All releases to date have a bug where they treat certain segment number
fields as an int, not a segment number. Storing segment_none (0xffff)
into the save file causes affected releases to crash in various places
because it fails to recognize that this is segment_none.
Current code correctly treats segment_none as a non-segment and works
correctly without this hack. The hack is only required to get past
releases to work correctly after loading a saved game written by current
code.
All releases to date have a bug where they treat MarkerObject as an int,
not an object number. Storing object_none (0xffff) into the save file
causes affected releases to exhibit several problems. Most obviously,
they crash if the automap is opened, because they try to show
Objects[0xffff] as a marker, but no such object exists. Additionally,
they refuse to let the player drop a marker because none of the elements
of MarkerObject are -1, so they incorrectly tell the player that all
marker slots are busy. Finally, they will crash if the guidebot is
killed, because the guidebot death code deletes marker 0xffff because it
fails to recognize that this is object_none.
Current code correctly treats object_none as a non-object and works
correctly without this hack. The hack is only required to get past
releases to work correctly after loading a saved game written by current
code.
Fixes: 9125ae32cd ("Make objnum unsigned")
All releases to date have a bug where they treat
Dead_controlcen_object_num as an int, not an object number. Storing
object_none (0xffff) into the save file causes affected releases to
crash when treating object_none as a valid object number.
Current code correctly treats object_none as a non-object and works
correctly without this hack. The hack is only required to get past
releases to survive destroying the reactor after loading a saved game
written by current code.
Fixes: 9125ae32cd ("Make objnum unsigned")
Fixes bug where in the editor, you insert a robot with normal behaviour then play the level - it fails the check on the hide_segment index in init_ai_objects(). This is because hide_segment was written as the poisoned value of 0xfdfd if the behaviour of the robot is AIB_NORMAL - because hide_segment wasn't initialised. Therefore, always setting hide_segment to suppress this exception (it's an inexpensive operation seldom called).
zicodxx reported a problem where poison=overwrite builds caused robots
dropped by other robots not to spawn. I diagnosed it as a problem
caused by not setting `dying_start_time` and proposed using
`init_ai_object` to set this (and other) fields. Kreatordxx posted pull
request #311 to implement this for dropped robots. However, his pull
had the unfortunate side effect of zeroing the dropped robot's velocity.
This change is based on his pull, but with the `init_ai_object` call
moved higher so that the velocity is (unnecessarily) zeroed by
`init_ai_object`, then initialized to `new_velocity` by `drop_powerup`
(as it always was).
Reported-by: zicodxx <https://github.com/dxx-rebirth/dxx-rebirth/issues/293>
Patch-inspired-by: kreatordxx <https://github.com/dxx-rebirth/dxx-rebirth/pull/311>
This is where a client in a multiplayer game hitting a wall fails an assert. Initialise obj->mtype.phys_info.flags in multi_reset_player_object() instead of adding flags. (Tested and still works OK without specifying PF_THRUST.)
Initialise ConsoleObject->mtype.phys_info.flags in reset_player_object() instead of adding flags. The only place this was initialised properly was when reading the player object from disk.
Use std::equal_range to find the upper and lower bounds in a single
binary search, rather than relying on a linear search to find the first
sought element.
Also happened when a client to a multiplayer game dropped out due to some network error. Delay call of multi_leave_game() until responding to EVENT_WINDOW_CLOSE, so the game isn't in an unstable state between handling the network event and the game closing.
Commit 4cc801f changed `object_move_one` to return `window_event_result`
instead of `void` and added a default return value at the bottom.
However, it added the value inside a `#if D2` block, so the D1 build now
fails with:
similar/main/object.cpp: In function 'dcx::window_event_result d1x::object_move_one(d1x::vobjptridx_t)':
similar/main/object.cpp:1627:7: error: variable 'result' set but not used [-Werror=unused-but-set-variable]
similar/main/object.cpp:1866:1: error: control reaches end of non-void function [-Werror=return-type]
Move the return statement out of the conditional block to fix both these
errors.
Fixes: 4cc801f42f ("Remove calls to window_close(Game_wind) when game finished or over")
In multi_do_frame(), replace call to window_close(Game_wind) with returning window_event_result::close whenever multi_quit_game is true. Only using this return value where multi_do_frame() is directly called by GameProcessFrame(). multi_quit_game will only be set back to 0 when a new multi game is started.
Closing a window within its handler is problematic - it can result in an unstable state.
Replace call to window_close(Game_wind) with returning window_event_result::close to game_handler. Applies to when there is a failure in net_udp_level_sync(). Closing a window within its handler is problematic - it can result in an unstable state.
Replace call to window_close(Game_wind) with returning window_event_result::close to game_handler. Applies to whenever newdemo_stop_playback() is called. Closing a window within its handler is problematic - it can result in an unstable state.
`struct object_rw` is poisoned prior to initializing and sending it.
However, some fields are legitimately unininitialized (other than their
memset or poison value) at send time. Add and use a poison variant that
can clear those fields, without marking them unreadable.
This is in a redundant check, as we shouldn't (and don't) call newdemo_record_start_frame if nd_record_v_no_space is true, i.e. we ran out of disk space recording a demo. We want to return window_event_result::close with every call to newdemo_stop_playback() to close the game, as this is safer than calling window_close on the game window within its handler. In this case, it's much simpler (and safe) to just remove it.
Even if newdemo_record_start_frame is called when nd_record_v_no_space is true, we don't want to close the game, just exit the function.
Replace calls to window_close(Game_wind) with returning window_event_result::close to game handler. Applies to when DoEndGame() is called, DoGameOver() is called, aborting in the kmatrix screen (multiplayer game) during AdvanceLevel() and playing one demo frame causes playback to stop in GameProcessFrame(). Closing a window within its handler is problematic - it can result in an unstable state.
Old versions of Rebirth sometimes saved games with bogus object data,
such that `obj->type == OBJ_NONE`, but `obj->movement_type`,
`obj->control_type`, or `obj->render_type` were not their respective
*_NONE constants. This confused the game loader code such that it would
load invalid data from the `mtype`, `ctype`, or `rtype` unions and use
it without checking.
Try to counter some of this by exiting early if the object has type
OBJ_NONE.
Valgrind warns when writing uninitialized data to a file. The Descent
savegame format requires writing certain fields that are no longer used.
Set those fields to 0 to prevent leaking stack data into the file.
Levels which fit in MAX_SEGMENTS_ORIGINAL read exactly
MAX_SEGMENTS_ORIGINAL segments worth of light_subtracted, even though
fewer segments were defined. Prepare light_subtracted into a temporary
stack buffer instead of doing byte-at-a-time writes. Initialize this
stack buffer to 0 when necessary. Write the buffer once it is fully
prepared.
player_info::cloak_time is only defined if the player is cloaked.
player_info::invulnerable_time is only defined if the player is
invulnerable.
In both cases, the game save code tried to read the ::*_time variable
even when it was undefined. Modify the saving code to check player
flags before reading the associated timer.