[lvm-devel] number conversion in lvm

[Mikulas Patocka]

> > > > BTW. how is that function supposed to handle integer overflow - it
> > > > doesn't
> > > > seem to do that.
> > > > 
> > > 
> > > it returns  LONG_MIN/MAX for underflow/overflow - but assuming errno could
> > > be checked for ERANGE and fail command for this case instead.
> > > 
> > > Zdenek
> > 
> > The major problem is that the same function _get_int_arg parses both
> > 32-bit and 64-bit arguments. It would be natural to return error on
> > integer overflow but you can't because you don't know if the caller will
> > use the 32-bit value or the 64-bit value.
> > 
> > Another question is why there is av->i_value and av->ui_value when the
> > values are the same and not dependent on the sign? What if the value fits
> 
> There are separate function for checking sign and accepting signed ints
> (int_arg_with_sign) - otherwise if SIGN_MINUS is detect - log error
> is reported.
> 
> Zdenek

Why are there two values av->i_value and av->ui_value if they have the 
same value and the sign flag is stored separately in av->sign?

Mikulas