Re: and/or missbehaviour in embedded lua 5.1

[Andreas Stenius <kaos@...>]

Interesting, I implemented luac into my monitor to see the output, and 
get this:

> c"local a = true and 0"

main <(string):0,0> (2 instructions, 8 bytes at 0x148f7a)
0+ params, 2 slots, 0 upvalues, 1 local, 1 constant, 0 functions
       1       [1]     LOADK           0 -1    ; 2.1436356971487e-228
       2       [1]     RETURN          0 1
constants (1) for 0x148f7a:
       1       2.1436356971487e-228
locals (1) for 0x148f7a:
       0       a       2       2
upvalues (0) for 0x148f7a:
> c"local a = true or 0"

main <(string):0,0> (6 instructions, 24 bytes at 0x14563a)
0+ params, 2 slots, 0 upvalues, 1 local, 1 constant, 0 functions
       1       [1]     JMP             3       ; to 5
       2       [1]     LOADK           0 -1    ; 2.1436356971487e-228
       3       [1]     JMP             2       ; to 6
       4       [1]     LOADBOOL        0 0 1
       5       [1]     LOADBOOL        0 1 0
       6       [1]     RETURN          0 1
constants (1) for 0x14563a:
       1       2.1436356971487e-228
locals (1) for 0x14563a:
       0       a       6       6
upvalues (0) for 0x14563a:
>

Any lua internals knowledgable have any conclusions to make from this?

Btw, I've completely forgot to mention some info regarding my target env:

The sizes for the various data types is as:
       ["char"] = 1,
       ["void *"] = 4,
       ["int"] = 2,
       ["short"] = 2,
       ["double"] = 8,
       ["float"] = 4,
       ["long"] = 4,

Thanks any input you may have!

//Andreas

Kein-Hong Man skrev:
> Your sample:
>
>   local a = true and 0
>
> on Lua 5.1/cygwin gives 0. It compiles to:
>
> [1] loadk      0   0        ; 0
> [2] return     0   1
>
> But on your platform, 1.0890312344636e-60 is however, E0FFFFFF05007C33 
> hex, and the first 4 bytes looks *very* suspicious to me.
>
> But:
>
>   local a = true or 0
>
> is not optimized to a LOADK. It compiles to:
>
> [1] jmp        3            ; to [5]
> [2] loadk      0   0        ; 0
> [3] jmp        2            ; to [6]
> [4] loadbool   0   0   1    ; false, to [6]
> [5] loadbool   0   1   0    ; true
> [6] return     0   1
>
> Looks like some kind of optimization thingy barfed on you. :-)