Hidden windows don't receive events, so the only way to close is outside its handler, which there should be no cases of here. The function window_exists could return a false positive if a new window was created with the same pointer value as a deleted one. Will later disable window_exists function.
Replace with a disabled check for window_event_result::deleted, which will be enabled when all newmenu subfunctions return a window_event_result. There are currently no instances of a subfunction closing the window itself (which is preferred).
This ensures the event system will know what's going on. Also remove the 'likely' branch prediction macro from close_wall_window because close_wall_window will only be called outside of the wall dialog's handler (to open a different dialog or close the editor).
Make the listbox handlers and all sub-handlers (when_selected callbacks) return a window_event_result. Makes code more readable and removes one use of window_close, making it easier to inform event system if a window closes in future.
Return window_event_result::deleted if Game_wind was closed somewhere in response to an event handled by game_handler, so in future the event code can use that instead of window_exists to avoid tripping up.
gcc permits strlen("") in a constexpr context, but clang prohibits it.
Kreator reports that this broke the OS X build. Use sizeof(), which
produces approximately the same result and works in both compilers.
Reported-by: kreatordxx <https://github.com/dxx-rebirth/dxx-rebirth/issues/251>
Fixes: 63ca86698e ("Added communication for tracker written by A Future Pilot")
Kreator reports that clang warns that `i` may be uninitialized in
write_key_text. This is a valid warning, but gcc does not generate it.
Refactor write_key_text to eliminate the warning and simplify the code.
Reported-by: kreatordxx <https://github.com/dxx-rebirth/dxx-rebirth/issues/252>
Fixes: 20776fcfbc ("Use range_for/vcwallptr for write_key_text")
Some custom levels attempt to use texture overrides that reference
textures that are not yet paged in. It is not legal to access the data
of a paged-out texture, but ogl_loadbmtexture_f attempted to access it
anyway, causing a crash. Page in the texture before calling
ogl_loadbmtexture_f. Old versions of the engine would have cached
garbage data when this happened.
Reported-by: kreator <https://github.com/dxx-rebirth/dxx-rebirth/issues/228>
Fixes: 3c20c24ac0 ("Disable piggy_bitmap_page_out_all")
Mako88 reports that e335cdacb3 broke
reading historical high score files.
Code analysis shows that the size of the high score file must match the
size of struct all_scores, but this was not enforced anywhere. Commit
e335cdacb3 changed the size of struct
all_scores. Restore it to its previous size, and add a static_assert to
break the build if anyone tries to change it again.
Reported-by: Mako88 <https://github.com/dxx-rebirth/dxx-rebirth/issues/241>
Fixes: e335cdacb3 ("Reverted 81d7db279b and rather removed the terminator byte from string length in nm_set_item_input(). This also fixes possible 9 character length for pilot callsigns. Had to fix up COOL_SAYING_LEN as it was not considering the termintor but considering the original D2 source, this was a mess to begin with (def = 50, text string = 60, input field len = 45). Made score saying 50 characters as this seems to be the maximum witdth without overlapping.")
Mako88 reports that 2a19da88d4 changed
robot movement. That commit reordered the operations in a way that
tends to truncate small values prematurely. Restore the original order
of operations.
Reported-by: Mako88 <https://github.com/dxx-rebirth/dxx-rebirth/issues/247>
Fixes: 2a19da88d4 ("Pass object_base &to move_towards_vector")
multi_do_reappear is overloaded to be used both on ghosts and on live
players. This causes a diagnostic because the *_id checks expect
exactly one type: either test for ghost or test for player, but not test
for and accept both. Open code the type check to support both.
kmatrix_draw_coop_item computes various values that are always the same
for every iteration of the calling loop. Expand it inline in its one
caller so that the caller can cache those values.
The only caller for get_multi_endlevel_poll1 was DoEndLevelScoreGlitz,
which only called it on an unreachable path. That path is now gone, so
remove get_multi_endlevel_poll1.
Commit 63ca866 reduced the number of elements in UDP_Socket[] from (2 +
tracker) to (2), but added a write to UDP_Socket[2], which is now out of
bounds. Fortunately, UDP_Socket[2] is never read. Remove the
initialization of it.
Fixes: 63ca86698e ("Added communication for tracker written by A Future Pilot")
Conflicts:
common/arch/sdl/window.cpp - Remove unused EVENT_WINDOW_CLOSED and w_callback local var
common/ui/dialog.cpp - Remove unused EVENT_WINDOW_CLOSED
Rename symbol WORDS_NEED_ALIGNMENT to DXX_WORDS_NEED_ALIGNMENT to show that it is a DXX
symbol, not one inherited from a library.
This is a mostly automated transform, but the changes to SConstruct were
manual.
git grep -wl WORDS_NEED_ALIGNMENT -- '*.h' '*.cpp' | xargs sed -i -e 's/^\s*#ifdef \(WORDS_NEED_ALIGNMENT\)\>/#if DXX_\1/' -e 's/\s*#\(el\)\?if \(.*\)defined(\(WORDS_NEED_ALIGNMENT\))/#\1if \2DXX_\3/' -e 's/^\s*#ifndef \(WORDS_NEED_ALIGNMENT\)\>/#if !DXX_\1/'
Fixes: 435aa5a020 ("Move USE_UDP to dxxsconf.h; rename to DXX_USE_UDP")
Fixes: 7c8c2b5aa4 ("Move USE_TRACKER to dxxsconf.h; rename to DXX_USE_TRACKER")
Saved games dump robot ai_local state for all objects, regardless of
whether the object is a robot. Loading a game restored all ai_local
values, regardless of whether the object was a robot on restore. This
caused memory corruption in objects that were not robots when the
`ctype.`_T_ for that object was large enough to overlap
`ctype.ai_info.ail`.
Fix this by loading the ai_local into a junk buffer when the object is
not a robot. It would be better to skip over the bogus ai_local
entirely, but the structure of ai_local_read_swap makes that difficult.
Use the easy fix for now.
Kreator reports that "Pyramids of Luxoran" crashes due to a valptridx
sanity check. The level incorrectly attempts to open a wall on a
segment/side which has no wall.
Add an error check and a con_printf suggesting that the level author
needs to fix the invalid trigger.
Kreator reports that a level from back when Descent was new fails the
test for bogus matcen triggers. This level cannot be played when that
test throws an exception, so downgrade the exception to a CON_URGENT
message. This will hopefully deter level authors from creating any new
levels with this problem, but will allow users to play existing levels.
