[Bug] gamecrash mid map noShip::StartDrivingToHarborPlace

[codeboat-dev]

Return To The Roots v20260309-55154d71fc55d009a59af4ca00e4b25151098a98

At some point in MISS219.WLD (mission 10 of the Berciq "Roman Campaign II") game crashes. Tried 3 times, after some time 1 or 2 hours it crashes. There is an old todo in the code here

WARNING: Bug detected (GF: 67938). Please report this with the savegame and replay.
noShip::StartDrivingToHarborPlace: Schiff hat keinen Weg gefunden!
player 2 state 11 pos 127,125 goal coastal 141,105 goal-id 10 goalpos 141,104
RttR crashed. Backtrace:
0x10276acdc
0x102753b10
0x102759299
0x7ff8192b837d
0x6150040200000085
0x102781d27
0x102a13e34
0x1028236f9
0x102a61d3a
0x102a398d8
0x1029ea497
0x1029f0162
0x1029e31ed
0x1029e2ede
0x10277968a
0x102752c4e
0x2034c5781
RttR crashed. Please restart the application!

Savegame and map attached

Archive.zip

[Noseey]

Tried 3 times, after some time 1 or 2 hours it crashes.

One question: when you tried it 3 times, was it also (at least once) from a fresh start instead of loading a save? If so, did it crash also when no saving / loading was involved from start to crash?

So far, I tested and was able to reproduce the crash with the provided savegame in both the nightly builds v20260309 and v20260222 after a few minutes into the game. With the current master debug build (asserts enabled), it already crashes right away in the following assertion, thanks to the newly added StrongIdVector.h:

    typename Parent::const_reference operator[](index_t id) const
    {
        RTTR_Assert(id && id.value() <= this->size());
        return Parent::operator[](id.value() - 1);
    }

.. by trying to access harbor_pos with harborID 17 (seems to be related to position x = 127 and Y = 126) while the harbor_possize is only 16, which is very odd. I will try to see where it fails without assertions in debug mode, but this observation might already indicate something went wrong a while before the crash.

Edit:
So without Assertions, the Segfault causing the crash occurs for me in FindPathImpl:

.. as it seems related to the destruction of a building, here the call stack:

.. which seems to happen according the log after the noShip::StartDrivingToHarborPlace related message. Will see if I can find some more clues.

[Flamefire]

Found the issue: I basically broke loading newly saved save games. Fix in #1905, I hope @Flow86 has soon time to merge this.