[PATCH] Fill up plot_data with temperature all the way

[Jef Driesen]

On 2013-11-13 03:05, Dirk Hohndel wrote:
> My problem with this patch is that it does the wrong thing for
> situations where you have only a very small number of temperature
> readings (e.g.; ONE). I know that I have seen XML files that have a
> single reading about half way through the dive (at the deepest spot,
> IIRC). So for dive computers like this it does something rather odd -
> you get no temperature for most of the dive, and then you get a
> potentially wrong temperature for the rest of the dive.

I don't think I have ever encountered a dive computer, that has only one 
(or very few "discrete") temperature sample(s). There are either no 
temperature values at all, a value every X samples, or a new value 
whenever the temperature has changed. Thus it should be fine to fill in 
missing values by keeping the previous values, interpolating, or 
whatever estimation method you prefer.

There is however one exception. For some Suunto's, which don't have any 
temperature values in the samples, libdivecomputer will artificially 
inject the temperature data from the header (at begin/end of the dive 
and at maximum depth) in the sample data. I assume these are the 
temperature samples you are referring to?

This was implemented long time ago as a stop-gap solution for the lack 
of support for the temperature values in the header. This is a very 
often requested feature. At that time, it seemed a good idea to add 
support it like this. But nowadays, I think we should somehow do this 
differently, because it makes it difficult to tell apart a real 
temperature profile from a few discrete values.

Jef