test binaries for Beta 5

[Jef Driesen]

On 2014-08-05 17:13, Miika Turkia wrote:
> On Tue, Aug 5, 2014 at 5:49 PM, Dirk Hohndel <dirk at hohndel.org> wrote:
>> On Tue, Aug 05, 2014 at 05:33:24AM -0700, Richard DePas wrote:
>>> Unfortunately I am still getting a group of error messages when 
>>> importing
>>> from my Petrel. The dives still show up.
>>> 
>>> unlikely dive gas data from libdivecomputer: o2 = 0 he = 0
>> 
>> Well, arguably these ARE rather unlikely dive gases :-)
>> 
>> Remind me what our conclusion was last time we talked about this...
>> 
> I am not sure whether a final conclusion was ever reached. Jef informed 
> us
> that this should be fixed now on libdivecomputer. He also wanted a 
> memory
> dump of the dive computer, but I was not able to grab one. A dump fails
> every time when I attempt to grab one from my Stinger.

I requested this warning message in order to help find parsing bugs in 
libdivecomputer [1-2]. If subsurface filters out the bogus gas mixes, 
then nobody will notice (and hopefully report) the libdivecomputer bug, 
and it will never get fixed.

The most likely reason for those odd o2 and he percentages (e.g. zero or 
0xFF values) is that the device uses those special values to indicate a 
disabled gas mix. I already fixed some of these issues, but certainly 
not all. So if you still spot such issue, please file a bug report and 
I'll have a look.

And because this is related to how the device stores the data, that 
means that if you are affected by such a bug, you'll get the warning for 
each disabled gas mix on every dive. Hence the large number of repeated 
warnings.

> Anyway, Showing same message only once would make sense, no matter if 
> the
> error message is this or something else in the future.

I think suppressing duplicate warnings is an excellent idea. In that 
case, the warning is still there, but at least users are no longer 
flooded with the same warning.

As I mentioned before [1], I would prefer to keep the warning if 
possible, for the reason explained above. But if they are really getting 
too annoying for users, go ahead and disable them before the release. Or 
maybe log to stdout, where most end-users won't see it?

[1] http://lists.hohndel.org/pipermail/subsurface/2014-June/012642.html
[2] http://lists.hohndel.org/pipermail/subsurface/2014-June/012508.html

Jef