Planned dives were still added by directly calling core code.
This could confuse the undo-machinery, leading to crashes.
Instead, use the proper undo-command. The problem is that as
opposed to the other AddDive-commands, planned dives may
belong to a trip. Thus, the interface to the AddDive command
was changed to respect the divetrip field. Make sure that
the other callers reset that field (actually, it should never
be set). Add a comment describing the perhaps surprising
interface (the passed-in dive, usually displayed dive, is
reset).
Moreover, a dive cloned in the planner is not assigned a
new number. Thus, add an argument to the AddDive-command,
which expresses whether a new number should be generated
for the to-be-added dive.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Currently, the filter is recalculated if a filter-entry is changed.
This also happens if the counts of a filter-entry changes. This
is to be avoided, as it causes unnecessary churn.
Therefore, send the proper role with the dataChanged() signal
and add a new slot, which invalidates only if a field with the
Qt::CheckStateRole is changed.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Instead of reloading all the filter, only increment / decrement the
count of the entries of added / removed dives.
Originally, this was planned to be done via the signals from the
divelist, but it turned out that this was suboptimal, because
if the filter decides that the new item is selected, this has to
be done *before* adding the dive. Otherwise, it wouldn't be shown.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
When adding dives to the list, set the filter flag accordingly.
Thus, dives that are hidden by the filter are not shown on
redo/undo.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Ultimately, we want to use a single dive-list and not replicate
it in the Qt-model code. To this goal, let's start with using
the same sort function.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
The current code cheats when it comes to move dives inside
a trip or move dives between trips: Instead of using the
*MoveRows() functionality, the dives are removed from and
re-added to the respective trips. This loses the selection.
Therefore, remember which of the moved dives are selected
and select them manually after they are re-added.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
The command-objects select a current item, but this selection
was not propagated to the front-end. The current item is the
base for keyboard-navigation through the dive-list and therefore
should be set correctly.
It took some experimentation to get the flags right:
QItemSelectionModel::Current
Hopefully, these are the correct flags across all supported
Qt versions!
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
If dives are deleted, the trip(s) containing the dives are expanded.
Thus, on undo it seems natural to re-expand the trip.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
For undo, it is crucial that commands don't modify existing dives.
Unfortunately, dive merging would write into the data-structures
of the to-be-merged dives. To prevent it from doing so, make the
input dives const-pointers.
This led to a whole cascade of functions that had to take const
and significant churn.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
The MergeDives and SplitDive commands used addDive() and removeDive()
calls to manage their dives. Unfortunately, these calls don't send
the proper signals and thus the dive-list was not updated. Instead,
use one- and two-element vectors, which are passed to addDives()
and removeDives() [note the plural].
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Select the proper dives after the add, remove, split and merge
dives commands on undo *and* redo. Generally, select the added
dives. For undo of add, remember the pre-addition selection.
For redo of remove, select the closest dive to the first removed
dive.
The biggest part of the commit is the signal-interface between
the dive commands and the dive-list model and dive-list view.
This is done in two steps:
1) To the DiveTripModel in batches of trips. The dive trip model
transforms the dives into indices.
2) To the DiveListView. The DiveListView has to translate the
DiveTripModel indexes to actual indexes via its QSortFilterProxy-
model.
For code-reuse, derive all divelist-changing commands from a new base-class,
which has a flag that describes whether the divelist changed. The helper
functions which add and remove dives are made members of the base class and
set the flag is a selected dive is added or removed.
To properly detect when the current dive was deleted it
became necessary to turn the current dive from an index
to a pointer, because indices are not stable.
Unfortunately, in some cases an index was expected and these
places now have to transform the dive into an index. These
should be converted in due course.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
In DiveListView, we have a very fundamental problem: When
On the one hand, we get informed of user-selection in the
DiveListView::selectionChanged() slot. This has to set the
correct flags in the C-backend.
On the other hand, sometimes we have to set the selection
programatically, e.g. when selecting a trip. This is done
by calling QItemSelectionModel::select().
But: this will *also* call into the above slot, in which
we can't tell whether it was a user interaction or an
internal call. This can lead to either infinite loops or
very inefficient behavior, because the current dive
is set numerous times.
The current code is aware of that and disconnects the
corresponding signal. This is scary, as these signals are
set internally by the model and view. Replace this
by a global "command executing" flag in DiveListNotifier.
The flag is set using a "marker" class, which resets the flag
once it goes out of scope (cf. RAII pattern).
In DiveListView, only process a selection if the flag is not
set. Otherwise simply call the QTreeView base class, to reflect
the new selection in the UI.
To have a common point for notifications of selection changes,
add such a signal to DiveListNotifier. This signal will be
used by the DiveListView as well as the Command-objects.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Don't delesect dives, when unregistering them from the backend.
If a previously selected dive is added, select it in the dive-list.
For this purpose introduce a SELECTED_ROLE to query the DiveTripModel
for selected dives.
Unfortunately, when adding multiple selected dives, current_dive_changed
is called for each of them, making this very slow. This will have
to be fixed in subsequent commits.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Now, that pointers to dives are stable, we might just as well
use dive * instead of the unique-id. This also affects the
merge-dive command, as this uses the same renumbering machinery.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
We have to avoid that undo/redo removes the currently edited
dive from under our feet. This code can be removed once proper
undo/redo (including editing) is implemented.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
If the autogroup flag is set, search for appropriate trips in
DiveAdd() and add the dive to this trip. If no trip exists, add
a new trip.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Previously, each dive-list modifying function would lead to a
full model reset. Instead, implement proper Qt-model semantics
using beginInsertRows()/endInsertRows(), beginRemoveRows()/
endRemoveRows(), dataChange().
To do so, a DiveListNotifer singleton is generatated, which
broadcasts all changes to the dive-list. Signals are sent by
the commands and received by the DiveTripModel. Signals are
batched by dive-trip. This seems to be an adequate compromise
for the two kinds of list-views (tree and list). In the common
usecase mostly dives of a single trip are affected.
Thus, batching of dives is performed in two positions:
- At command-level to batch by trip
- In DiveTripModel to feed batches of contiguous elements
to Qt's begin*/end*-functions.
This is conceptually simple, but rather complex code. To avoid
repetition of complex loops, the batching is implemented in
templated-functions, which are passed lambda-functions, which
are called for each batch.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
insert_trip() adds a trip to the backend, but merges trips if
there exists a trip with the same date. This is a disaster
for the MergeTrips command, because this command adds a new
trip and removes the previous two. Of course if the added trip
is merged, this cannot work.
Therefore, add an insert_trip_dont_merge() function, which
adds the trip, but doesn't merge.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
This refactors the undo-commands (which are now only "commands").
- Move everything in namespace Command. This allows shortening of
names without polluting the global namespace. Moreover, the prefix
Command:: will immediately signal that the undo-machinery is
invoked. This is more terse than UndoCommands::instance()->...
- Remove the Undo in front of the class-names. Creating an "UndoX"
object to do "X" is paradoxical.
- Create a base class for all commands that defines the Qt-translation
functions. Thus all translations end up in the "Command" context.
- Add a workToBeDone() function, which signals whether this should be
added to the UndoStack. Thus the caller doesn't have to check itself
whether this any work will be done. Note: Qt5.9 introduces "setObsolete"
which does the same.
- Split into public and internal header files. In the public header
file only export the function calls, thus hiding all implementation
details from the caller.
- Split in different translation units: One for the stubs, one for
the base classes and one for groups of commands. Currently, there
is only one class of commands: divelist-commands.
- Move the undoStack from the MainWindow class into commands_base.cpp.
If we want to implement MDI, this can easily be moved into an
appropriate Document class.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
AddDivesToTrip, CreateTrip, AutogroupDives, RemoveAutogenTrips
and MergeTrips basically all did the same thing as RemoveDivesFromTrip,
which was already implemented. Thus, factor our the common functionality
and hook it up to make all these functions undo-able.
Don't do the autogroup-call everytime the dive-list is rebuilt
(that would create innumberable undo-actions), but only on dive-load /
import or if expressly asked by the user [by switching the autogroup
flag].
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
The dive list is fed data by means of a sorted "DiveTripModel".
There are two modes: list and tree. This was implemented rather
elegantly with a general "TreeModel", which can represent trees
of arbitrary depths.
Nevertheless, we have at most two levels and on the second level
only dives can reside. Implementing proper model-semantics
(insert, delete, move) will be quite a challenge and implementing
it under the umbrella of a very general model will not make it
easier.
Therefore, for now, hardcode the model:
At the top-level there are items which may either be a trip
(can contain multiple dives) or a dive (contains exactly one dive).
Thus, we can completely de-virutalize the DiveItem and TripItem
classes, which are now trivial wrappers around dive * and dive_trip *.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
The undo-system now guarantees that pointers to dives are stable
throughout their lifetime. Therefore, replace the unique index by
pointers. This is a small performance improvement, but much more
importantly, it will make it more natural to transport a pointer
to the dive inside QModelIndex's private pointer.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
The whole undo system assumes that the indexes in the dive table
do not change under its feet. On desktop, there seems only one
exception left: editing of the dive time. To circumvent this,
hook editing of the dive-time to the already existing UndoShiftTime
command.
This introduces a temporary UI-inconsistency: this is the only
edit that is reflected in the undo-list. This will be fixed in
due course, when other edit actions are also made undoable.
UndoShiftTime is changed to take pointers to dives (which should
be stable by now) instead of uniq-ids.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
On dive editing, for every changed field the code looped through
the whole dive-list and modified the selected dives. Instead,
get the list of selected dives once and use that.
Whereas this may look like a gratuitous optimization, it will
make things easier for subsequent commits. Notably, we can
pass the list of selected dives to an "UndoObject".
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Keeping undo-history across load makes little sense. The user was
expressly reminded that they have unsaved work.
For import (from other logs or the dive-computer) an undo-functionality
would be desirable. Nevertheless, this is rather complex since
new and old dives are merged. Implementation would require a finer
backend<->undocommand interface. Thus, leave this for now until more
experience with the undo system is acquired.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
For this, an output-parameter was added to the backend merge_dives()
function. When non-zero, instead of adding the merged dive to
the preferred trip, the preferred trip is returned to the caller.
Since the new UndoObject, just like the delete-dives UndoObject,
needs to remove/readd a set of dives, the corresponding functionality
was split-off in a helper function.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
For this, the core functionality of the split_dive() and
split_dive_at_time() functions were split out into new
split_dive_dont_insert() and split_dive_at_time_dont_insert(),
which do not add the new dives to the log. Thus, the undo-command
can take ownership of these dives, without having to remove them
first.
The split-dive functionality is temporarily made desktop-only
until mobile also supports "UndoObjects".
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Play manual addition of dives via an UndoCommand. Since this does in
large parts the same thing as undo/redo of dive deletion (just the
other way round and only a single instead of multiple dive), factor
out the functions that add/delete dives and take care of trips.
The UI-interaction is just mindless copy&paste and will have to
be adapted.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Both callers have a dive * and transform that into an id,
the callee transforms it right back to the dive *. Simply pass
the dive directly. This will allow us to use the function for
dives that have not yet been added.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
The original undo-code was fundamentally broken. Not only did it leak
resources (copied trips were never freed), it also kept references
to trips or dives that could be changed by other commands. Thus,
anything more than a single undo could lead to crashes.
Two ways of fixing this were considered
1) Don't store pointers, but unique dive-ids and trip-ids.
Whereas such unique ids exist for dives, they would have to be
implemented for trips.
2) Don't free objects in the backend.
Instead, take ownership of deleted objects in the undo-object.
Thus, all references in previous undo-objects are guaranteed to
still exist (unless the objects are deleted elsewhere).
After some contemplation, the second method was chosen, because
it is significantly less intrusive. While touching the undo-objects,
clearly separate backend from ui-code, such that they can ultimately
be reused for mobile.
Note that if other parts of the code delete dives, crashes can still
be provoked. Notable examples are split/merge dives. These will have
to be fixed later. Nevertheless, the new code is a significant
improvement over the old state.
While touching the code, implement proper translation string based
on Qt's plural-feature (using %n).
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Disable the Download button when one of the fields vendor, product,
connection is not filled in. The app will crash when trying.
In addition, make the underlying core code to actual download
more safe by checking this, and silently fail instead of crash.
And, yes, this is a double fix in this scenario, but the core code
is used in more places, so better safe than sorry.
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
The code to disable a quick button has moved to the DC matching logic,
in order to inactivate the correct button also for USB DCs.
Signed-off-by: Joakim Bygdell <j.bygdell@gmail.com>
Since a known DC will have the name prepended to the BT/BLE addresss
we need to substring match the BT address.
Signed-off-by: Joakim Bygdell <j.bygdell@gmail.com>
By saving the device address together with the vendor and product we fix the
corner case where a user with two DCs would not get quick select buttons if they
where the same vendor and model.
Signed-off-by: Joakim Bygdell <j.bygdell@gmail.com>
For increased maintainability, use the same columns, roles and
the same accessor function for both dive-site models.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
The LocationInformationModel added two dummy sites to the front
of the list (add new dive site). This was never used - desktop
uses its own model, mobile only extracts the list of dive site
names with a custom function. Remove this functionality.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Editing of dive sites does not work via this model and the function
was broken anyway (it didn't subtract 2 from the index).
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Under certain conditions the user was presented an ugly
"invalid dive site" message. The condition would arise because
the proxy-model which selects the list of dive sites and the code
which creates a proposed dive site name had different filter
conditions:
- The proxy would select any dive site containing the text
- The name-proposing code searched for dive sites *starting*
with the text.
If the user entered a text contained by a dive site name, but
no dive site would start with the second line was filled with
a dummy text. This text would be kept if it contained the text
entered by the user.
To avoid this problem, if no dive site is found, use an empty
string instead. This will be filtered out by the proxy because
it does not contain the user-entered string.
Yes, that's horribly subtle, therefore add a comment. But ultimately,
this should be solved in a less brittle way.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
On both Mac and Linux cmake 3.12 complained that there were "no sources given
to target" for the Subsurface-mobile target, which made no sense at all (easy
enough to add debug output to ensure there were, in fact, sources given in the
call to add_executable()). But splitting this across two lines like this seems
to make it work both for older and newer cmake versions.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
After the previous commits, we now have a preference that nicely
preserves the state of the UI, and we have the well known git_local_only
global, that is used to denote whether we want to use to local repo
only, or we want to interact with the online cloud as well.
This commit gets rid of the now superfluous syncToCloud logic. Instead
we simply set the git_local_only directly.
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
