GNOME Bugzilla – Bug 140561
Task links SF and FF are missing
Last modified: 2006-05-02 14:02:31 UTC
Seems OK with negative or zero lag but hangs (CPU high 90's which looks like the same use of CPUs as the task recursion bugs (when those occur)) when a positive lag input (>=1h for SF and >=1d for FF). There is no logical reason why you can't have task2 finish laging after task 1 finish if you have a task1->task2 FF relation.
Would be good to get this fixed for 0.12 since it's a new feature and it would look a bit bad if we just added a bug instead of a feature ;)
I can't redproduce any hang, bug the patch that added the other relation types is not entirely correct. It tries to use the start time of a task before it's actually calculated.
Ah - OK on my particular machine (2.6.5 kernel + SMP) combination maybe it just caused the program to go walkies. It was a consistent bug and poorly setup memory can cause indeterminate behaviour.
For 0.12, we either have to fix this or disable SF and FF. The fix is to add a function that calculates the start from the finish value with resources assignments/calendars in consideration. There is a function that does the opposite already, it should be pretty straight-forward to adapt that code.
I'll have a go at fixing this. Probably get completely lost on a voyage of discovery of the guts of Planner so I may be some time. Hopefully I find the issue at the tonsils as opposed to the r-send.
The problem is that task_manager_calc_relation() assumes that both start and finish is known, which is not the case. task_manager_calculate_task_finish() works from the start time to get the finish time, and we need task_manager_calculate_task_start() that works from the finish time for the cases where the task is anchored at the finish time.
I've disabled SF and FF for now, moving this off 0.12.
*** Bug 318465 has been marked as a duplicate of this bug. ***
SF and FF added. Will be part of v0.14 release.