Add bitwise AND, OR and XOR modes to math block #91
Conversation
|
Not sure whether this might be relevant, but maybe you might want to have a look into the bit operations library of LuaJIT https://bitop.luajit.org/ and see whether this could be applicable to your implementation. |
I can't seem to find the source code for that library (and even if I were able to do so, it's unlikely the C++ implementation would be applicable to my Lua implementation). What's interesting is the page describing the semantics of BitOp. They've decided to only support arithmetic in the range of signed int32 ( Setting a maximum range would allow to implement the bitwise not operation and bitwise operations on negative integers, but would make the implementation trickier and would require deciding on the behaviour for numbers above 2^52. |
|
If I'm not mistaken the bitwise operation library is exposed to the environment within the table I'm not exactly sure if the SM devs ended up using a different bitwise operations library or the one shipped with LuaJIT. |
Hey, thanks for this suggestion. The latest commits change the implementation to use the |
Scripts/libs/math.lua
Outdated
| -- parses user-provided number strings | ||
| -- - removes all spaces | ||
| -- - parses binary numbers with the "0b", "0B" prefixes or "b" suffix | ||
| -- - parses hexadecimal numbers with the "0x", "0X" prefixes | ||
| -- returns nil for incorrectly formatted strings | ||
| -- | ||
| -- Examples: | ||
| -- math.parsestring(" 1 000 000 ") == 1000000 | ||
| -- math.parsestring("0xFF") == 255 | ||
| -- math.parsestring("0b1000") == 8 | ||
| -- math.parsestring("1000b") == 8 | ||
| -- math.parsestring("test") == nil | ||
| function math.parsenumber(input) | ||
| normalized = string.gsub(input, "%s+", "") -- remove spaces: | ||
|
|
||
| prefix = string.sub(normalized, 1, 2) | ||
| suffix = string.sub(normalized, -1, -1) | ||
| if suffix == "b" or suffix == "B" then | ||
| return tonumber(string.sub(normalized, 1, -2), 2) | ||
| elseif prefix == "0b" or prefix == "0B" then | ||
| return tonumber(string.sub(normalized, 3, -1), 2) | ||
| elseif prefix == "0x" or prefix == "0X" then | ||
| return tonumber(string.sub(normalized, 3, -1), 16) | ||
| else | ||
| return tonumber(normalized, 10) | ||
| end | ||
| end | ||
|
|
There was a problem hiding this comment.
I think it's best to move this function outside of the math namespace, as it's shared with all mod/vanilla scripts. I know we've done it with more functions, but this could interfere with other mods that define a function with the same name.
There was a problem hiding this comment.
Right, I've moved math.parsenumber to a new lib input (maybe there is a clearly better place to put it?) as mp_parseNumber
This PR implements the bitwise AND, bitwise OR and bitwise XOR math block modes from #90.
I've decided not to implement the bitwise NOT operation, as it would require establishing the number of bits in the integer (if Lua used 32-bit or 64-bit integers it would be simple, but it uses doubles instead). The bitwise not operation is equivalent to a bitwise XOR with a number consisting of only binary ones.
The current behaviour for numbers that are not non-negative integers is:
negative numbers are treated as 00.1is treated as0,0.5and0.9are treated as1)Maybe the behaviour should be different? Should a
math.flooroperation be used instead ofmath.round?