a couple of crashes with the test dives

[Lubomir I. Ivanov]

On 24 November 2014 at 15:28, Robert Helling <helling at atdotde.de> wrote:
>
> On 23.11.2014, at 21:54, Lubomir I. Ivanov <neolit123 at gmail.com> wrote:
>
> Hi,
>
> i was testing something by importing all test dives and i think i
> found some bugs.
> not sure if these are known or related to WIP code - both patches mention
> CCR.
>
>
>
>
> We were (I was I should say) too clever when comparing floating point
> numbers resulting in 0 not being considered equal to 0. This occurred when
> computing the scaling factor for a y-axis of an O2 plot without data which
> had a max and a min value of 0 (which happens for those test dives).
>
> Problem is that it does not crash reliably on Mac (i.e. not on Mavericks but
> on Yosemitee), but I tried to test this on Linux as well.
>
> Could you please check that this solves the problem?

hello,

with the patch it still crashes with the second backtrace when
selecting #36 -> #1 (from the same, first trip).
the problem in DiveGasPressureItem::paint(), again looks like the
number of entries is not correct (probably less than the number of
elements to plot).

>
> On a related matter: dives/test35.xml is declared to be a CCR dive but it
> lacks both o2 sensor data or set-point. This results in an OC calculated
> ceiling which is heavily violated. I suggest adding po2=‘1.2 bar’ to one of
> the samples.
>

with or without the patch selecting #36 -> #35 doesn't crash - i think
while bisecting i found a bug that was already fixed between fa181d1de
and HEAD.
or...it could be conditional.

lubomir
--