Re: Bug in hook(luaV_finishOp:OP_LE)?

[云风 <cloudwu@...>]

2015-04-09 23:34 GMT+08:00 Roberto Ierusalimschy <roberto@inf.puc-rio.br>:

>
> --- lvm.c       2015/03/30 15:45:01     2.238
> +++ lvm.c       2015/04/09 15:30:13
> @@ -275,9 +275,14 @@
>      return l_strcmp(tsvalue(l), tsvalue(r)) <= 0;
>    else if ((res = luaT_callorderTM(L, l, r, TM_LE)) >= 0)  /* first try 'le' */
>      return res;
> -  else if ((res = luaT_callorderTM(L, r, l, TM_LT)) < 0)  /* else try 'lt' */
> -    luaG_ordererror(L, l, r);
> -  return !res;
> +  else {  /* try 'lt': */
> +    L->ci->callstatus |= CIST_LEQ;  /* mark it is doing 'lt' for 'le' */
> +    res = luaT_callorderTM(L, r, l, TM_LT);
> +    L->ci->callstatus ^= CIST_LEQ;  /* clear mark */
> +    if (res < 0)
> +      luaG_ordererror(L, l, r);
> +    return !res;  /* result is negated */
> +  }
>  }

I saw this patch in bugs page (http://www.lua.org/bugs.html) now.

But, Is this the patch base on the Lua 5.3.0 ? I can't patch it to the
official release.

The line 275 of lvm.c (http://www.lua.org/source/5.3/lvm.c.html) is
the function luaV_lessthan, not luaV_lessequal .

-- 
http://blog.codingnow.com