This commit fixes three different things:
- a memory leak in WeightModel::setData()
- getSetting() calling strdup() on a QByteArray
- a possible usage of memory after deallocation
Here's an explanation of the last issue (taken from the mailing list, slightly
adapted):
toByteArray(), as well as others "toSomething()" methods, returns
a new object which the compiler allocates on the stack. The compiler
will consider it a temporary data, and destroy it on the next line. So,
when one does
char *text= value.toByteArray().data(); // line 1
if (strcmp(description, text)) { // line 2
the compiler creates a QByteArray on line 1, calls ::data() on it, which
returns a valid char *, and assigns its value to "text". So far, so
good. But before jumping to line 2, the compiler destroys the temporary
QByteArray, and this will in turn invoke the QByteArray destructor,
which will destroy the internal data. The result is that on line 2,
"text" will point to some memory which has already been freed.
One solution is to store a copy of the temporary QByteArray into a local
variable: the compiler will still destroy the temporary QByteArray it created,
but (thanks to the reference-counted data sharing built in QByteArray) now the
destructor will see that the data is referenced by another instance of
QByteArray (the local variable "ba") and will not free the internal data.
In this way, the internal data will be available until the local variable is
destroyed, which will happen at the end of the {} block where it is defined.
Please note that when one uses the data in the same line, one doesn't need to
worry about this issue. In fact,
text = strdup(value.toString().toUtf8().data());
works just fine.
Signed-off-by: Alberto Mardegan <mardy@users.sourceforge.net>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
We now detect if a weight / cylinder with this description exists and if
not add it on the fly. We also remember the additional values (weight and
size / workingpressure) for new entries and take the values for these
fields into account when autocompleting.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
This is very much analogous to the way cylinders are implemented.
That means that just like with cylinders, if the user enters a new type
and hits 'tab' before hitting 'enter', Subsurface will crash.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
With this we should have tank editing mostly done.
See #122
(it's not quite fixed, we need the equivalent code for weight systems)
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
I had to immprove the TankInfoModel with two new methods,
insertRows and setData, because the delegate used this
model to show what kind of Tanks we are offering.
Since the user can enter a new type of Tank, it's important
to add this tank to all lists using the delegates.
I Also added two new methods on the delegate itself,
to correctly shows the data, and set the data on the
model. This also will help dirk with a working example
on how to edit things while using a delegate.
Signed-off-by: Tomaz Canabrava <tcanabrava@kde.org>
This is a fun one.
We only want to mark the divelist changed if the user actually changed
something. So we try really hard to compare what was entered with what was
there and only if it is different do we overwrite existing values and
record this as a change to the divelist.
An additional challenge here is the fact that the user needs to enter a
working pressure before they can enter a size (when in cuft mode). That is
not really intuitive. We work around this by assuming working pressure is
3000psi if a size is given in cuft - but then if the user changes the
working pressure, that changes the volume. Now going back and changing the
volume again does the trick. Or enter the working pressure FIRST and then
the volume...
This also changes the incorrect MAXPRESSURE to WORKINGPRESSURE and uses
the text WorkPress in English (Gtk code used MaxPress which was simply
wrong - this is just the design pressure or working pressure, not some
hard maximum. In fact, people quite commonly "overfill" these tanks.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
This patch adds basic editing functionality for Cylinders and Weigthsystems,
it still doesn't use delegates to show the data to the user in a better
way, and it does not take in consideration user preferences yet, but
it's a starting point.
Signed-off-by: Tomaz Canabrava <tcanabrava@kde.org>
So, the Cylinders and Weigthsystems got a new Trash icon,
and the interface already intercepts the clicks ( on all
columns ) and send this to the 'remove' method on boch
models. On the model I'm just filtering the indexes that
are not 'DELETE' and creating a stub method to be filled
later.
Signed-off-by: Tomaz Canabrava <tcanabrava@kde.org>
This patch adds support showing and for editing weigthsystems in the equipment tab,
so, now the two things that are missing are 'edit' and 'delete', wich are quite easy to do.
Signed-off-by: Tomaz Canabrava <tcanabrava@kde.org>
i
Added the code to show the cylinders from a dive,
this code also already permits additions from the
interface, so the user can click 'add' and insert
what he wants there.
Signed-off-by: Tomaz Canabrava <tcanabrava@kde.org>
This makes the DiveList be similiar to the GTK one, comparing
the size of the text. The current code makes text on the
table be 30% smaller.
Signed-off-by: Tomaz Canabrava <tcanabrava@kde.org>
The weight management widget added 500 grams / 0.5 lbs
when a new entry was added.
Signed-off-by: Henrik Brautaset Aronsen <subsurface@henrik.synth.no>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
A nonexisting temperature (mkelvin==0) was displayed as -273°C.
Weight was always displayed with an extra 500 grams/0.5 lbs.
Signed-off-by: Henrik Brautaset Aronsen <subsurface@henrik.synth.no>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Cosmetic commit to clean up some of the annoying typos in qt-ui
Signed-off-by: Henrik Brautaset Aronsen <subsurface@henrik.synth.no>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
And rip out all the code that Dirk put there to do that.
Signed-off-by: Tomaz Canabrava <tcanabrava@kde.org>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
A left click in the treeview header leads to a call to createIndex which
results in a null pointer dereference.
Signed-off-by: Amit Chaudhuri <amit.k.chaudhuri@gmail.com>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Establish some useful helpers and use them when updating the values.
One of the helpers (from statistics.c) puzzlingly doesn't link - so that's
ifdefed out.
Also had to re-arrange the settings reading code (it came too late) and to
extract the expanding code of the top dive from the settings reading code
(as it had no business being there to begin with).
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
This was written with massive hand-holding by Tomaz - actually, this was
mostly proposed via IRC by Tomaz and then implemented by me...
Right now because of the list-of-lists nature of the model we have the
small issue that every trip starts with a dark background dive, even if
the trip itself has a dark background.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
This code seems rather crude to me. I'm sure this could be done better.
This also makes the column alignment work again.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Now the list and dives will work in the same way that the GTK
version does. The code got changed heavly because the old one
was just looking at the dives and didn't worked like a tree.
small adaptations on the list view and model delegates because
of the changes done on this model.
Signed-off-by: Tomaz Canabrava <tcanabrava@kde.org>
For the stars on the dive table I had to rework a bit my
StarRating widget, because it used a pixmap for each widget
that were created. Not it uses only 2 pixmaps: the active
and inactive ones.
A new file was created named modeldelegates(h, cpp) that
should hold all delegates of the models. For the GTK / C
folks, a 'Delegate' ia s way to bypass the default behavior
of the view that's displaying the data.
I also added the code to display the stars if no delegate
is set ( good for debugging. )
Signed-off-by: Tomaz Canabrava <tcanabrava@kde.org>
Thanks to commit bdbfdcdfa0fb ('Ask Qt 4 to use the UTF-8 codec as the
"codec for C strings"') we no longer need the explicit UTF-8 conversion
when creating QStrings from char *.
Suggested-by: Thiago Macieira <thiago@macieira.org>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Just use the dive struct directly.
Suggested-by: Dirk Hohndel <dirk@hohndel.org>
Signed-off-by: Henrik Brautaset Aronsen <subsurface@henrik.synth.no>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
The DiveItem constructor had 13 variables. By passing it the full
dive we reduce that to 2.
[Dirk Hohndel: changed to use "struct dive *" instead of just "dive *"]
Signed-off-by: Henrik Brautaset Aronsen <subsurface@henrik.synth.no>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
We have these data structures for a reason. They provide context
about the units used and prevent mistakes. And of course they
are used everywhere else so we should use them here, too.
This also tries to display some more data and make things look
a bit more like the Gtk version when it comes to alignment and
formatting.
My guess is this will make Qt developers' eyes bleed. My apologies.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Amend the DiveItem class to avoid float in favour of int. Add getters
which return display friendly QStrings reflecting user preferences for
(e.g.) depth.
Modify DiveTripModel to support controlled alignment by column; right
align for depth and duration.
Fix problems with utf8 encoding on rating stars, degree symbols and
O2 subscript.
Signed-off-by: Amit Chaudhuri <amit.k.chaudhuri@gmail.com>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Moves the DiveTrip model related code to models.h
The DiveTripModel was implemented in three parts:
DiveTripModel.h
DiveTripModel.cpp
MainWindow.cpp (the code to populate the model)
This patch changes the DiveTripModel from it's original
implementation to the file models.h, wich should store
all models (Dirk requested the Qt developers to not create
2 files per class, but instead to use a file for functionality,
and data-models are one functionality.)
Besides that, this code cleans up a bit the style:
removed operator<< for .push_back, const references where they apply,
moved the internal DiveItem class to the .cpp since it should be visible
only to the DiveTripModel class, and redesigned the current interface of
the model to be identical of the GTK one (used the UTF8-star and 2
subscribed, for instance).
Amit (the creator of the original code) should comment here if it's ok
with my changes.
Signed-off-by: Tomaz Canabrava <tcanabrava@kde.org>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Added the code that will load and populate the Tank Info
ComboBox that`s used by the user to select the Cylinder
description.
Code curerntly implements more than the GTK version since
the GTK version of it was a plain-list, this one is a
table based model that can be used in ListViews ( like
we use now in the ComboBox ) but also in TableViews
( if there`s a need in the future to see everything
that`s catalogued in the Tank Info struct. )
Signed-off-by: Tomaz Canabrava <tcanabrava@kde.org>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
The Qtr_ hack isn't needed as in commit 720fc15b2dcd ("Introduce
QApplication") had already made sure that we are using gettext.
I didn't revert the two commits as I wanted to keep the added header
comments and fix the tooling in the Makefile as well.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
This should wrap gettext nicely and replace the "_()" macros we use in C
code.
Also added comments to the top of all the new files.
Suggested-by: Thiago Macieira <thiago@macieira.org>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
1 - Open File already open files, it tries to not break the Gtk version,
but some methods on the GTK version still need to be called inside Qt
because the code is too tight-coupled.
2 - Close file already close files, same comments for the open file dialog
applies here.
3 - The code for adding new cylinders in the cylinder dialog is done,
already works and it's integrated with the system. There's a need to
implement the edit and delete now, but it will be easyer since I'm
starting to not get lost on the code.
4 - Some functions that were used to convert unities have been moved to
convert.h ( can be changed later, put there because it's easyer to
find something that converts in a convert.h =p ) because they were
static functions that operated in the GTK version but I need those
functions in the Qt version too.
[Dirk Hohndel: lots and lots of whitespace and coding style changes]
Signed-off-by: Tomaz Canabrava <tcanabrava@kde.org>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
