It make debugging much easier if the function signature tells you
that a parameter is not altered.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
When plotting profiles with surface segments, there were strange
artifacts. As we found out with Robert, these were due to the fact
that the calculated maxtime was set to the last event which is just one
second inside the surface segment. This terribly confused the profile
code. For example, it didn't properly allocate samples for the surface
segment.
Thus, when calculating maxtime, consider the last sample beyond the
last event.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
When we had fixed-sized cylinder arrays, the planner used the last
empty cylinder for "surface air". This was not recognized by the UI
as a separate cylinder, because "empty cylinder" was the sentinel for
the end of the table. The conversion to dynamically sized cylinder
tables broke this code: everytime the surface segment is changed,
a new dummy cylinder is added, which is visible in the UI.
As a very temporary stop-gap fix, emulate the old code by creating
a cylinder and then setting the end-of-table to before that cylinder.
This means that we have to loosen the out-of-bound checks.
That's all very scary and should be removed as soon as possible.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
This will be temporarilly used by the planner to mark consumption of
air at the surface. Do this by creating a new function add_cylinder,
which replaces add_to_cylinder_table() and takes care of always adding
a dummy cylinder at the end of the table. Make the original
add_to_cylinder_table() local, so that it cannot be accessed anymore.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
These two Q_ASSERTs made no sense - their expression (a string
literal) always evaluated to true. A qWarning() was intended here.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
When the dive mode is changed, the profile has to be replot. This
is by a function of the TabDiveInformation. However, that function
was also executed when populating the tab. Thus, when changing dive,
the profile was plot twice.
Move the profile plotting out of the function. Ultimately, the profile
should listen to the appropriate signals itself.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
There's no point in doing that since the app directory this creates is broken
on older macOS versions, anyway (and we create a working DMG through a
webhook).
Additionally, lately this has started to fail on GitHub, so let's just rip this
out.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
The profile data was not properly cleared when not showing a
dive.
Fixes#2787
Reported-by: Robert C. Helling <helling@atdotde.de>
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
This file is copied from the QBS project. It's under LGPL and therefore
compatible with our licensing. While it would be possible to retrieve
this file at build time from the original project, for now it seemed
easier to include it in our repo.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
This reverts commit cd47499694.
The signal removed in the reverted commit was used in Subsurface-mobile.
So removing it, caused a not running app.
Signed-off-by: Jan Mulder <jan@jlmulder.nl>
Instead of using the Subsurface-divelog user on GitHub, we now use an org that
was generously donated to us.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
To check wether the tab widgets should show the trip view, they called
the selectedTrips() function. The trip view was shown if that contained
only one trip. However, the selectedTrips() function was very slow,
because it has to query to core models.
Change the function to singleSelectedTrip(), which returns a trip
if there is exactly one trip selected. The function returns early
if there is more than one trip selected. This makes the select-all
case much faster.
There are two cases which are still very slow:
- List mode, because here all top-level items are queried.
- Dive log with many only top-level items.
Ultimately, we will have to cache the trip selection because
querying the model is too slow.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
The old code would call QItemSelectionModel::select() once for every dive.
Instead collect the selection in a QItemSelection and only call
QItemSelectionModel::select() once. This makes selecting multiple dives
significantly faster.
The loop also expanded the trips with selections. This has now to be
done in an extra loop.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
When determining the selected dive sites to highlight them on the
map, the DiveListView code used the local indices of the selected
dives. However, that was unreasonably slow. Even though a layering
violation, let's access the core data structures directly. In my
tests this improved from 700 ms to 0 ms!
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
This was used by the divelist to check wether a selection change is
programmatical or user-initiated. However, since there is only one
entry point for programmatical selection changes, this is not needed
anymore. Remove it - this removes an inter-module dependency.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
It makes reasoning about code so much easier when one knows that
functions cannot be accessed from the outside. Therefore, make
a number of DiveListView slots private. Moreover, unslotize
functions that never were used in connect calls.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
The DiveListView widget has to differentiate between programmatical
und user-initiated selection changes. It did so by using the
DiveListNotifier::inCommand() flag.
However,
1) There is only one point of entry for such selection changes,
viz. the MultiFilterSortModel::selectionChanged() signal
2) This signal is not only emitted in command-context.
Another source is for example dive-map selection changes.
Therefore, move the programmatical-selection-change status down
to the widget and set/reset it in the diveSelectionChanged() slot.
This makes "select all visible dive sites" somewhat faster. Sadly,
not as much as expected.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Outside callers had no business calling these functions.
Especially some functions that were simply there to override
default DiveListView behavior. Mark these as overridden.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
This was used by the map to select dives. However, the map now calls
the core function directly, so this can be removed.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
The map widget called the dive list to select dives. This is
inconsistent and complex. The dive list has to call down to
the core anyway. Therefore, change the code to call the common
core function.
This means that we have to transform integer ids into dive-pointers.
That is a bit sad, because the dives were just transformed into
indices. Let's address that in a future commit.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
This is no longer created using GitHub actions (but all the necessary
information is still included in this repo). We need to be able to
shrink this container so our GitHub Action runs don't run out of disk
space.
Adjust the path where the resulting binaries are found with this build.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
We no longer use libusb to access USB devices on Android, therefore
there's no point including libusb in our build. Also, we have never even
attempted to run the tests on Android, so let's not even pretend to
support building them.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
If we install our support libraries into the NDK we later run into
include path order issues that result in strange errors around the
inclusion of math.h (because we find the C version of that include
file that ships with the NDK before we find the libstdc++ version
of math.h (because the include path for our support libraries is
listed before the libstdc++ include search path). By having a distinct
install-root for our libraries we can avoid this problem.
Remove the previous hack that tried to work around the symptoms of
this issue.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
The official installer now requires the user to log in which we can't
really do in a scripted manner. Let's see how long this way of
installing things will be available.
While doing this remove an ancient hack of some Qt settings that we no
longer need.
This also tries to prune some things that we don't need in the Docker
image to reduce image size.
The mapbox plugin is removed as it would add a dependency to QtSql which
we otherwise don't need. And since the plugin isn't used, no point in
installing it.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Qt 5.13.2 fixes a few bugs.
We are NOT switching to Qt 5.14 as that breaks a ton of things with
Android support. At this point even with qmake it near impossible to get
working Android binaries, no one appears to have a solution for cmake.
With qt-android-cmake current master fixes the problem that held us back
at the earlier commit, so let's go back to using master.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
It appears that on those two platforms you now need a paid API key in order
to have this feature work. Certainly not something I'm going to do.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
The qml-widget seems to catch selectedDivesChanged signals of the
MapWidgetHelper and reemit them. However, there seems to be nobody
listening to this signal?
Let's remove this for now to make debugging of the signals easier.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
This was erroneous, as it should only be called at the end of
a selection change, not after every single dive. It made selection
of multiple dives extremely slow.
Reported-by: Jan Mulder <jlmulder@xs4all.nl>
Debugged-by: Robert C. Helling <helling@atdotde.de>
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
DiveListView::selectDive() is an overloaded function. The second
version was only called by the first version, so we can fold one
into the other.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
The undo-machinery makes sure that a current_dive always exists
after an undo command. This part of the code should never be
called.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
This appears to be an artifact. Collapsing does not unselect
the current dive, so reselecting it appears pointless.
Also it is unclear why the selection should be restricted to
a single dive after collapsing.
Probably that was originally meant to expand only the trip
with the current dive in it?
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Thus, it is made clear that these member functions can only be
accessed from DiveListView itself. This should make debugging
easier.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Move this to the core so that desktop and mobile don't have
to call this explicitly. Matter of fact, mobile didn't call
this. It is unclear, whether that was even used on mobile,
though.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
It makes no sense to keep the device nodes if all the other data
is cleared. Let's do this automatically and not explicitly.
This ensures that the function is also called on mobile.
Currently it was only called on desktop.
Weirdly, the parser-tests were expecting that the device nodes
were not reset by clear_dive_file_data() and therefore divecomputers
were accumulating in the test results. Thus, the additional
computers had to be removed from the expected test results.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
The filter sets the maximum date to now. This is so confusing when
you manually add a dive and it isn't shown, because it is slightly
in the future. Add seven days, that should help.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
This only checks the first divecomputer as the semantics for
multiple dive computers with different dive modes are not
clear. Should we check them all?
The implementation is a bit hackish: the indexes [0...n] of the
combobox are mapped onto [-1...n-1], where -1 means don't filter
and n-1 is the last valid dive mode.
Implements #2329
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
It appears that some misguided compiler / library combinations crash
on &vector[0] for empty vectors. Even though very unfriendly, they are
technically correct, so let's remove these constructs.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
TabDiveInformation::updateProfile() does some statistics via the
per_cylinder_mean_depth function. It passes down arrays with one
entry per cylinder, which are allocated by means std::vector.
To pass the array, the expression "&vector[0]" is used. It seems
like some compilers through an assertion violation if vector
has no elements. They are technically correct in that this is
undefined, but still this appears like very unfriendly behavior.
After all, std::vector should behave just like a dynamic C-array
that is automatically freed, when going out of scope.
Replace the "&vector[0]" by "vector.data()" and don't do the
call if there aren't any cylinders for good measure.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
