Motivation / Problem

peter1138 found a problem while investigating a pathfinder issue, which is that stopped/stuck trains do not correctly update their viewport hash and thus partial sprite problems occur. Example image attached.

Description

The viewport hash calculation is set up to use the sprite cache from before the most recent bounding box update. This was done to fix a specific problem:

• When the vehicle is first placed on the map, it has no valid sprite cache.
• A valid sprite cache is needed as a starting point, so in this case it is set to the current coordinates.
• Because this will result in a new vehicle having "no updates" (current position = cached position) and thus not getting a viewport hash calculated, the coordinates from before this update are used.
• In most situations the coordinates being one vehicle movement step out of date is not noticeable, all that happens is the dirty rectangle is slightly larger than necessary.
• However, when a vehicle is reversed twice without otherwise moving, the co-ordinates of the train's current position will match the co-ordinates of its position from 2 moves ago, not 1 move ago. This causes the viewport hash to become out of sync with the actual position of the vehicle, and thus vehicles to not be drawn when a region becomes dirty.

This fix corrects the problem by only using the "old" position in the one case it is actually needed, when it contains an invalid coordinate.

Limitations

I've run my usual series of checks - against the game demonstrating the issue, against some Timberwolf's Trains games from the original PR, against the abuse.grf size-changing train, and against ships doing traversing through rotations of >180 degrees on the spot.

No regressions observed, although these are only visual checks against currently known scenarios.

Checklist for review

Some things are not automated, and forgotten often. This list is a reminder for the reviewers.

• The bug fix is important enough to be backported? (label: 'backport requested')
• This PR affects the save game format? (label 'savegame upgrade')
• This PR affects the GS/AI API? (label 'needs review: Script API')
  • ai_changelog.hpp, gs_changelog.hpp need updating.
  • The compatibility wrappers (compat_*.nut) need updating.
• This PR affects the NewGRF API? (label 'needs review: NewGRF')
  • newgrf_debug_data.h may need updating.
  • PR must be added to API tracker