I've made some changes to the Scenery handling. Basically just tidy ups.
The main difference is in tile.[ch]xx where I've changed list<fgFRAGMENT> to
vector<fgFRAGMENT>. Studying our usage patterns this seems reasonable.
Lists are good if you need to insert/delete elements randomly but we
don't do that. All access seems to be sequential. Two additional
benefits are smaller memory usage - each list element requires pointers
to the next and previous elements, and faster access - vector iterators
are smaller and faster than list iterators. This should also help
Charlie Hotchkiss' problem when compiling with Borland and STLport.
./Lib/Bucket/bucketutils.hxx
Convenience functions for fgBUCKET.
./Simulator/Scenery/tile.cxx
./Simulator/Scenery/tile.hxx
Changed fragment list to a vector.
Added some convenience member functions.
./Simulator/Scenery/tilecache.cxx
./Simulator/Scenery/tilecache.hxx
use const fgBUCKET& instead of fgBUCKET* where appropriate.
./Simulator/Scenery/tilemgr.cxx
./Simulator/Scenery/tilemgr.hxx
uses all the new convenience functions.
- use strings for fg_root and airport_id and added methods to return
them as strings,
- inlined all access methods,
- made the parsing functions private methods,
- deleted some unused functions.
- propogated some of these changes out a bit further.
- saving the entire vertex list for each tile with the tile records.
- saving the face list for each fragment with the fragment records.
- code to intersect the current vertical line with the proper face in
an efficient manner as possible.
Fixed a bug where the tiles weren't being shifted to "near" (0,0,0)
Working on rewriting the tile management system so a tile is just a list
fragments, and the fragment record contains the display list for that fragment.
data file rather than calculating it on the fly.
Fixed polygon winding problem in scenery generation stage rather than
compensating for it on the fly.
Made a fgTILECACHE class.
