Ticket #179 (new defect)
Silly semeai reading with double-ko
| Reported by: | arend | Owned by: | gnugo |
|---|---|---|---|
| Priority: | normal | Milestone: | 3.9.x |
| Component: | source | Version: | |
| Severity: | normal | Keywords: | |
| Cc: | patch: | no |
Description
In the attached position, deep_semeai_reading.sgf, the semeai reading produces a non-sensical reading once one of the two groups can only be killed by filling in one of two kos.
This leads to an assertion failure when compiling with MAX_BOARD = 9 due to running out of MAX_STRINGS, but thats a little besides the point.
deep_semeai_reading-2.sgf shows the variation that leads to a crash.
Attachments
-
deep_semeai_reading.sgf
(200 bytes) - added by arend
3 years ago.
- Game position
-
deep_semeai_reading-2.sgf
(2.1 KB) - added by arend
3 years ago.
- Semeai reading variation leading to crash (start at move 22, always take lowest variation)
-
deep_semeai_reading-3.sgf
(225 bytes) - added by gunnar
3 years ago.
- Another position where the semeai reading is deep enough to crash with MAX_BOARD = 9.
Regression Results
| Attachment | Rev. | PASS | FAIL | Nodes | Status |
Change History
Note: See
TracTickets for help on using
tickets.
