Fixed another couple of bugs revealed during automated testing.
Been redoing flat buttons:
They are still far from perfect. Next step will be to change grey menu backgrounds into something stone/leather/plaster-like. Probably darker shade.
Changed new Map Editor terrains selection from buttons to ListBox. Looks and fits much neater:
Some terrains don’t have textures yet (and I also plan to at least double the count of terrains in future).
What do you mean by automated testing? Just unit tests? Or do you do something more?
I have a special tool to run just the games logic, without rendering. In those tests it set up a skirmish map between 4 AIs and simulated gameplay for 1.5 hours (which takes ~1 minute of real time). The game is peppered with many internal checks for impossible situations (e.g. 2 units being on the same tile). After the run I inspect the result (game not crashed) and stats, to see that all AIs have developed (not got stuck). Running these simulations in background for 20-30min allows to catch some bugs – e.g. when a unit dies of hunger when entering an Inn, or an AI tending to run out of timber, etc.
The game has little to no unit tests, since it takes a lot of time to rig and actualize them. Above approach works much better.
Awesome, thanks for the reply :).