which >> applications are generating which log messages, and I'm pretty sure >> that this is true in general for it. >> >> Really, there's nothing good you can say about it other than it might >> save you the 721K commons logging dependency. > > > I agree here. Jul is painfully broken and doesn't work well with others. > Even bridging from $x to Jul is painful. > > However I am currently preparing to put some effort into a new Logging JSR: > https://java.net/projects/newlogging > So far I didn't haven the time yet, but I plan to push this around xmas. > > Imagine a logging proxy like slf4j or log4j2 in the JDK. Not implementation > like > JUL, just the interfaces. This is the goal. Unfortunately this will be a > tough road > and if it ever happens is not sure. > > > > > > > --- > http://www.grobmeier.de > @grobmeier > GPG: 0xA5CC90DB
