Re: regression on zOS between 5.4.8 to 5.5.0 on zOS

> Maybe there are other undetected overlapping assignments in lcode.c;
> it uses a lot the structure 'expdesc', and it does several assignments
> among the fields of that structure.

Another suspect is this one:

    t->u.ind.t = cast_byte((t->k == VLOCAL) ? t->u.var.ridx: t->u.info);

In this case, the fields u.ind.t and t->u.info do indeed overlap. The
code does not assign one to the other---it assigns the result of a
conditional operator to the field u.ind.t---but it is easy to imagine
that a compiler can get this wrong. Whether the bug is in the code or
in the compiler seems a question for language lawyers. (Again, we can
avoid this code, just in case.)

-- Roberto

-- 
You received this message because you are subscribed to the Google Groups "lua-l" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].
To view this discussion visit https://groups.google.com/d/msgid/lua-l/20260207182500.GA33568%40arraial.inf.puc-rio.br.