search

WillFlame14 / hanabi-bot

A bot that plays on the hanab.live interface.
GNU General Public License v3.0
15 stars 9 forks source link

Fix clue issues #23

Open RandoPerso opened 1 year ago

RandoPerso commented 1 year ago

Version/Most recent commit (check git log): 5a7175266c483e99af0b24b2f235ca3de7161d8b Convention settings: HGroup 5 Steps to reproduce or replay link: https://hanab.live/shared-replay/979042#29 Additional information: g5 is inferred, why was a fix clue given? image this is a fix clue on p3, why was a collateral card given an inference? The bot correctly identifies this as a fix clue, but using the hand command reveals it also added p4 as an inference, possibly from gtp? image

WillFlame14 commented 1 year ago

The fix clue is no longer (needed to be) given after https://github.com/WillFlame14/hanabi-bot/commit/f286aaad1d79ca342e35170973a0b4934cfd9325.

I believe Fix Clues always have touched cards follow GTP (see this example, where b1 isn't marked as a possibility), which is why that card ended up being inferred as p4. I could forbid Fix Clues that look like they reveal another playable card, or simply remove GTP if it seemingly reveals another playable.. not really sure what is better. Maybe even a new system where it "waits" a round for it to be fixed and otherwise assumes GTP.