[RFC] wrc: remaining TODOs

Alexandre Julliard julliard at winehq.com
Wed Mar 26 12:46:31 CST 2003


"Dimitrie O. Paun" <dimi at intelliware.ca> writes:

> Yes, as I noted, I realized that the option would be needed if
> we use native endianess. But I would strongly argue that this
> is Not Right (TM). FS design has tought us that data should be
> stored in a single endianess. It makes sense -- you could move
> the file from one machine to the other, etc.
> 
> I think the right thing to do is:
>   -- modify Wine to always read stuff in little endian
>   -- remove the option, and force little endian output

I don't think it's feasible, and it's not clear at all that it is the
right way. Resource data is visible from the application, and most
apps will handle it as native endian. If you want to change that, you
will have to add byte swapping code all over Wine plus all over the
applications, for the very small benefit of exchanging .res files
across platforms. I don't see any need for that; .res can be
considered more like an object file format, and you don't expect to
share .o files across platforms either, you simply recompile for your
target.

> P.S. Didn't NT 3.1 use to support MIPS? Isn't that big endian?

No, NT was always little endian; on MIPS and PowerPC you can have both
modes. On Sparc you are stuck with big endian, and by a strange
coincidence NT has never been ported to Sparc...

-- 
Alexandre Julliard
julliard at winehq.com



More information about the wine-devel mailing list