|Re: [cross-project-issues-dev] Obscure org.eclipse.platform dependencies? Where are they?|
Whew, thanks to both of you for your responses. While the explanation itself is opaque to me -- why does platform have a dependency that isn't supported, or if it doesn't where did it sneak in from? ;-) I made the change that Paul suggested and it worked. I had copied the milestone reference for Eclipse from "another project" which just goes to show the perils and pitfalls of trying to assemble a working build from a set of other attempts.
In general, we have *got* to come up with ways to make these things more transparent/consistent/maintainable if nothing else for the benefit of others who haven't faced the trauma yet so they don't have to go through every possible failure mode too. One thought is to have some common shared rmap redirects for builds. We could have one for Helios baseline, Indigo, Modeling, Modeling + XText. etc.. In fact, we could even think about projects maintaining consumer rmaps in the same way that we now maintain our update sites and build rmaps. For example, I'd love to be able to just do a redirect to Indigo M2T/TMF/MWE and not have to worry about changes that might happen when a build dependency or update site location changes. I *don't* want to copy their rmap because I'm looking for a somewhat more stable (say I rather than N) build and I don't need their
One issue with this is that you have to specify the pattern for the redirects and there is no failure pass through. This means that you couldn't combine rmaps that have the same search patterns to create a sort of catch all and this would prevent consumer rmap providers from dealing with special cases such as TMF's use of com.google.. Thomas, has the Buckminster team thought about supporting some kind of "includes" style functionality?
On Nov 17, 2010, at 11:21 PM, Thomas Hallgren wrote:
Back to the top