update fix/general-strike #836
Conversation
make less aggressive should close #3779
| local bld = dfhack.items.getHolderBuilding(v) | ||
| if bld and bld:isFarmPlot() then | ||
| v.flags.in_building = true | ||
| for _,i in ipairs(bld.contained_items) do | ||
| if i.item.id == v.id then | ||
| i.use_mode = 2 |
There was a problem hiding this comment.
I take it you discovered the seeds had the wrong use_mode?
There was a problem hiding this comment.
in rare cases, yes
when seed planting starts, the seed is added to the building's inventory vector in use mode 0 (the same as for any workshop, and farms do act like a workshop in this regard), with in_job and in_building set. when the plant seed job completes, the in_job flag is cleared and the seed's use mode changes to 2
the problem that this tool is seeking to correct is when the plant seed job is cancelled incorrectly. normally, when the job is cancelled, the in_job flag is cleared but the use mode is unchanged. in most cases the seed is immediately removed from the building and everything is fine, but in uncommon cases the seed gets stuck in the building. i'm guessing that there's a cancellation code path in toady's code that doesn't cancel plant seed jobs properly leaving them in a limbo state where they can neither be removed nor completed, which results (i suspect) in planters spamming trying to replant the seed but not being able to do so because it's locked into that building somehow. my assumption is that toady has (as he is wont to do) cut and paste job cancellation code into diverse locations, and updated some but not all such cancellation pathways in a way that has an differential impact here
the script attempts to detect and fix this latter case by changing the seed to "planted", but the version putnam provided is both too aggressive and insufficient: it tries to change them to "planted" while they're still being planted, and it fails to change their use mode to 2, which results in the game being inconsistent about whether the seed is actually planted or not. this results in overplanting the field (i've seen them shove as many as nine seeds into a field with four spaces), and inconsistent on-screen displays it appears that the seeds in use_mode 0 actually do grow and eventually get harvested, but they don't look like they're in the farm and are not, at least for some purposes, treated as if they are in the farm, and i can understand why players feel like their seeds have disappeared because that's what it looks like on the map view
the determination that use_mode 2 was required was based both on experimenting with poking other use_modes into the inventory vector using gm-editor and by watching seeds that appeared to be planted correctly. every mode other than mode 2 yielded the "purple infinity" on the farm plot detail view, while only mode 2 resulted in the map view showing the plot as occupied, and only mode 2 prevents the dwarfs from trying to plant another seed in the same spot
i ran a test fort for two seasons with the code above, during which i saw only a very small number (but not zero) of corrections and saw none of the weirdness i was seeing with the previous version
make less aggressive
should close DFHack/dfhack#3779