ghost.zh
Overview
Creator:
Emily
Updated: 08 Jan 2020
Tags:
Global,
Library
Downloads: 433
|
Download Package (2.11 MB) |
Information
An enemy scripting library.
Description Setup Reviews Comments
Timelord
Posted 16 July 2019 - 05:16 AM
I see that you changed default to defaultVal. I'd also suggest using ptr in all places where you use array as an identifier. In future versions, array may become a token, for complex arrays; as will struct. We have already fixed std.zh (repo for std.zh, here), renaming all instances of array and string to ptr, for this very reason.
(A real array datatype will also then use proper C-style pointer tokens, so you'd have &ptr or *ptr.)
This does look as if it will fix the issue for now. I forced a size check in my patch to avoid issues where the value stored as (enemy->Misc[__GHI_NPC_DATA]&0xFFFF) is non-zero, but is not actually an array.
While this may be an edge case, it's theoretically possible. Validating the size ensures that the function moves to its else block and ensures that there is never access to an invalid pointer: The array size here is always 24, hence why I tested for equality in my patch.
All that aside, how would you feel about making a repo for ghost?
I've considered refactoring it for 2.55, to produce ideal ASM, for performance reasons; and future versions could migrate from ffcs to npcs. An open repo for it would allow community enhancement, while remaining under your supervision.
(A real array datatype will also then use proper C-style pointer tokens, so you'd have &ptr or *ptr.)
This does look as if it will fix the issue for now. I forced a size check in my patch to avoid issues where the value stored as (enemy->Misc[__GHI_NPC_DATA]&0xFFFF) is non-zero, but is not actually an array.
While this may be an edge case, it's theoretically possible. Validating the size ensures that the function moves to its else block and ensures that there is never access to an invalid pointer: The array size here is always 24, hence why I tested for equality in my patch.
All that aside, how would you feel about making a repo for ghost?
I've considered refactoring it for 2.55, to produce ideal ASM, for performance reasons; and future versions could migrate from ffcs to npcs. An open repo for it would allow community enhancement, while remaining under your supervision.
