> > Of course, our way here is to come to agreement and it looks like we have > > not done that in this case. My original vote (+1) was partly for > > consistency with the rest of commons and out of fear of "jar hell" > > scenarios. I was swayed by Niall's argument and reflection on use cases > > that I am aware of. I am sure Niall as well is open to enlightenment if > > others can point to practical use cases (observed or likely) involving > > [math] that demonstrate that the "pain-minimizing" alternative is to change > > the package name. > > Didn't someone say that math already had a "jar hell" situation > encountered by users?
The example situation (that triggered the vote proposal) was the inclusion of the commons-math JAR v1.2 into the released JAR of a project's library. [This was done to help the users grab all the dependencies at once. Actually, the only parts of commons-math currently in use in that project are "MathException" and "special.Erf"!] All they would need to do is replacing that JAR with the new release. Moreover, they will start using "Ivy" to manage dependencies, so actually the JAR won't be included anymore. So, this is hardly an inextricable jar-hell situation. Best, Gilles --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org
