Ben Rockwood writes: > I'm excited to see these changes. The way I interpret the bug is an effort t > o get away from having to export the manifest in order to investigate operati > on or diving into the property lists directly (I admit that I find an export > easier than digging through the layers of properties).
Yeah, svcprop <fmri> is currently a bit unwieldly to sort through and get everything I want. (It's still my preference, but can see how the contextual XML might be more useful to some.) > When interpreted in that way it would seem that the next logical inclusion wo > uld be context on each method; Good catch. :) I was thinking about the method contexts as well, but concluded that the quick effort to get in the simple method stuff would be a sufficient win for folks that I should just push it through. > but perhaps that makes the output a bit bloate > d. Perhaps adding -v to compliment -l? That seems like one way to deal with a more significant expansion of contextual data. Once we've got basic templates support complete, I'd like to explore this space more. That is, we'll be able to explain application data as well, and can actually do more significant annotation. It'd be good to be able to experiment with output formats and do some (formal or informal) usability testing on them, then decide on whether to shoehorn all or part of that into svcs -l, or use a different option or command. Hopefully you'd be OK with deferring this part a bit? (Templates library code was exactly the project I was procrastinating on. :) ) liane
