On Wed, Aug 06, 2014 at 09:27:14AM -0700, Linus Torvalds wrote:
> On Wed, Aug 6, 2014 at 9:12 AM, Linus Torvalds
> <torva...@linux-foundation.org> wrote:
> >
> > So try this. I have *not* tested it with the odd cases that I tested
> > things originally, but the code hasn't really changed and it should
> > all be ok. And I updated the commit message, so it should be better.
>
> So to show some example breakage, here's a stupid dive based on
> dives/test1.xml that shows an example.
>
> It has three gases, but the second has doesn't have any gas switch events.
>
> What *used* to happen is:
> - the "get_gaslist()" thing shows gases 1 and 3
> - the gas *usage* shows just gas 1, and incorrectly at that ("0.00
> cuft", when it's actually unknown and should be empty)
> - the SAC-rate shows just gas 1, which is unknown (and properly empty)
>
> What happens with my patch is:
> - it shows gases 1 and 3 sanely
>
> What we *might* want to add is to show gas2 because it does have
> start/end pressure information. Right now it's not shown, because it
> doesn't have any actual use time (no gas switch event to it). That
> would be easy to change in my patch, because it's one coherent loop
> that does all these things.
>
> But this is a pretty made-up example, just to show the breakage that
> the old code had. So really not a big deal.
I don't know what to do with a dive that shows pressure change on a
cylinder that we didn't switch to. And given how few people dive with more
than one pressure sensor and how likely this is a sign that the user
either mistyped the pressure values OR forgot a gas change...
I'm much happier ignoring gases that weren't used on the dive.
/D
_______________________________________________
subsurface mailing list
subsurface@hohndel.org
http://lists.hohndel.org/cgi-bin/mailman/listinfo/subsurface
