This is a messy commit, because the "qPref" system relies
heavily on QString, which means lots of conversions between
the two worlds. Ultimately, I plan to base the preferences
system on std::string and only convert to QString when
pushing through Qt's property system or when writing into
Qt's settings.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
This is a humongous commit, because it touches all parts of the
code. It removes the last user of our horrible TABLE macros, which
simulate std::vector<> in a very clumsy way.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
There were only two of them, from the time C-code had to access
the divelog: clear_divelog() and delete_single_dive().
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
The old code was leaking memory. Use std::unique_ptr<> for
ownership management.
This is still very primitive and divetags are kept during
application lifetime. There should probably be some form
of reference counting. And the taglist should not be global,
but attached to the divelog.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Let's use std::string in the core. Notably, I'd like to make
the numerous main() functions mostly independent of Qt. Some
things will have to remain, such as argument parsing, of course.
This changes the API: instead of returning an error code and
taking a pointer to the actual return-value, return an
std::optional<std::string>> that is set if the function succeeds.
Returning an empty string in the error case might be simpler,
but oh well...
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
The memory management and string concatenation was hard to follow.
Since the mobile files ios.cpp and android.cpp were already
converted to C++, let's do the same for Unix, Windows and MacOS.
Simply store the default directory and filename in a function-level
static string. Thus, it will be initialized on first call and
freed on application exit. Since the std::string data is
guaranteed to be contiguous and zero-terminated, it can be used
from C code.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
The parser API was very annoying, as a number of tables
to-be-filled were passed in as pointers. The goal of this
commit is to collect all these tables in a single struct.
This should make it (more or less) clear what is actually
written into the divelog files.
Moreover, it should now be rather easy to search for
instances, where the global logfile is accessed (and it
turns out that there are many!).
The divelog struct does not contain the tables as substructs,
but only collects pointers. The idea is that the "divelog.h"
file can be included without all the other files describing
the numerous tables.
To make it easier to use from C++ parts of the code, the
struct implements a constructor and a destructor. Sadly,
we can't use smart pointers, since the pointers are accessed
from C code. Therfore the constructor and destructor are
quite complex.
The whole commit is large, but was mostly an automatic
conversion.
One oddity of note: the divelog structure also contains
the "autogroup" flag, since that is saved in the divelog.
This actually fixes a bug: Before, when importing dives
from a different log, the autogroup flag was overwritten.
This was probably not intended and does not happen anymore.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
A member function had a minute name change.
The SceneGraphBackend is now set via a string argument, not a magic constant.
Thankfully that appears to be backwards compatible.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
The backend infrastructure will soon be able to support more than one
cloud server which automagically stay in sync with each other.
One critical requirement for that to work is that once a session was
started with one of the servers, the complete session happens with that
server - we must not switch from server to server while doing a git
transaction. To make sure that's the case, we aren't trying to use DNS
tricks to make this load balancing scheme work, but instead try to
determine at program start which server is the best one to use.
Right now this is super simplistic. Two servers, one in the US, one in
Europe. By default we use the European server (most of our users appear
to be in Europe), but if we can figure out that the client is actually
in the Americas, use the US server. We might improve that heuristic over
time, but as a first attempt it seems not entirely bogus.
The way this is implemented is a simple combination of two free
webservices that together appear to give us a very reliable estimate
which continent the user is located on.
api.ipify.org gives us our external IP address
ip-api.com gives us the continent that IP address is on
If any of this fails or takes too long to respond, we simply ignore it
since either server will work. One oddity is that if we decide to change
servers we only change the settings that are stored on disk, not the
runtime preferences. This goes back to the comment above that we have to
avoid changing servers in mid sync.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
The profile had a static variable which prevented animation
when first showing the profile. It appears more logical to
don't show the animation when switching from the empty state.
This removes global state, as a function static variable
exists only once, even if there are multiple objects.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
This is obviously a pure code-hygiene thing. But with the new
dynamic tank info table, this becomes trivial, so let's do it.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
The list of known tank types were kept in a fixed size table.
Instead, use a dynamic table with our horrendous table macros.
This is more flexible and sensible.
While doing this, clean up the TankInfoModel, which was leaking
memory.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
So far, the PreferencesDialog emitted a settingsChanged signal.
This meant that models that listened to that signal had to
conditionally compile out the code for mobile or the connection
had to be made in MainWindow.
Instead, introduce a global signal that does this and move
the connects to the listeners to remove inter-dependencies.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
The parse_* functions should probably be declared in parse.h.
Arguably, parse_xml_init() and parse_xml_exit() should be moved
to an init.h file, however that doesn't yet exist.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Instead of deleting the Application explicitly in exit_ui(),
use a unique_ptr to have the instance automatically freed.
This is the pattern given in the Qt-docs:
https://doc.qt.io/qt-5/qapplication.html
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Move the declarations of the "report_error()" and "set_error_cb()"
functions and the "verbose" variable to errorhelper.h.
Thus, error-reporting translation units don't have to import the
big dive.h header file.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Make dive.h a bit slimmer. It's only a drop in the bucket - but at
least when modifying tag functions not the *whole* application is
rebuilt anymore.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
A number of small cleanups to the color-table:
1) Make the profile_color map of static linkage - it is not
used outside of this file.
2) Remove the third color, which originally was planned for
printing. It was not accessed anywhere.
3) Replace QVector<QColor> by std::array<QColor, 2>. Using a
reference-counted, copy-on-write, dynamic container for static
data seems like overkill. std::array<QColor, 2> has exactly the
same run-time impact as QColor[2], but allows for assignment.
4) Use brace-initialization and remove the unneeded COLOR macro.
5) Remove the fill_profile_color function. Simply use static
initialization.
6) Move #includes from .h to .cpp file.
7) Remove text_render_options(_t), which were not used anywhere.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Simplify starting application.
instead of declaring var. application, and the setting unused,
simple do the "new QApplication" without a variable.
Signed-off-by: Jan Iversen <jani@apache.org>
Replace constructs of the kind
s.toUtf8().data(),
s.toUtf8().constData(),
s.toLocal8Bit().data(),
s.toLocal8Bit.constData() or
qUtf8Printable(s)
by
qPrintable(s).
This is concise, consistent and - in principle - more performant than
the .data() versions.
Sadly, owing to a suboptimal implementation, qPrintable(s) currently
is a pessimization compared to s.toUtf8().data(). A fix is scheduled for
new Qt versions: https://codereview.qt-project.org/#/c/221331/
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Internal floating point (FP) calculations should be performed using double
unless there is a very good reason. This avoids headaches with conversions.
Indeed, the vast majority of FP calculations were already done using double.
This patch adapts most remaining calculations. Not converted where things
that were based on binary representations and variables which weren't used
anyway.
An analysis of all instances follows:
core/plannernotes.c, l.404:
This was a comparison between two floats. On the left side, first an integer
was cast to float then multiplied with and integer and divided by a constant
double. The right hand side was an integer cast to a float. Simply divide by
1000.0 first to convert to double and continue with calculations. On the right
hand side, remove the cast, because the integer will be implicitely cast to
double for comparison. This conversion actually emits less instructions,
because no conversion to double and back is performed.
core/planner.c, l.613:
Same analysis as previous case.
subsurface-desktop-main.cpp, l.155:
A local variable representing the version OpenGL version. Turn this into
integer logic. Not only does this avoid dreaded FP rounding issues, it also
works correctly for minor version > 10 (not that such a thing is to be
expected anytime soon).
abstractpreferenceswidget.[h/cpp]:
A widget where the position is described as a float. Turn into double.
desktop-widgets/divelogexportdialog.cpp, l.313:
total_weight is described as float. Use double arithmetics instead. This
instance fixes a truncation warning emitted by gcc.
The Windows auto-verbose + log file creation if starting
from a non-terminal has the problem that the print_version()
call is never made becase 'verbose' is updated programatically
in windows.c and not by the user (by passing -v).
To work around the issue:
- move the windows console creation call before *everything* else
- then immediatelly install the message handler
- then see if 'verbose' is set and explicitly call print_version()
print_version() now also has a flag (version_printed), to avoid
printing the version multiple times, if the user decided to add
an extra -v to the Desktop shortcut.
Signed-off-by: Lubomir I. Ivanov <neolit123@gmail.com>
Currently one has to explicitly use --win32console and/or
--win32log to enable a dedicated console (a console window
that opens next to the Subsurface window) or to enable file
logging on Win32.
This patch makes the following changes:
- removes the --win32* command line arguments
- removes the dedicated console window support
- if the app starts from a shortcut and not from a console, always
redirect stderr and stdout to _err & _out log files
- if the app starts from a console redirect stderr and stdout to that
console
Signed-off-by: Lubomir I. Ivanov <neolit123@gmail.com>
When verMinor is extracted with the glGetString() method,
use roundf() to round the GL MINOR version to an integer.
2.1 is split in 2 and 1.
Without this patch rounding issues are present as for 32bit IEEE
the default rounding makes (0.1f * 10.f) into 0.999999f.
Given the MINOR is a single digit for GL, calling roundf() on
0.999999 gives the expected result 1 for the MINOR.
Reported-by: Murillo Fernandes Bernardes <notifications@github.com>
Signed-off-by: Lubomir I. Ivanov <neolit123@gmail.com>
First, attempt to use glGetString(GL_VERSION), which works
on legacy profiles but has to be parsed (unsafe and vendor specific?).
If the above fails attempt to use the newer version API from
glGetIntegerv().
If both fail, fall back to a software renderer or exit.
If GLES is detected from glGetString() (for some odd reason -
e.g. emulated driver) show a warning and return early as we don't
handle it's versioning.
Signed-off-by: Lubomir I. Ivanov <neolit123@gmail.com>
Without this patch the user can pass a custom backend
like 'software' and the validation will still be performed.
In a case where the validation fails Subsurface will still exit.
With this patch the validation is performed *only* for the
standard backend which is OpenGL (or an empty QT_QUICK_BACKEND)
env. variable.
Signed-off-by: Lubomir I. Ivanov <neolit123@gmail.com>
Add a new static function to validate the availability of OpenGL
on this particular desktop system. It makes some calls to create
a platform agnostic GL context that renders to a offscreen surface.
Then it makes a couple of glGetIntegerv((GL_xxx_VERSION, ...) calls
to see if the GL profile version is at least 2.1.
In case any of the steps fail, a stderr message is shown and all
QtQuick based widgets would be rendered using a software renderer.
Testing was done in the case of the Google maps plugin and the
fallback seems to work on Windows, but further testing will be
required on all OS.
For the time being, the automatic fallback is only supported
on Qt 5.8.0 or later.
Signed-off-by: Lubomir I. Ivanov <neolit123@gmail.com>
We only do this in verbose mode to make sure a report contains the information
about the existing files used.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
report_error() now does this automatically. So all these odd places in which we tried
to make sure that we show errors are no longer needed.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
fill_computer_list() creates a Qt friendly
structure that contains all of the necessary
information about dive computers and it's
devices, and it's needed both in Qml and Widgets
to allow the user to download their dives.
This patch makes it possible to use the code
in QML without duplication.
Signed-off-by: Tomaz Canabrava <tcanabrava@kde.org>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Maintab is one of our most complex classes, and it's
something I'm not actually proud of. But it currently
works and the idea of splitting it was in my head for
quite a while.
This is the third or fourth tentative of splitting it,
and this time I let the most complex part of it untouched,
the Notes and Equipment tab are way too complex to untangle
right now on my limited time.
A new class 'TabBase' should be used for any new tab that
we may create, and added on the MainTab (see the new lines
on the MainTab constructor).
Also, Extra Info, Information, Photos and Statistics where
ported to this new way helping reduce the number of
lines and functions on the MainTab quite a bit.
Overall this is a step in the right direction for the future.
Signed-off-by: Tomaz Canabrava <tcanabrava@kde.org>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Adding --win32log as the first command line option on Windows
will now log all stdout and stderr output to the files
subsurface_err.log and subsurface_out.log in the working directory.
This change required a new argument 'bool logfile' to be added to:
subsurface_console_init() which is defined in all platform files
(linux.c, macos.c, etc.)
Example usage:
subsurface.exe --win32log -v -v -v
Signed-off-by: Lubomir I. Ivanov <neolit123@gmail.com>
No point in having it defined in each main program's .cpp. Especially
since the unit tests don't define them.
Signed-off-by: Thiago Macieira <thiago@macieira.org>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
