Content
View differences
Updated by Jens Ulferts about 6 years ago
### Acceptance criteria
* Whenever a work package scheduling is triggered (creation, update by children, follows/precedes, ...)
* work packages that are manually scheduled are not updated
* work packages that are connected to a scheduled sub tree only via manually scheduled work packages are not updated, e.g.
* A precedes B (manually scheduled) precedes C (automatically scheduled) -> If A is updated, B and C are not affected
* A is child of B (manually scheduled) is child of C (automatically scheduled) -> If A is updated, B and C are not affected
* A is child of B (manually scheduled) precedes C (automatically scheduled) -> If A is updated, B and C are not affected
### Implementation notes
* Currently, whole subtrees are fetched from the database via typed\_dag. Doing this is efficient but disregards the attributes of the transient work packages. Upon fetching a subtree, whenever a work package with manual scheduling is returned, all transitively connected work packages must be disregarded.
* If possible, improve the performance of fetching the work packages
* Check using CTEs
* Check fetching fewer attributes
* Whenever a work package scheduling is triggered (creation, update by children, follows/precedes, ...)
* work packages that are manually scheduled are not updated
* work packages that are connected to a scheduled sub tree only via manually scheduled work packages are not updated, e.g.
* A precedes B (manually scheduled) precedes C (automatically scheduled) -> If A is updated, B and C are not affected
* A is child of B (manually scheduled) is child of C (automatically scheduled) -> If A is updated, B and C are not affected
* A is child of B (manually scheduled) precedes C (automatically scheduled) -> If A is updated, B and C are not affected
### Implementation notes
* Currently, whole subtrees are fetched from the database via typed\_dag. Doing this is efficient but disregards the attributes of the transient work packages. Upon fetching a subtree, whenever a work package with manual scheduling is returned, all transitively connected work packages must be disregarded.
* If possible, improve the performance of fetching the work packages
* Check using CTEs
* Check fetching fewer attributes