On Fri, Apr 22, 2016 at 12:10:07PM -0400, Jan Harkes wrote:
> On Fri, Apr 22, 2016 at 11:55:46AM -0400, Jan Harkes wrote:
> > This just needs to be fixed by padding the strings out.
> 
> Actually I'm going to fix this by reshuffling the packed fields
> alignment issues pop up when an N-byte type is not read at an N-byte
> boundary. I think if I move the strings to the end everything else
> should fall into place.

For my taste consequently treating the data as a byte array and
packing/unpacking explicitly is safer in the long run and easier to
modify without bothering about alignment.

But of course as long as the constraints on data member ordering are
documented/commented, just go ahead, aligned access is faster and the
code smaller (for cfs this hardly makes any difference though :).

There are no compatibility problems either as cfs and venus being in
use together are assumed to be built together (which does not sound
positive to me, I would prefer very much to have a venus<->cfs protocol
specification instead, but this is the status quo anyway).

Rune