Community
Participate
Working Groups
Created attachment 122854 [details] CompoundQueryable There are a number of places that take a set of queryables and perform queries across them all. For ContextQueries this does not work. For example: S1 = {1,2,3,4} S2 = {3,4,5} getMax{S1 + S2} != getMax{S1} + getMax{S2} In order to properly query the combined sets when ContextQueries are used, you need to: getMax(getMax{S1} + getMax{S2}) I have created a CompoundQueryable to make this easier for clients. The patch includes CompoundQueryable and the 10 test cases. 81% coverage in the compound queryable (there are a few unreachable methods because there is a private inner class that implements an interface, but not all the methods are used). The patch is ready for review.
Also, this patch adds a dependency on org.eclipse.core.tests.harness. If this is determined to be a bad idea (I posted a question about this), i will rework the test case that causes problems.
Created attachment 122873 [details] Updated patch Not all queries can be evaluated by first calling the query on the sub-queryables and then calling the query on the results. This patch fixes this. It allows query implementers to specify the query to use on the sub-queryables. There is a default implementation in all query superclasses. The default implementation is not optimal, but it works in all cases. I added a bunch of new tests cases for a variety of compound queries and compound queryables (and combinations of the two).
Created attachment 123152 [details] Compound Queryable that does not support transitive queries p2 currently doesn't have any transitive queries, so supporting them right now is not necessary. I have a patch that will do this (if needed) and I will attach it here. This patch adds CompoundQueryables that do not support transitive queries.
Thanks Ian. I committed the lates patch to head with a couple tweaks to JavaDoc. There remains the issue about non-transitive queries on compound queryables. Do you want to cover that in this bug or open another? Also, a question. Do CompositeRepos qualify as CompoundQueryables?
(In reply to comment #4) > Thanks Ian. I committed the lates patch to head with a couple tweaks to > JavaDoc. I noticed a few javadoc typos just now. If they still exist I will fix them. > > There remains the issue about non-transitive queries on compound queryables. > Do you want to cover that in this bug or open another? > I will do this. > Also, a question. Do CompositeRepos qualify as CompoundQueryables? > YEP! well sort of. Because we can't have multiple inheritance, they will likely not be compound queryables, but they will use them. See the patch in Bug 261460 for an example.
Bug 261758 includes a patch for non-transitive queries.
I am closing this bug as Fixed. I have set the target milestone to 3.5 M5.
