On Mon, Apr 2, 2012 at 1:12 PM, Dawid Weiss <[email protected]> wrote: >> I don't think that you (carrot as a library) should do any shading. I >> think that when someone prepares a Solr component including carrot, >> that they should avoid dependencies that possibly conflict. Jackson is >> a particular risk factor, since 'minor' versions are mutually >> incompatible, so the presence of Jackson x.y precludes the use of some >> other package that requires jackson x.z. > > I realize this but then I think shading should be used exactly when > somebody encounters this kind of problem (incompatible versions must > coexist on classpath). It is a pain. It is wrong. But it's a less > error-prone solution than shading everything possible for every > possible package out there. > > I don't rule out the possibility that we will be preparing a > "self-contained" release in the future because JAR conflicts are a > common issue, but I also kind of know for a fact what kind of pain it > is because we _are_ trimming the distribution for .net > cross-compilation (which in the essence is similar to package > renaming/ obfuscation). It is an unbelievably tedious process to check > if everything works after each upgrade (even with tests and > everything).
Of course, some would say that Solr should be based on OSGi just to avoid this sort of conflict. (ducks under desk) > > Dawid > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [email protected] > For additional commands, e-mail: [email protected] > --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
