Add assertions to the model parsing macros to guard against accessing out-of-bounds memory. Patch the model parsing routines to avoid going out of bounds, even if the out of bounds situations are temporary and benign.
* Update nightly build to use windows 2022
* Update build-release.yaml
* Update build-test.yaml
* Update test-pull_request.yaml
* Upgrade to CMake 3.21+
temp commit
* remove cotire due to incompatibilities
* Update dependencies to compile on CMake 4 with no warnings
* Update ANTLR4
* Update GTest
* Fix Windows CI
* Fix case sensitive ANTLR include
* Fix naming issues and disable shared ANTLR+
* Include Mess for MSVC
* fix new windows signed/unsigned compare warning
* FRED2 linking settings
* Don't clang tidy gtest
---------
Co-authored-by: Tyler Blount <[email protected]>
FSO doesn't appear to use this any longer. And even when it was enabled the
requirements made it largely useless anyway. Removing to clean up cfile a
bit and so we don't have to keep maintaining it.
Some QoL improvements:
1. Update `$fvec` and `$uvec` parsing to also allow parsing `$rvec` and to allow only `$fvec` to be specified; also make parsing more robust against failure
2. Allow `$fvec` and `$uvec` to specify the "facing" direction if `$look_at_offset` is not specified
3. Add a few additional warnings for incomplete or invalid POF data
Although normalized vectors in FSO are supposed to stay normalized, occasionally small floating point errors can pop up. In most cases these do not cause a problem, but in certain cases these errors can rapidly accumulate and corrupt an otherwise well-behaved matrix. This can most easily be seen with daisy-chained dock sequences or docked ships with angled dockpoints.
This PR audits the usage of `vm_vector_2_matrix_norm` to re-normalize vectors in cases where floating point errors could accumulate. There are also a few new Assertions for extra vigilance.
Follow-up to #6596. Fixes#6634.
Most places in the code that use `vm_vector_2_matrix` do so on already-normalized vectors. Changing these to `vm_vector_2_matrix_norm` avoids redundant normalization.
1. Trim trailing whitespace on any property values, just as leading whitespace is trimmed. All places where `get_user_prop_value()` is used implicitly or explicitly require this.
2. Warn the user if an invalid `$special` value is parsed. This would have caught #1, and will now catch related errors, such as misspellings, in the future.
3. Enhance `string_lookup()` with the ability to print the list being searched, if desired. Remove an unused version of `string_lookup()`.
4. Check for lower-detail submodels being recognized as subsystems.
* Fix incorrect ship legacy particle curve assignment
* Setup trigger radius
* Properly attach particle sources to objects and subobjects
* Add trigger velocity for weapon impacts
* Onetime particle source timing fix
* Warnings
* Allow global_to_local to work with no submodel
* Allow global_to_local to work with no submodel, again
Several changes related to the volumetrics model fix:
1. Update the `model_load()` prototype with better parameter names, including a new `ErrorType` enum defined in pstypes.h
2. Update the `MIN`, `MAX`, and `CLAMP` macros, all of which were susceptible to performance issues because they duplicated their tokens. This could cause functions or calculations to be re-run unnecessarily. Since the codebase is now on C++17, `std::clamp()` can be used.
3. Update the rest of the codebase to fix any associated issues.
Make these convenient functions even more convenient:
1. Change `object_get_model()` to `object_get_model_num()` and `object_get_model_instance()` to `object_get_model_instance_num()` to more accurately reflect what they do
2. Add new `object_get_model()` and `object_get_model_instance()` functions to return `polymodel*` and `polymodel_instance*` values.
As discovered by WouterSmits, campaigns with periods in their filenames caused issues with FSO's campaign handling. Specifically, in The Babylon Project, the EA-Minbari war v1.0.fc2 campaign would not save progress because FSO expected EA-Minbari war v1.fc2 in certain places. This is because the `_splitpath()` function attempted to remove an extension from a filename that already had it removed. For most campaigns this did not matter, but WouterSmits found a campaign where it did.
To solve this, there is now a new `util::get_file_part()` function to remove the path without removing the extension. This replaces `_splitpath()` for campaign files, as well as the similar `clean_filename()` and `GetFilePart()` functions in other places. All code paths were inspected for any chance of inadvertent extension adding, and Assertions were added to verify future code. The only potential problem area was `player:loadCampaignSavefile()` which now pre-removes the campaign extension if needed.
Model textures that are too long will cause a very unfriendly and hard-to-track down "ERANGE: String" error. The situation can be greatly improved by checking the string length ahead of time, printing a friendly Warning, and truncating the filename. This will alert the modder to the source of the problem while also allowing gameplay to continue.
* base working version of unlimited debris types
* Proper FRED support
* make qtFRED handle asteroids properly
* fix qtFRED assertion
* further clean out MAX_RETAIL_DEBRIS_TYPES
* These should be sizes not POFs
* Move this to it's own method
* convert asteroid subtypes to a vector
* unlimit asteroid subtypes
* fix some global shadowing
* signed/unsigned mismatch
* Address feedback
* Add control_info driver sources
* Properly deal with cockpits and their 'objnum's
* Add driven animation sets to weapons and cockpits
* enumify objnums
The general expectation is that model filenames can be specified as 'none' or '<none>' for no model, but as discovered by no tea, this was not consistent across the code base. For example, `$Model File: none` was handled as expected but `$External Model File: none` was not. This PR makes the handling consistent for all ship, weapon, and shockwave POF files, and also adds some extra checks when models are loaded.
In consultation with @BMagnu, code that validates data in the POF file itself should happen in `read_model_file_no_subsys`, as virtual POFs rely on their source files having correct data. I've added comments and slightly rearranged things to make this clearer, and also moved the new code from `read_and_process_model_file`. Additionally, I fixed two bugs by changing `parent > pm->n_models` to `parent >= pm->n_models` and adding a newline to the `nprintf`.
1. make `check_ok_to_fire` return a boolean (like `check_los`)
2. make `check_ok_to_fire` and `check_los` both use a position argument since we don't need the whole turret
3. remove the redundant position prediction code from `ship_get_global_turret_gun_info` (since that needs to be a "pure" function) while making sure "share fire direction" still works
4. skip calculation of the vector in `ship_get_global_turret_gun_info` if we don't need it
5. move the position prediction code out of `aifft_rotate_turret` and into its own function
6. prune some gun position calculation calls since in most cases one will suffice
7. fix position "jitter" with weapon subsystem damage in missions that use gravity
8. `model_rotate_gun` now uses a nullptr position for a reset instead of an extra boolean argument
9. use a more efficient method to calculate gun barrel length
10. a couple variable name changes for clarity
* Add global to local for matrices
* Add lua integration
* Get Submodel from Submodel instance
* Add submodel functions
* Add model instance from submodel instance
* Fix index
* Fix copy paste error
BMagnu