3 Commits (9df4ede3cc12e9285a402d4e77fe4eea75ea5606)

Author	SHA1	Message	Date
ulferts	33081e1f97	increase performance of scheduling by avoiding paths aggregation ... It is no longer necessary as we are no using `UNION` and not `UNION ALL` so that even if a path is traversed back again, that will not lead to an infinite loop since the `UNION` will work as a DISTINCT canceling out paths that we have already traversed	4 years ago
ulferts	655f021b92	add derived dates to the work package resource ... The derived dates are the minimum/maximum of the descendant's dates. As long as the work package is scheduled automatically, the derived and non derived dates should be equal althoug the derived dates will be null in case non descendants exists. But if the user decides to activate manual scheduling the dates can deviate. The derived dates can then help to display the deviation to the user. This commit however does not alter the front end. The dates are eager loaded on the work package index commit using the same mechanims already introduced for spent time. The patched cost attributes, which are also eager loaded, are now eager loaded in the core as well so the patch is removed. As those attributes are now part of the core (although in a different module) it no longer makes sense to rely on patching for this.	4 years ago
ulferts	7065c162fd	disregard manually scheduled nodes when scheduling work package graphs ... Instead of fetching all work packages from a subset of the whole graph constructed by hierarchy and follows relationships upon scheduling, only work packages eligible for automatic rescheduling are returned. In order to avoid quering the database multiple times, the whole subgraph is returned at once using a recursive CTE.	4 years ago