Follow-up Comment #2, patch #3839 (project freeciv):
Heh, perhaps my definition of "lightweight" was adjusted by the remainder of
the nativity stuff :) I've no objection to splitting this into bits: it was
assembled from bits as I tried to clean up nativity everywhere: these were the
ones that were tiny patches, showed no behaviour changes in autogame testing,
and didn't trigger me to add TODO items reminding me to investigate later.
For base_assess_defense_unit(), I'm very certain of the semantics, and there
are a few other places in the code that have had that change made (this one
appears to have been missed).
For assess_danger_unit(), there are no precedents that I found, but the
behaviour is precisely the same for rulesets without UMT_LAND units granted
CanAttackNonNative (and the change seems sensible).
For assess_danger(), I'm less certain, and it may be worth performing autogame
testing with rulesets specifically designed to hit this condition (presence of
UMT_LAND units with !CanOccupy and UMT_SEA units with CanOccupy).
For kill_something_with(), I was pleasantly surprised not to have differences
in autogame testing: I was expecting this to fall into the more complicated
category. Of everything collected here, I'm least confident in this change.
For dai_rampage_want(), I'm pretty sure the current code won't work usefully
even in the presence of "Big Land" from the experimental ruleset, ignoring
more interesting nativity (e.g. Hut on Mountains). That said, I'm not
absolutely sure my solution happens to be correct.
For the SDL client fix, it's mostly a random swipe to remove it from grep
results. That function should really be rewritten entirely to do things
correctly, and if the nativity issue is a good reminder about that, there's no
reason to patch it away.
Reply to this item at:
Message sent via/by Gna!
Freeciv-dev mailing list