Re: Lua bytecodes and endian-ness

["D Burgess" <dburgess@...>]

I am of the same view as Asko.
Is it also possible to make the bytecode either independent
or cross-compilable for sizeof(lua_Integer)?

DB
On 6/11/06, Asko Kauppi <askok@dnainternet.net> wrote:
> Could you reconsider the endianess issue?  That is, to have the
> modified lundump.c in the main sources, it could be #ifdef'fed like
> readline support is, that would be fine (= we need to specifically
> enable multi-endianness support).
>
> This comes up at least in fink (OS X) packaging; now, we'll have to
> include & deliver a separate patch. Otherwise, we could be building
> just based on the official .tar.gz.
>
> Thanks for the consideration, :)
>
> -asko
>
>
> Luiz Henrique de Figueiredo kirjoitti 11.6.2006 kello 2.23:
>
> >> Roger.  So the 5.1.x luac will always generate (say) little-endian
> >> bytecode and any other platforms will need to cross-compile that
> >> bytecode (or the client platform requires the modified C code).?
> >
> > luac will always generate native-endian bytecode. There is no cross-
> > compiler
> > (yet). If you want clients to load bytecode of any endiannes, just
> > build Lua
> > with the modified lundump.c that I posted. I'm sorry if I wasn't
> > clear about
> > that before.
> > --lhf