Instead of assigning sensor ids in the order that values are reported,
actually use the sensor ids reported by libdivecomputer. This will fix
the problem that for some dive computers (e.g. Shearwater) the dive
computer calculated ppO2 is currently reported first, thus pushing out
all actual sensor values.
A new fixed id (7) outside of the range of currently supported sensor
IDs is used for these dive computer calculated values.
Signed-off-by: Michael Keller <github@ike.ch>
We appear to consistently assume that clamp can be called with 0, size as
interval, but that actually results in a possible out of bounds access at the
upper end.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Mostly NFC; this commit is mainly to get familiar with the codebase and to meet
the people who will review these changes.
I hope to make some changes to the DAN parsing code to eventually extract more
metainfo from my aqualung divecomputer's `.zxu` formatted logs. To do so, and
for me to be able to work on this efficiently, I've refactored the DAN parsing
code using a bit more modern C++-style, as well as being more true-to-spec wrt.
the (...ancient) DAN file format documentation that i could dig up... hopefully
that's an alright tradeoff for the project.
This more true-to-spec parsing also fixed a bug with the number being parsed
from the incorrect index in the ZDH vector (or, atleast i consider it a bug -
the "Export sequence" number was being used as the dive number, instead of the
"Internal Dive Sequence" number. The latter, described in the spec as: `The
sequence number assigned to the dive by the recording computer`).
Also contains some unrelated formatting changes; i tried to keep these minimal
(i presume these files haven't been touched in a while by `clang-format`).
Signed-off-by: Morten Borup Petersen <morten_bp@live.dk>
Fix the script used to generate version numbers for CICD builds. If
there is a race condition and we are not the first to create the branch
named after the lates changeset id, go and do a `git pull` before
attempting to read the branch that was created by another process.
Signed-off-by: Michael Keller <github@ike.ch>
Move the import of .FIT files into the 'Import log files' menu item,
where most people will be looking for it. This also naturally opens a
file selection dialog, which is more intuitive than having to select
this in the dive computer import dialog.
Also fix a bug affecting file imports if the log files contain
coordinates - the dive log needs to be set in the import data structure.
And refactor the file dialog file filters to make it more natural to add
more entries.
Requires https://github.com/subsurface/libdc/pull/72 to work.
Signed-off-by: Michael Keller <github@ike.ch>
This was using delete for a buffer allocated with calloc().
BTW this definition is more simple than previous.
Signed-off-by: Salvador Cuñat <salvador.cunat@gmail.com>
Import can be done from .script files generated by LogTrak software or
from .asd files generated by SmartTrak and LogTrak export option.
This code was writen in C, and has just been "ported" to some extent to
C++, so it can work with recent rework of Subsurface to C++.
I'm not a C++ guy, so this is mostly C with some make-up.
Signed-off-by: Salvador Cuñat <salvador.cunat@gmail.com>
Add a check that will fail whenever there is a change in the list of
supported dive computers in libdivecomputer that is not reflected in the
`SupportedDivecomputers.*` lists.
Also add a script for a simplified update of these lists.
From a discussion with @dirkhh in https://github.com/subsurface/libdc/pull/71#issuecomment-2565384338.
Signed-off-by: Michael Keller <github@ike.ch>
For dives in CCR mode, show 'bailout' and 'on loop' events whenever a
gas switch from a diluent gas to a bailout gas and vice versa happens.
Signed-off-by: Michael Keller <github@ike.ch>
Clarify that the gradient factor displayed in green in the profile is
the gradient factor that is set in Subsurface, and not the one used by
the dive computer.
Fixes#4396.
Signed-off-by: Michael Keller <mikeller@042.ch>
In the decostop table, we had deco stops at fractional feet
(e.g. 333.33 ft). This was somewhat awkward, because the
user defined literals do not yet(?) support non-integers
and therefore, these entries were given in mm instead.
Just round to the nearest integral feet values.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Commit is longer than expected, because a few of the callers
were converted to use depth_t instead of int.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Also un-inline it. There seems no reason for exporting the
implementation details in the header file.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
This function had a horrendous interface: The caller would have to
allocate two arrays of the correct size to be filled with data.
The callee couldn't even check the size, because the data was passed
as raw pointers.
Instead, use std::vector<>, construct everything in the called
function and do size-sanity check in the calling function.
Use depth_t and duration_t instead of plain integers to represent
mean depth and time.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Allow multiplication/division of unit types with scalars,
such as
depth * 4
or
depth / 4
Multiplication the other way round (4 * depth) is not currently
supported, because that needs some template trickery.
Might do this later or wait for our switch to C++20, where we
could use concepts to make that trickery more palatable.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
No point in this being a macro. Make it return a depth_t - it
was unclear that this returns a depth in mm.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Calls the global interpolate() function for integer types.
For now, the template is enabled if the arguments are not
integer types. Might want to refine that in the future.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Make the type the interpolate() function works on a template type,
so that this can be used for arbitrary unit types without warnings.
Internally, it does its math using floating point arithmetics anyway,
so no risk of overflow. (Exception: when just taking the middle point,
but I would hope that no part of the code lives on the edge such that
addition of two values gives an overflow.)
Moreover, add an "math.h" include so that the header becomes
independent of others.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
