On 18 May 2014 04:54, chris beck <beck...@gmail.com> wrote: > So despite what the code comments say, we are continuing to use the legacy > arithmetic in more places than just the terrain_builder code. > > It seems like for the most part this function's main responsibility is > deciding which way units will face when they move, attack, etc. > > Anyways I would like to propose that in 1.13 we phase out all of the > legacy arithmetic -- if it's absolutely necessary to preserve the current > behavior of the terrain_builder object, then maybe we can write similar > characterization tests for it and then try to refactor using the regular > vector arithmetic. (Hopefully that might not be necessary though -- I > imagine that the last time this stuff was changed, the legacy ops were > preserved because of a nearing release / freeze of some kind or something?) >
Hi, I kept the "legacy" ops long tiem ago because simply fixing things broke terrain_builder badly (terrains were being placed / rendered incorrectly). IIRC I concluded that the bad assumptions the code made were worked around in the actual terrain data files, and decided to just leave the old-broken map_location operations there since that was far out of the scope of the editor reworks I was doing, and the code in terrain_builder was, uh, uninviting. In the end I may have only switched things in close proximity to the editor and brushes there, where I knew what was going on and knew I wanted proper map_location algebra. Switching to the "correct" operations everywhere will probably take some work, but might be for the best. Regards, -- ilor
_______________________________________________ Wesnoth-dev mailing list Wesnoth-dev@gna.org https://mail.gna.org/listinfo/wesnoth-dev
