Re: is it possible to make longjmp-free Lua?

["Juris Kalnins" <juris@...>]

On Tue, 28 Jul 2009 06:18:04 +0300, David Manura <dm.lua@math2.org> wrote:

> I was thinking of posting something on this exact performance problem
> with lua_cpcall allocating a new function object on each call.  The
> allocation can be avoided.  In brief....first construct a Lua C
> function that when executed calls the C function pointer stored in the
> lightuserdata passed as the first argument.  This is done only once.
> Now, whenever you need to call a C function, lua_pushvalue that Lua C
> function mentioned previously, lua_pushlightuserdata the C function
> pointer you want to call, and finally do the lua_pcall.  All three
> operations never longjump and are safe, assuming you have sufficient
> stack space.  The approach does assume that C function pointers can
> fit inside a lightuserdata, which is usually safe in practice.  I had
> done some micro-benchmarking of this and got a good speedup.

May I ask how big was your speedup?
I just tried microbenchmark with lua_cpcall and the following:

int my_auxcpcall(lua_State *L) {
    lua_CFunction f = (lua_CFunction) lua_topointer(L, -1);
    lua_pop(L, 1);
    return f(L);
}

int my_cpcall(lua_State *L, lua_CFunction func, void *ud) {
    lua_pushvalue(L, myauxcpcallIdx); /* global reference index that has  
my_auxcpcall CFunction value */
    lua_pushlightuserdata(L, ud);
    lua_pushlightuserdata(L, (void *) func);
    return lua_pcall(L, 2, 0, 0);
}

and got 0.15 vs 0.22 seconds on 1000000 calls on core2, and 40 vs 47  
seconds on mips.
I was expecting something more spectacular...