Sascha Vogt wrote: > First of all, thank you very much for the explanation. I wasn't aware of > that up to know. Will definitely look into a different solution. > Nevertheless I want to understand a bit more to evaluate possible ways > to go. > > Am 27.09.2012 16:06, schrieb Jörg Schaible: >>> Am 27.09.2012 15:07, schrieb Jörg Schaible: >>>> Sascha Vogt wrote: >>> The idea was to have one Mojo extend another Mojo. >> In M2, it is simply not reliable and should therefore never be done. > Ok, for M2 the stuff we did is bad :) For now let's concentrate on M3 > >>> There is also a >>> maven-inherit-plugin out there >>> (https://github.com/ops4j/org.ops4j.pax.construct/tree/master/maven- >> inherit-plugin) >>> which suffers from the same issue. >> >> Yes, I complained about that before. >> >>> What do you mean by different classloader model? >> >> M3 uses isolated classloaders for the individual plugins. > So with the isolated classloaders the extension of a plugin could/should > work as expected (aka even if there are multiple different versions > referenced)? Or do you mean that in M3 if I extend plugin a, my plugin > shouldn't see classes from plugin a?
In M3 the plugins are no longer shared within the reactor i.e. each plugin is using its own classloader and can therefore have any arbitrary dependency. So if your plugin depends on another plugin, there should be no difference to any other dependency. However, I never tried this in M3 and I have therefore no idea how the plugin-plugin will actually react when you build yours. - Jörg --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
