I've run into a bit of a feature-ordering dilemma, where two features can't be consistently ordered to produce the desired results. I'd thought of a rather ad-hoc hack to solve this, but was wondering whether this might potentially cause problems.
What I'd thought of doing was something along the lines of the following, where a 'dummy' feature is declared simply as a placeholder for a lookup:
feature QRXC { # dummy feature
lookup X {
# do X
} X;
} QRXC;
feature ss01 {
# do Y
} ss01;
feature ss02 {
lookup X;
# do Z
} ss02;
This would force the relevant lookups to be ordered such that feature 'ss02' would be able to do some stuff both before and after feature 'ss01' is applied. Feature 'QRXC' presumably would never be activated on its own since 'QRXC' isn't a registered feature and presumably therefore no application will be aware of it.
Has anyone ever made use of this approach? Are there potential problems which might arise from using 'dummy' features (apart from the obvious problem that someone might someday register such a feature)?
André