AsyncAppender can very well leverage this. Am I missing something? For the ones that do not implement these two methods, we can define a default interface method for them -- given log4j3 will target Java 8+.
On Sat, Nov 17, 2018 at 1:27 AM Ralph Goers <[email protected]> wrote: > > > > On Nov 16, 2018, at 1:14 PM, Gary Gregory <[email protected]> > wrote: > > > > On Fri, Nov 16, 2018 at 12:30 PM Volkan Yazıcı <[email protected]> > > wrote: > > > >> Hey Gary, > >> > >> *Package Name* > >> > >> Once every couple of months I found myself helping out people > >> for JAR Hell problems since they included wrong Log4j artifact. > >> The artifact and package names of Log4j 1 and Log4j 2 are > >> pretty similar looking. Hence I really encourage you to explicitly > >> state the version in artifact and package names. For instance, > >> log4j3-core and org.apache.logging.log4j3, etc. It goes without > >> saying, this will also aid SEO too, which is a pain right now. > >> > > > > IMO, we should change the package names and artifact IDs to contains a > "3" > > postfix, like we did in HttpComponents for version 5, so probably > > "log4j-core3", "log4j-api3" and so on. To be discussed... > > > > > >> > >> *Allowing Batches in Appender Interface* > >> > >> Is it also possible to extend the Appender interface such that in > >> addition to append(LogEvent), batched append(LogEvent[]), > >> append(LogEvent[], int offset, int length) are allowed as well? > >> > > > > Sounds OK, PRs welcome. Ralph, any thoughts? > > What would call the append(LogEvent[]) method? Nothing in the Logger > interface or implementation currently would. I suppose it could be useful > when one Appender is wrapped with another, but then wouldn’t every Appender > that can wrap other Appenders need to be modified to support this? > > Ralph > > >
