Just pushed to SVN (5280):
- Fixed some rounding errors in building evaluation that could lead to non-useless (but not very good) buildings rounding down to a value of 0 (could prevent myth builds in some circumstances)
- Fixed inappropriately tuned threshold on building value for AI to decide it's worth constructing by a unit (subdued animals mostly)
- Modified inappropriate base BTS code that reduced the value of unit-built buildings in small cities (made it think subdued animal builds were not worthwhile in small cities, even if it only had small cities!) [note - left old calculation intact for Wonders, which shouldn't target small cities usually]
- Tweaked priority of hunter unit builds to be more progressive (was very high for first then fell right off for the rest, up to the number it felt it needed)
- Fixed (or more precisely made work in all necessary contexts) code to decide to build exp-adding buildings before building hunters if the loss rate of hunters has been high recently
- Modified city build order priorities slightly to make very cheap production buildings pretty much always come first (i.e. - things like rock gatherers will now pretty much always be built before a food building, etc.)
- Possible fix for a recently reported CTD (weak evidence and no repeatable test case so cannot be certain)
- Removed the recently added code that prevented AIWeights having any effect now that the XML has been changed to mostly remove them
With this version I cannot now get to Sedentary Lifestyle first (on immortal, snail, large), or at least in the latest test game one AI beat me there by a LONG time (50+ turns) and at least one other also got there before me. As such I think it can be considered 'open for business' again, so feel free to report cases where the AI still seems to be lagging through prehistoric (preferably with save + AI logs)