This isn't a complete refwctoring, just the part than can be done with
transparetly correct capture of inlinre logic into macros.
No logic changes. Tests pass, 100% coverage.
Note: there is a tiny risk that this could break savefile
compatibility, as I swiped an unused int member in the game structure
and turned it into a bool. int and bool *should* be the same sixe...
There was a very old bug, probably predating the OpenAdventure port,
that would poke a stashed value of -1 into the snake object if you did
a save in endgame, and then fail the savefile validation on resume.
This was masked for a long time by a bug in put() just fixed a couple
of revisions ago.
To enable use with online Bulletin Board Systems (BBSes) where users
may be disconnected unexpectedly, but would naturally want to resume
playing their same game, added support for an optional save game
path/filename to be specified on the command-line (very similar to
"-r <filename>"), except this save/restore file is:
1. automatically loaded/restored if it exists
2. automatically created when starting a new game
3. automatically updated when exiting a game for any reason
4. cannot be changed to a different path/filename by the user
Since a BBS server program can be expected to send a SIGHUP or SIGTERM
to the game process upon user disconnection (or timeout), those
signals are caught and a graceful termination will occur which saves
the current game state.
Build with ADVENT_AUTOSAVE defined to enable this option.
BUG:
The 'info' command still reports the save/suspend/pause commands as
valid, though they are not when this build option is used (same is
true of ADVENT_NOSAVE, and that doesn't apparently bother anyone).
Note. because we used fixed-length declrations in the right places, this
shouldn't break saves.
(Besides being right, this will someday make a port to Go a touch easier,
if we decide to do that.)
