Re: Locking bug in lua_error()?

[Colin <share-lua@...>]

On Sun, Oct 21, 2007 at 04:55:15PM +0200, Colin wrote:
> Hi,
> 
> digging around in the Lua 5.1.2 source code I noticed that
> lua_error() contains a call to lua_unlock() _after_ the
> call to luaG_errormsg(), although luaG_errormsg() will never
> return (and in fact the comment at the end of lua_error()
> about the "return 0;" only to suppress warnings [because
> luaG_errormsg(), and therefore lua_error(), never return]),
> and the lua_unlock() never called...?

Ok, I think I've found the matching unlock in luaD_throw(),
so the only thing that is amiss is that the lua_unlock() in
lua_error() is superfluous.

Sorry for the noise.

Colin


PS: Gee, how much easier is this with C++ where an automatic
    mutex_lock object that locks in the constructor, and 
    unlocks in the destructor, automatically (sic) cleans
    up in case of all errors (= exceptions)...