cochran_emc.h: packed structs

[Dirk Hohndel]

On Tue, Oct 28, 2014 at 11:29:48AM -0700, Thiago Macieira wrote:
> On Tuesday 28 October 2014 09:34:33 Dirk Hohndel wrote:
> > On Tue, Oct 28, 2014 at 09:28:31AM -0700, Thiago Macieira wrote:
> > > Are we sure of the sample size? It's an odd number of bytes...
> > 
> > We have dive computers with non-constant sample sizes. And pretty much any
> > random number you can think of. Yes, 8 is common, as are 12 or 16.
> 
> 8, 12, 16, 24, 32, 64 are fine.

I meant it when I said "any random number". Yes, we USUALLY see multiples
of 4, but there are others. Like 37 (0x25).

> uemis_sample_t is 37 (0x25) bytes in size with the packing and is used like 
> this:
> 
>         /* first byte of divelog data is at offset 0x123 */
>         i = 0x123;
>         u_sample = (uemis_sample_t *)(data + i);
>         while ((i < datalen) && (u_sample->dive_time)) {
> [...]
>                 i += 0x25;
>                 u_sample++;
>         }
> 
> Given the i += 0x25, it seems that the size is as intended.

It certainly is. If there is anything insane, broken or stupid that could
be done in a data protocol, chances are good that it has been done in the
Uemis protocol. Including an offset of 0x123 for the divelog data. Of
course, 3 x 97. And then 37 byte samples on top of that. What could
possibly go wrong?

/D