Add coverity[GLOBAL_INIT_ORDER] annotations to ~50 global variables
across 27 files. These were all analyzed and confirmed safe:
- ~48 OptionBuilder::finish() globals use a Meyers singleton
(OptionsManager::instance()), which guarantees safe init order.
Destruction is also safe due to shared_ptr ref-counting.
- 6 default-constructed container globals (vectors, maps) have no
cross-TU dependencies at construction or destruction time.
Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>
When searching to see if there is an extension mismatch for a soundtrack audio file, go ahead and substitute the mismatched extension if found. Matching extensions is important, so continue to display a Warning; but for the benefit of players and out-of-maintenance mods, it is more friendly to play the mismatched file even if the audio data is inaccurate.
The warning for mismatched names-and-indexes implicitly assumed that retail-style sounds would only be added with numeric indexes. Since adding named sounds in the retail style is, in fact, allowed, update the logic to not warn in that case.
Fixes#6354.
Fix a small oversight from #6112. In the empty/placeholder indexed sound situation, although the sound should not be appended (since it indexed), it should still be treated as a new sound, in order for all the first-parse and first-initialize logic to run properly. Set the right value and add some comments to illuminate the logic.
Fix the inadvertent swapping of the two utility functions for parsing sounds. While we're at it, streamline a few lines, remove a few C-style casts, and remove an unused function prototype.
Follow-up to #6314. Fixes a crash encountered by Inferno on 24.2-RC1.
Although sound indexing is now done by name, mods which index by number (sc. a number that is treated as a name) expect the indexes to be in numeric order. When multiple levels of -snd.tbm files add mixed retail-style and new-style sounds, this assumption can be violated due to named sounds being placed into indexes that later files expect to be available.
This PR refines the parsing so that sounds which use a numeric index can pre-empt existing named sounds that have been assigned to that index. Since named sounds can go anywhere, they are just moved to the end of the list. This allows the ranges 192-199 and 202-249 to be available for numeric indexes as expected under the old style. MediaVPs named sounds which were previously parsed into those indexes are bumped to the next available index.
Additionally, this PR doubles the range of numbers that are treated as indexes, to more realistically accommodate additional numbered sounds. (The number/index distinction is rather arbitrary and is primarily used to avoid creating large sections of unused sounds if a mod were to specify a very large number, e.g. 10001, for a sound name.)
Finally, this changes parse warnings to use `error_display()` so that the warning displays the parsed file and line number.
* coverity - using `std::move` instead of copying, and change some parameters from value to reference
* coverity - fix or mark potential copy-paste mistakes
* coverity - various checking of parameter and return values
* coverity - miscellaneous fixes, and change some 'intentional' markers to 'false positive' markers
Patches the gamesnd system in several ways to properly fix the bug identified in #6196. Due to the way the new sound system was written, sounds with numeric IDs were not necessarily added in their indexed location, even though the code assumed (even prior to #6112) that this would be the case. The code now uses the proper index for both old *and* new style sounds. This is done by ensuring, in `gamesnd_parse_entry()`, that the vector contains an entry for the specified numeric ID before doing a search for that ID.
Due to the slightly modified population of the gamesnd vectors, the `Snds_iface_handle` vector is now populated after parsing the various sounds.tbl files rather than during parsing.
Separately, this makes consistent the handling of the `sound_entries` vector which could potentially be empty but also treated an empty state as an assertion failure.
When game sounds were changed to allow named entries, it became possible for named entries to unexpectedly replace numbered entries. This was noticed most obviously in the ballistic rearm sound effect, which started using explosions rather than the standard rearm sound effects. This occurred because MV_Tiny and MV_Small were added into indexes 200 and 201.
To restore the original behavior, the code now prevents sound effects added via TBM from being placed into reserved indexes. The reserved indexes are instead assigned placeholders, which can subsequently be updated or left alone as the modder prefers.
This also makes the code a bit more robust including patching some sound validity checks in `gamesnd_game_sound_try_load` and `snd_load`.
* add extra ways to modify background elements in missions
This adds Lua API methods for getting suns and bitaps by index, as well as modifying their scale and position. This removes the need to completely add and remove a background element that only needs to be modified.
* big pruning of (int) casts
* Create the ingame options game state
* Base working version of in-game options
* Cleanup + Optimization
* Localization pass
* more cleanup + documentation
* tooltips on option hover
* documentation and better inbounds check
* match case
* format strings instead of direct strings
* unused parameter
* address feedback
* localization id conflict with master
* Add this line back in
* pass the new option flags as well
* resolve xstr id conflict
Ever since retail, the code has assumed that a soundtrack would always have arrival music. This removes that requirement. If arrival music is missing, it will just be skipped.
1. Update soundtrack message to make it easier for modders to track down missing files
2. Update sexp help for ai-stay-near-ship to show the right argument count
The fix in #5578 assumed that a previously playing soundtrack would properly set the timestamp, but if no soundtrack was previously played at all, this wouldn't happen. So if we deliberately don't play a soundtrack, be sure to set the timestamp as if `event_music_do_frame()` had been called.
Fixes#5705.
This adds several fixes to make the event and menu music more robust:
1. The game will now run with zero soundtracks. All accesses into a potentially empty vector should now have checks.
2. Soundtracks and menu music are now checked for validity before being played (not just when importing a FSM mission).
3. The parsing validity checks and the in-game event music now check for skippable patterns using the same function.
4. Zero measures and zero samples-per-measure now cause the game to skip a pattern. Negative values now trigger a warning.
5. Omitting some or all of the `$Name:` lines for a soundtrack will not cause an assert.
Fixes#5576. Follow-up to #5572.
There is a built-in delay for the event music to start playing, but this delay gets stepped on if `change-soundtrack` is called when the mission first starts. So don't set the delay unless it has already elapsed at least once before.
Since event music soundtracks use the exact file extension specified (see #5567), the validity check should also use the exact file extension specified.
Also, move the validity check to after parsing has completed because extensions may be overridden.
This allows warp parameters to be specified in species_defs.tbl, rather than requiring them to be specified for each ship in ships.tbl. Since this introduces a cyclic dependency between species_defs.tbl and sounds.tbl, the parsing of sounds.tbl is broken into two stages.
Implements request from #5283.
When iterating through `Ship_obj_list`, check the `Should_be_dead` flag before using the object. This fixes#4990 and adds defense-in-depth checks to guard against similar errors elsewhere.
I considered moving the `Ship_obj_list` removal from `obj_delete` to `ship_cleanup`, but decided it would be prudent to avoid fundamental changes to lifecycle management.
Follow-up to #3743. A sound that was originally parsed as empty, but then re-parsed as non-empty, would still have the `GAME_SND_NOT_VALID` flag from the original attempt. This fixes that by clearing the flag each time the filename is read.
When the code checks for the existence of an audio file, be sure to check all relevant extensions.
This fixes#4495. Tested with both ST:R personas as well as the Vasudan persona.
A few assorted fixes for timestamps:
1) The pre-player delay should be fast-forwarded by 1/4 second per frame regardless of time compression
2) If time compression is very slow, there could be a 0-millisecond change in timestamps, so don't adjust this
3) Properly handle a change in soundtrack that occurs during a mission, not just when the mission is parsed
The timestamp fix/cleanup in #3878 revealed a longstanding bug with music timestamps--the timestamps were never set to something valid later.
This is a widespread issue in the code, fortunately Goober is modifying the timestamp code to accommodate invalid timestamps.
Tested and works as expected. Fixes#3930.
BMagnu