Hi,
A bit of background.
Maven allows control over which profiles are activated either explicitly by using -Pprofile1,!profile2, or specifying certain activation conditions, usually by system properties.
However, there is a subtle difference in how -P and -D profile activation happens.
-P will only activate profiles on the projects that are being built, while -D (or any other activation mechanism) will also activate profiles in dependencies.
Consider an example of three projects A, B and C. A has a profile which specifies B as a dependency. B has a similar profile with the same id which adds C as a dependency. Both profiles are also activated by the same property.
Trimmed output of dep:tree commands:
$ mvn dependency:tree
[INFO] test:a:jar:1.0
$ mvn dependency:tree -Palt
[INFO] test:a:jar:1.0
[INFO] \- test:b:jar:1.0:compile
$ mvn dependency:tree -Dalt
[INFO] test:a:jar:1.0
[INFO] \- test:b:jar:1.0:compile
[INFO] \- test:c:jar:1.0:compile
We have a setup where activating a certain profile results in an 'alternative' build with a different dependency tree.
This is needed for allowing a third party to access a limited set of codebase. Some artifacts are effectively removed by specifying a dummy version (empty jar) and some are added.
Now, in M2E explicit profile activation (analogous to -P) works the same way as in cli, except that you can be more detailed in which profiles are active for each project.
But the only way I have found to specify profile activation properties is putting them into eclipse.ini, which is not that easy to force people to do.
So my question is: what if there was some ui element in m2e preferences that would allow setting what would be considered
system properties by underlying maven runtime?
Do you know of any potential problems with such a solution?
I do understand that these properties are global, they affect the whole workspace and changes should trigger full rebuild of all the projects.
--
