- `speech`
- `ambient`
- `water-h`
- `vol-h`
- `generic-obs`
- `carry-h`
- `pilot-h`
- `board-h`
- `gun-h`
- `flut-h`
- `indax-h`
- `lightjak-h`
- `darkjak-h`
- `target-util`
- `history`
- `collide-reaction-target`
- `logic-target`
- `sidekick`
- `projectile`
- `voicebox`
- `ragdoll-edit`
- most of `ragdoll` (not added to gsrc yet)
- `curves`
- `find-nearest`
- `lightjak-wings`
- `target-handler`
- `target-anim`
- `target`
- `target2`
- `target-swim`
- `target-lightjak`
- `target-invisible`
- `target-death`
- `target-gun`
- `gun-util`
- `board-util`
- `target-board`
- `board-states`
- `mech-h`
- `vol`
- `vent`
- `viewer`
- `gem-pool`
- `collectables`
- `crates`
- `secrets-menu`
Additionally:
- Detection of non-virtual state inheritance
- Added a config file that allows overriding the process stack size set
by `stack-size-set!` calls
- Fix for integer multiplication with `r0`
- Fixed detection for the following macros:
- `static-attack-info`
- `defpart` and `defpartgroup` (probably still needs adjustments, uses
Jak 2 implementation at the moment)
- `sound-play` (Jak 3 seems to always call `sound-play-by-name` with a
`sound-group` of 0, so the macro has been temporarily defaulted to use
that)
One somewhat significant change made here that should be noted is that
the return type of `process::init-from-entity!` was changed to `object`.
I've been thinking about this for a while, since it looks a bit nicer
without the `(none)` at the end and I have recently encountered init
methods that early return `0`.
Previously, `object` and `none` were both top-level types. This made
decompilation rather messy as they have no LCA and resulted in a lot of
variables coming out as type `none` which is very very wrong and
additionally there were plenty of casts to `object`. This changes it so
`none` becomes a child of `object` (it is still represented by
`NullType` which remains unusable in compilation).
This change makes `object` the sole top-level type, and the type that
can represent *any* GOAL object. I believe this matches the original
GOAL built-in type structure. A function that has a return type of
`object` can now return an integer or a `none` at the same time.
However, keep in mind that the return value of `(none)` is still
undefined, just as before. This also makes a cast to `object`
meaningless in 90% of the situations it showed up in (as every single
thing is already an `object`) and the decompiler will no longer emit
them. Casts to `none` are also reduced. Yay!
Additionally, state handlers also don't get the final `(none)` printed
out anymore. The return type of a state handler is completely
meaningless outside the event handler (which is return type `object`
anyway) so there are no limitations on what the last form needs to be. I
did this instead of making them return `object` to trick the decompiler
into not trying to output a variable to be used as a return value
(internally, in the decompiler they still have return type `none`, but
they have `object` elsewhere).
Fixes#1703Fixes#830Fixes#928
* mips 2 c basic version, not yet tested
* calling works without crashing, but the function doesn't
* it works
* add test
* cleanup and actually add the test
* dont use mips2c by default for font
* clean up formatting
* add support for non virtual states
* typecheck go
* start on virtual states
* more support for virtual states
* offline passes
* fix tests
* use behavior shortcut instead of lambda
* final cleanup of virtual go
* unused var warnings and fix inconsistent enum decompile order on win vs linux
* fix thread safety bug with goal symbol table and vif1 interrupt handler
* fix type mistake
* get gkernel and gkernel-h at least somewhat working in the offline tests
* strip comments from json
* switch hints to casts. online tests passing, offline passing up to gkernel
* variable retyping is added
* fix up casts in lets
* update
* up to ash
* add more expressions
* fix some return variable usage nonsense
* bfloat print working
* basic-type working
* type working, fix decompiler on all files
* clang format
* wip decompiler ir
* add AtomicOp stuff
* fix windows build and warnings
* add instruction parser
* include
* make minilzo shared
* odr fix
* a
* fix merge conflicts
* move decompiler into namespace
* update the code coverage to include the decompiler
* add demo test
* add register use test to example test
* begin framework for expressions
* more
* clean up warnings
* small fixes
* update
* wip type prop improvements
* see if nasm works
* fix format strings
- Recognize new type definitions/parents/type flags in the decompiler
- Analyze autogenerated inspect methods and dump guesses at fields to a file
- Utility functions for accessing static data by label
- Better ordering in the decompiler to go through functions in the order they appeared in the source
- Added a decent number of types to `all-types.gc` based on the new field analyzer
- Correct a few `int`/`integer` mistakes in `gcommon.gc` (this should really be a warning)
- Correct a few type issues in `gcommon` and `gkernel-h`
- Option in the decompiler to be strict about `define-extern` redefining a type of a symbol
- Add a test to check consistency in types between `all-types.gc` (used by decompiler) and `goal_src` (used by the compiler)
* implement basic framework for decompiler type analysis
* before type system changes
* add some decompiler stuff to the type system
* try algorithm on a few functions
* w/early-return-and-break
* more edge cases
* all instructions in non asm functions now convert to ir
* cleanup
* implement rarely used control flow in IR
* clang format
* add some more cfg ir stuff
* add cond with else
* add type of recognition
* add cond to compare conversion
* finally all conds are passing
* started sc recognize, but ash min and max should be recognized first
* fix ash showing up as sc
* add abs
* fix merge issues
* try building goos with optimizations on
* sc mostly working, still need to fix right aligned nesting
* ands and ors are converting correctly now
* clean up
* framework for basic op
* started IR framework
* check in type info file
* add some basic operations to the first pass ir conversion
* use a single condition system
* add more basic op decoding
* more ir
Mostly revolved around the new MSVC check for functions that don't cover all paths (ie. a switch statement without a default case). It appears to not see an assert as a valid default case.
I switched assert(false) to exceptions in these cases. I believe this should also abort the program, but will also provide a hopefully useful message? Hopefully this is an improvement.
Resolves#32
Hat Kid
ManDude
water111
Tyler Wilding
water
Tyler Wilding