Openrct2: No entry signs ignored

[PhyxionNL]

OS: Windows 10
Version: 0.0.5
Commit/Build: 35ae7600f9647f7be01b4ec3f059f7061aa5f52a.

Guests go through the no-entry signs. This didn't happen before the pathfinding update (or at least I didn't notice it before).
Very easy to spot by running the game at turbo speed.

Screenshots / Video:

Save game:
Mines of Africa - Copy.sv6.txt

[IntelOrca]

@zaxcav Could you look at this?

[duncanspumpkin]

I made sure to check this when i was testing the update. I'm not sure how this has changed.

[duncanspumpkin]

Is this only happening on wide paths?

[zaxcav]

I'll have a look. I'm not yet familiar with how no-entry signs relate to the tile edges, so I'll need to get up to speed. At present the pathfinding search explores all set tile edges, so if no-entry signs are not reflected in the tile's set edges the AI can walk through them.

Edit: It should be taken care of by path_get_permitted_edges(), so I'll have to check that this is being used everywhere it should be.

[zaxcav]

Found the issue. It appears to be an original bug inherited from RCT2.
_peepPathFindIsStaff is not always set when necessary - only aimless peeps are affected.
I'll try to make a fix and submit a PR tonight or tomorrow.

[duncanspumpkin]

@zaxcav that sounds very unlikely. I might get time to fix it tonight if you don't get there first

[duncanspumpkin]

Bug has been in the game since i implemented the peep pathfinding last year. Issue caused by refactoring.

[zaxcav]

I've just committed a 1 line fix and was about to make a PR. Shall I continue or would you prefer to correct it?

[duncanspumpkin]

10 seconds too slow :(

[zaxcav]

That's ok. It's fixed, and that's what's important :-)