The "default" tech database flags indicate whether entries are in the tech room by default, that is when the table is originally parsed and the campaign is started. Flags may be changed later, so that when the tech database is "reset to default" the flags can be returned to the known state as specified in the tables.
A bug was introduced back in commit c924e3997a (!) based on a misinterpretation of what "default" meant. This fixes the bug and clarifies the flags.
Fixes#4390.
This takes a few of the existing hooks and converts them to the new style of hook definition which can include more extensive documentation and enumeration of supported hook variables.
This is an updated version of #3151 to work with the latest codebase.
NOTE: Before this can be integrated into the codebase, it needs an implementation for `bool script_state::IsActiveAction(int hookId)` which apparently only has support for old-style script hooks.
For script hooks which use conditions, be more consistent about evaluating both objects involved, as well as evaluating the different types of objects.
Fixes#3874.
Back in #104, I made an *attempt* at clearing the delayed SSM parsing data once it'd been validated; however, for some reason, I only cleared half of the variables, and I only used `.clear()` rather than `.clear()` followed by `.shrink_to_fit()` (which means the same amount of memory was still reserved for these variables that would never be used again). While making #4201 I happened to comes across the "swap with empty temporary" paradigm do to both at once, so I decided to apply it to this old thing, as well as clearing all four variables rather than just two. Additionally, since these mappings don't actually care about ordering, I made them `unordered_map`s while I was at it. It's probably not a significant memory savings, but since we'd never need to use that memory after parsing, I figured we might as well reclaim it anyway.
* beam_add_light_small() is only ever used in one place which always passes a point. Calculating the point inside is removed as redundant and obfuscating.
* Repeated beam light width calculation moved to function
* Repeated beam light state validation moved to function
* Repeated beam light noise generation moved to function
* cleanup superfluous parameter
Removed useless light_ignore_objnum and affected_objnum lighting data, light_add_point_unique() subsequently removed as well, and cleaned up unused parameters.
The `stuff_int_list()` call converts the weapon names into the appropriate index into the `Weapon_info` array; however, in `weapon_do_post_parse()`, we call `weapon_sort_by_type()`, which reorders the array; this makes these indices potentially point to the wrong weapon. Additionally, there's no protection against someone `+remove`ing one of the weapons in the list in a later TBM.
This commit makes it so we instead store the names into a vector of strings, then convert them into indices after `weapon_sort_by_type()` has already been called. Additionally, `Player_weapon_precedence` is changed from a fixed-size array to a vector, since it's usually not as long as `MAX_WEAPON_TYPES`.
Fixes#4199.
It's theoretically possible for `parent_type` to get out of sync with the parent's type, particularly if the parent is destroyed. Since the field is almost never used, it can be removed.
Should fix#4105.
There is a bit of code that detonates a missile if it misses impacting on a subsystem. On Nyctaeus's Arachnas model, missiles could reach the center point of the main beam turret submodel without actually hitting any part of the submodel. The fail-safe code would set the missile to detonate in that case, but on low-FPS systems the weapon would be able to actually hit the turret before it detonated. On high-FPS systems, there is a clean miss.
This reworks the code to delay the missile detonation until it reaches the other side of the submodel, so it has time to hit some other part of it. It also reduces the detonation time from 1/100 of a second (which is more than one frame on 120 FPS systems) to 1/1000 of a second (which is always less than one frame).
A standalone server can reset itself, which triggers game_level_close(),
before game_level_init() has been called. trails_level_close() didn't account
for not being inited first and would crash.
Move the initialization of `used_weapons` to `weapon_init` since that's where it belongs. Also call `weapon_level_init` in the lab since investigation revealed a few other things that needed to be done there.
* Switch Multimsg to use netsig for animations
* Unique-ify animation ID's
* Clang Tidy
* Implement first suggestions
* Implement suggestions 2
* Add get model instance safeguard
* version1 of 'head-on'lasers
* version 2 of head on lasers
* better handling of radius/length
* more adjustments and manual switchover angle
* add switchover rate and glow head on
* bugfix
* allow for instant switchover
* clean up and documentation for PRing
* misc fixes
* more fixes
* more fixes
* Scale distant combat related effects so they stay visible
- Effects scaled so that they always maintain a minimum pixel size
- Supports lasers, beams, all trails, muzzle flashes, and thrusters
- Disabled by default, but can be configured through the game_settings.tbl
* Added comments clarifying when min pixel size scaling happens.
* Convert some NULLs to nullptr to make CLANG happy
* Add custom data to ship tables
* add index check to shipclass.cpp
* fix hasCustomData name to use the proper casing
* Add custom data to weapon tables
* Add validity checks and remove superfluous empty check
* Attempt to appease clang-tidy
* Use F_RAW as recommended by @BMagnu
* Creates a function to parse maps and uses it in ship.cpp and weapons.cpp
* another attempt to appease clang-tidy
* Remove superfluous using
* Small style and grammar fixes
* Accidentally forgot this change. Rename 'delimiter' to 'end_marker'
This moves submodel rotation out of the AI pipeline and into object movement. With the exception of turret rotation (which remains in the AI code) this now allows all types of submodel rotation to happen in the same place. Many associated functions have also been cleaned up.
In addition to making the code cleaner and clearer, this will make it much easier for intrinsic rotations and new animations to be applied to non-ship objects, such as weapons.
* Deleted a line that blanked target_num on homing weapons but still let them home
* This whole block of code about missiles missing turns out to be a logical dead end anyway, ripped it out by the roots.
Goober5000