This makes it more obvious what we are doing. And won't make any difference
from a performance perspective.
Also converted the last call to connect using the old syntax to the new syntax.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Simply move the initialization of the logging function into its own method and
call that in the QMLManager constructor.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Instead of always showing info about the location, allow all data to be
captured in a more structured format - but only when the app is in
verbose mode.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Since switching to the mobile-models and removing grantlee,
DiveObjectHelper was demoted to a thin wrapper around string
formatting functions. The last user was removed in a previous
commit.
It was never a good idea, given QML's strange memory-management.
Let's remove it.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
When editing a dive, a DiveObjectHelper of the unmodified dive
was created to compare the edited with the old values. Since
the DiveObjectHelper is used here only as a pointless wrapper
around the formatting functions, call these functions directly.
However, note that the code is in principle wrong since the
change to the mobile-models, which do not use the DiveObjectHelper.
The real fix would be to reload the data from the model to prevent
going out-of-sync with respect to the formatting routines!
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>
Instead of programatically reload the completion models, listen
to the relevant signals in the models. To that goal, derive all
the models from a base class.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
These flags are not dive-related, therefore move their declaration
to the appropriate header file. Likewise, move their definition
from parse-xml.c to subsurfacehelper.c
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Desktop does not use mark_divelist_changed() anymore - all is done
via the undo machinery. Therefore move this function (and its
counterpart unsaved_changes()) to qmlmanager.cpp.
Ultimately, it probably should be removed from there as well, but
currently I don't dare to touch all the cloud-logic!
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
In analogy to the xml-parser add a device-table to git's parser-state.
Currently this is unused. In upcoming commits the git parser will
then be changed to add device nodes in this table instead of the
global device table. The long-term goal being to detach the
parsers from global state and to make dive-import fully undoable.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Add a device_table parameters to Command::importTable() and
add_imported_dives(). The content of this table will be added
to the global device list (respectively removed on undo).
This is currently a no-op, as the parser doesn't yet fill
out the device table, but adds devices directly to the global
device table.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
To include the device code in the undo system, we need functions
to check for the existence of devices and to add or remove them.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
If we want to avoid the parsers to directly modify global data,
we have to provide a device_table to parse into. This adds such
a state and the corresponding function parameters. However,
for now this is unused.
Adding new parameters is very painful and this commit shows that
we urgently need a "struct divelog" collecting all those tables!
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
We used a typedef "filter_preset_table_t" for the filter preset table,
because it is a "std::vector<filter_preset>". However, that is in
contrast to all the other global tables (dives, trips, sites) that we
have.
Therefore, turn this into a standard struct, which simply inherits
from "std::vector<filter_preset>". Note that while inheriting from
std::vector<> is generally not recommended, it is not a problem
here, because we don't modify it in any shape or form.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
This is mostly copy and paste of other git loading code. Sadly,
it adds a lot of state to the parser-state. I wish we could pass
different parser states to the parser_* functions.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
This is a bit painful: since we don't want to modify the filter
presets when the user imports (as opposed to opens) a log,
we have to provide a table where the parser stores the presets.
Calling the parser is getting quite unwieldy, since many tables
are passed. We probably should introduce a structure representing
a full log-book at one point, which collects all the things that
are saved to the log.
Apart from that, this is simply the counterpart to saving to XML.
The interpretation of the string data is performed by core
functions, not the parser itself to avoid code duplication with
the git parser.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Switch the mobile code to use the constraint-based filter. The one
thing that is still commented out is dive-site mode, since mobile
doesn't (yet) have a dive-site edit feature. And even if it had,
the dive list probably wouldn't be shown at the same time.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
If the dive timestamp changes, the dive could move in the dive list. But the
current dive actually doesn't change (it's still the same dive, right?). Yet
we need to update the dive list as well as the shown dive (especially if this
is after adding a dive, which is first inserted with the current time and then
updated with whatever the user enters).
Fixes: #2971
Suggested-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
The old code wasn't wrong, and likely the compiler turned this into something
that wasn't really terrible... but yeah, 5 unnecessary calls to a helper
function just bugged me. And I think the new code is much easier to read.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
This allows the mobile UI to reflect the three states that the dive list can be
in:
- changes that haven't been written to local storage
- there potentially are changes in local storage that were not synced with the cloud
- dive list is in sync with cloud storage
The last state could be misleading if the user access the cloud from a
different device and makes changes to the cloud storage from there, but from
the point of view of this device, the states are consistent.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
If we haven't connected at all to the cloud server we assume that there are
local changes. And whenever we save changes only locally, we also set that
flag.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
If you pass in the repo name, it isn't treated as cloud storage, but simply as
local git storage and imported (i.e., added) to the current dive list.
This allows the user to work around failed no-cloud->cloud transitions, merge
different accounts, and most importantly deal with situations were conflicts on
the server caused us to move a cache out of the way and potentially make dives
that were on the mobile device inaccessible to the user.
Once a UI is added, this allows the user to recover those dives (realistically
this is not really all that potentially 'dangerous' to do, but it's definitely
something that would best be done after talking to someone who understands the
cloud storage and can guide the user...).
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
I could have sworn that I have fixed this several times in various places,
but apparently (as shown by todays support emails) it's still possible to
setup a mixed case email address. So let's try to solve this problem at
the very top.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
In analogy to the timestamp -> QDateTime conversion, create a
common function.
1) For symmetry with the opposite conversion.
2) To remove numerous inconsistencies.
3) To remove use of the deprecated QDateTime::toTime_t() function.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
To reset the core data structures, the mobile and desktop UIs
were calling into the dive-list models, which then reset the
core data structures, themselves and the unrelated
locationinformation model. The UI code then reset various other
things, such as the TankInformation model or the map. . This was
unsatisfying from a control-flow perspective, as the models should
display the core data, not act on it. Moreover, this meant lots
of intricate intermodule-dependencies.
Thus, straighten up the control flow: give the C core the
possibility to send a "all data reset" event. And do that
in those functions that reset the core data structures.
Let each module react to this event by itself. This removes
inter-module dependencies. For example, the MainWindow now
doesn't have to reset the TankInfoModel or the MapWidget.
Then, to reset the core data structures, let the UI code
simply directly call the respective core functions.
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>
Calling qApp->processEvents() in QMLManager::setNotificationText()
caused crashes, because it could lead to the context-menu that
initialized the call being deleted. Something that QML apparently
doesn't like.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
If the user switches away from Subsurface-mobile and back while we are
in our initialization phase, that code might run more than once leading
to undesirable results.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
The strange enum casts (that apparently we needed in order to have strongly typed
enums in QML) are really ugly and confusing in the code. Since we had this three
times it seemed worth to create a little helper that hides this nonsense.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
We know that we have valid credentials, so calling openLocalThenRemote()
makes no sense, worse, it doesn't respect the special case of
transitioning between accounts. That's the whole point of having a
separate loadDivesWithValidCredentials() call.
Once we have transitioned state, we need to record the now current state
as the old state so that future transitions will work.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
The noCloud variable contains our desired state, we need to test for the
previous state for this to make sense.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
We haven't supported the GPS web service for a long time.
Also, add another message for the log to be able to more easily trace
one error case.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
If we're using the undo stack for determining the whether there are
unsaved changes, we have to mark the undo stack as clean after save.
This duplicates code from desktop. It is planned to unify this in
the near future.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
In analogy to desktop, also consider the state of the undo-stack when
testing for unsaved changes. This prevents us from missing changes.
This adds duplicate code, which will be unified in the near future.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
These appear to be the only cases where we forgot to ensure data is
saved. And pastDiveData is currently not called as the UI to use it has
been removed.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
The code was using the pre-editing dive site name, which appears
weird. Not sure if that was on purpose.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
QMLManager::checkLocation() would only return true if the
dive-site itself was changed, not if the dive-site was set
to a different site.
Thus, in some cases edit events could be lost.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
This way the launch screen is shown significantly shorter on a device,
and instead the user sees our progress notifications.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
We have convinced ourselves that only the main thread will ever trigger
a save operation, therefore the locking is not needed (and it has
recently started to cause user problems where local changes aren't saved
to storage and get lost).
Fixes#2718
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Instead of not showing a notification until after we are done loading
the data into our models, tell the user that we are about to do that.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
This simply passes the text to the existing notification text mechanism.
The call to processEvents() ensures that Qt processes the UI events
after we updated the notification and that the user gets timely updates;
otherwise the UI might look like the app is hanging.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
When editing the dive site of a dive, the dive-table of the
corresponding dive site was not properly updated by the undo
commands. Try to get this right.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
QMLManager::changesNeedSaving() behaves differently on iOS:
it only saves locally with saveChangesLocal(), whereas all
other OS save to cloud with saveChangesCloud(). Nevertheless,
even for other OS saveChangesLocal() is called even though
that will be called in saveChancesCloud anyway. Therefore,
compile the saveChangesLocal() call in changesNeedSaving
conditionally.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
This should deal with the rather confusing 'No dive in dive list' shown
while loading and processing the dive list.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Instead of the crude and error prone bool, let's just use the right tool
for this job.
In order to avoid issues with a goto across a mutex boundary, this
slightly restructures the code in one place - 'git show -w' makes it
clear that this is really rather simple in its changes.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
I'm suspicious that an issue that some people have seen around changes
not being saved is caused by our handling of alreadySaving.
When starting with Subsurface-mobile in no-autosync mode, we were able
to end in a scenario where alreadySaving was true, even though there
were no unsaved changes. And in that case no changes made during such a
session were saved unless the user forced a manual sync with the server.
This commit radically changes our approach to handling the flag. It
moves it right next to the actual calls into code that could modify git
storage, and ensures that the flag can never stay set.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
