<URL: http://bugs.freeciv.org/Ticket/Display.html?id=39476 >
second batch: base_type_iterate client_options_iterate This seems better to remain an indexing iterator (almost all usage is the index number, no pointer needed), but the name translation and other standard accessors were needed. Several obscure bugs fixed! specialist_type_iterate Others that seem better to remain indexing iterators, side-effects removed: shuffled_players_iterate sorted_event_iterate (city.h) city_map_iterate_outwards city_map_checked_iterate map_city_radius_iterate cities_iterate (map.h) iterate_outward_dxy circle_dxyr_iterate adjc_dirlist_iterate whole_map_iterate Committed revision 13169. === At this point, all the "_index" overloading has been eliminated or qualified (usually _tile##_index, where _tile is a variable parameter). Of course, there are more that use other (unqualified) overloading. Both base and specialist had (minor) name translation problems -- converted to use the new translation structure. Also, in client/tilespec, citizen specialist names were accessed (they don't exist, so it generated a random memory location) and then a quick test skipped using the bad value. When I added more tests, and returned NULL instead of something random, it became a reliable crash! So, I changed the order of the enum, and run the for loop only up to CITIZEN_SPECIALIST. I don't see any problem with the change, but the code assumptions aren't well documented. So, we'll wait and see.... Meanwhile, I added some error checking in client/packhand for passed array sizes, matching similar stuff in server/ruleset. Lots more to do in that regard. Folks, you cannot trust data that comes over the network! _______________________________________________ Freeciv-dev mailing list Freeciv-dev@gna.org https://mail.gna.org/listinfo/freeciv-dev
