Bug in util.strcasecmp() ?

[jfcap]

IMHO util.strcasecmp function is bugged

 util.strcasecmp('aaa','aba')==0 ??? 

-- Fast string case agnostic comparison
function util.strcasecmp(str1, str2)
    local r = 0;
    local b1,b2
    local i
    local len = ((#str1 > #str2) and #str1) or #str2 -- get the longer length

    for i = 1,len do

        b1,b2 = string.byte(str1,i), string.byte(str2,i)
        if b1 == nil then return -1 end
        if b2 == nil then return 1 end
        -- convert b1 and b2 to lower case
        b1 = ((b1 > 0x40) and (b1 < 0x5b) and bit.bor(b1,0x60)) or b1
        b2 = ((b2 > 0x40) and (b2 < 0x5b) and bit.bor(b2,0x60)) or b2
        r = b1 - b2
        -- add here something like : if r~=0 then return r end
    end

    return r
end

[kernelsauce]

This is something that was added by https://github.com/jsolman (Jeff).

I did not test this thing out as most of his code seems fine. I propose we just use the libc version which is already defined in cdef.lua, which probably also are faster:

 int strcasecmp(const char *s1, const char *s2);

so basically:

  function util.strcasecmp(str1, str2)
      return tonumber(ffi.C.strcasecmp(str1, str2))
  end

  --or for added safety:
  function util.strcasecmp(str1, str2)
      assert(type(str1)=="string" and type(str2)=="string", "Invalid arguments.")
      return tonumber(ffi.C.strcasecmp(str1, str2))
  end

Any reasons not to do this jsolman?

[kernelsauce]

Fixed now.

[jsolman]

Sorry for introducing that bug, it did need to break the loop if r ~=0. I must not have tested it properly myself.

[jsolman]

I don't know if it is actually faster or not, might depend on your libc, but since it is available might as well have used it as you did.