On 11/11/2014 22:55, Robert C. Helling wrote:
All deco and ceiling calculations have to know at some point the ppO2 to
determine the breathing gas. If there is sensor data, great, then use the value
from that (using voting logic) otherwise the set point is the best guess
(assuming the rebreather works as designed). The current code (at least before
your patch) uses the variable sample.setpoint as this best effort value for
ppO2 (admittedly the name might not be optimal). So even, when use user had a
set point, this variable eventually gets overwritten by the sensor data.
So with your patch, that makes a distinction, the sensor data does not reach
the deco/ceiling calculations anymore. It is not sufficient to set the
variables when reading from files, you also have to make sure, that the values
reach other points in code where they are expected.
Also note that code at different places tests for CCR character of that segment
(as we said, CCR is a property of a segment, not the dive as in bail out)
against 0. This (or a replacement) has to work after you separate set point and
sensor data.
There are several ways I see possible to achieve this (all involve Qt Creator
functions):
1) You refactor/rename the old sample.setpoint back to sample.ppo2 and make
sure it gets initialised properly from sensor data (as done in
fill_o2_values()) and lacking that set point information. (simpler I guess)
2) You go through the list of “Find usages” of sample.setpoint and update the
code there accordingly.
This patch as it is breaks stuff.
Best
Robert
-
Robert,
I am quite dumb because I struggle to understand the real detail of what
you are saying. I get the broad message but not the precise implication.
After reading your message a few times here is what I understand against
the context of what I know from the code. It is about the use of
variables plot_data->po2 and plot_data->setpoint. I am insistent that
setpoint should mean the setpoint given to the dive computer and not
anything else. To store a setpoint value in a variable called ppo2 just
destroys the understandability of the code to anyone else.
Let's talk about xml input. there is no sample->po2 variable. So, when
reading a "PO2='1.22' " attribute from xml, there is not a direct
variable against which to store that. So when we encounter "PO2=.." in
xml what should we do about this? If one wishes to keep the PO2 xml
attribute at all, there is only one logical way to treat it, and that is
as a po2 measurement (therefore store it in sample->sensor[0]), do not
treat it as a setpoint. I agree that the best way to check for a CCR
point on the dive profile is by checking for a setpoint. But then the
PO2 attribute is possibly not a good place to get that setpoint
information and the setpoint attribute is a better place for reading
that information.
Now, while the sample structure has no PO2 member, the plot_data
structure does have. We are free to translate the sample information in
any way when we transfer from samples to plot_data. My feeling is that
one should be fairly specific in reading and storing the raw dive data
in xml. If we want to set up a way to preliminarily store setpoint data
into the sensor values (to provide for the case when there are no sensor
data), then the proper way to do this translation is surely while
populating the plot_data structure and not while reading the raw xml
samples?
I suspect that you deal with equipment that stores setpoint values that
are reported in xml as "PO2=..". is this correct? If this is the case
then there is only one option and this is to keep compatability with
whatever xml that equipment produces. But this would be a critical bit
of information that I am not aware of at the moment.
Please spell out the issue a bit more, would you?? Please comment on
each of my paragraphs above. I am insistent in getting totally into your
frame of reference to really understand this.
Kind regards,
willem
_______________________________________________
subsurface mailing list
[email protected]
http://lists.subsurface-divelog.org/cgi-bin/mailman/listinfo/subsurface
