Gimmer / 27 / resource scheduling
Same work.
Different piles.
A task needs a whole bundle of people, for its whole duration. Move the priorities. Change the crew. See which blocks can actually fit.
Six fictional refuge tasks. Explicit prerequisites. Every displayed schedule is checked. Changes stay in this page and its link; nothing is uploaded.
Compare an extra builder, an extra climber, then both. Capacity only helps when the other constraints allow it.
Explore the crew choices ↓03 / Read the pile
Time rises. People are columns.
Select a coloured block to inspect the whole task. Empty cells are spare capacity. Scroll sideways to see every column.
Selected task
Frame assembly
Why this result is credible
CheckingThese are distinct results of all legal priority orders. Schedules with deliberate extra waiting are not listed.
The logic under the blocks
Haul → prepare the site → assemble the frame → fit walls and roof. The emergency cache needs the haul, but sits in a separate prepared recess; it can be fitted independently of the shelter. Completion means all six tasks have finished.
| Task | Days | Whole crew needed | Must wait for |
|---|
No task is interrupted. Resources renew when a task finishes. Whole-day durations are illustrative. Weather, fatigue, material shortages and engineering design are outside this small model.
Keep an inspectable result
The link restores your inputs and priority. The JSON export includes those inputs, the schedule, named resource assignments and verification results. A visible copy remains available if your browser cannot save a file.
This is a snapshot of the exported schedule. Changing the model clears it; export again for a fresh result.
Method, scope and lineage
A resource heap you can interrogate.
Scheduling first
This is the single-mode resource-constrained project scheduling problem: fixed task durations, renewable capacities and finish-to-start prerequisites. Each legal priority order is turned into an earliest feasible schedule. With just six tasks, we can try every such order and report a minimum finish for this model.
The method follows Karapetyan & Vernitski’s serial schedule generation scheme. A separate checker replays coverage, timing, prerequisites and resource use.
Geometry with a precise meaning
The display keeps the original Tetris-like idea. Every rectangle occupies one named resource unit over a verified interval; the pieces belonging to one task stay together in time. Physical prerequisites and chosen resource orders are different constraints.
The drawing is inspired by Gaubert & Mairesse’s task-resource heaps. This pooled-resource scheduler does not claim to implement their max-plus heap product, or a Petri-to-SMC translation.