To extend the undo system to dive sites, the importers and downloaders
must not parse directly into the global dive site table. Instead,
pass a dive_site_table argument to parse into.
For now, always pass the global dive_site_table so that this commit
should not cause any functional change.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
To enable undo of dive site functions, it is crucial to work
with different dive site tables. Therefore add a dive site table
parameter to dive site functions. For now, always pass the global
dive site table. Thus, this commit shouldn't alter any functionality.
After this change, a simple search for dive_site_table reveals all
places where the global dive site table is accessed.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
See https://www.kdab.com/goodbye-q_foreach/
This is reduced to the places where the container is const or can be made const
without the need to always introduce an extra variable. Sadly qAsConst (Qt 5.7)
and std::as_const (C++17) are not available in all supported setups.
Also do some minor cleanups along the way.
Signed-off-by: Rolf Eike Beer <eike@sf-mail.de>
printGPSCoords() returned a newly allocated C-style string. Most
callers simply made a QString out of it and freed the C-style string.
This is paradoxical, as printGPSCoords internally works with QStrings
and converts them to C-style on return.
Therefore, let printGPSCoords() return a QString and create a
printGPSCoordsC() wrapper for the two C-callers.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
The printGPSCoords() function returns a copied C-style string. Since
the owndership is transferred to the caller, the correct return type
is "char *" instead of "const char *".
Thus a number of casts when calling free can be removed.
Moreover a number of callers didn't free the string and thus were
leaking memory. Fix them. Ultimately we might want two versions
of the function: one for QString, one for C-style strings.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
And offer an option to show all devices in the settings. This is intentionally
not stored in the preferences as this should never be needed. We don't support
BT or BLE dive computers that we don't recognize. This is a last resort in case
a new firmware were to change the name or some other weird issue causes us not
to recognize a dive computer - and that should be fixed instead of worked
around.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
process_imported_dives() takes four boolean parameters. Replace these
by flags. This makes the function calls much more descriptive. Morover,
it becomes easier to add or remove flags.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Since process_imported_dives() can add dives to a newly generated
trip, this need not be done in the downloading code. This makes
data flow distinctly simpler, as no trip table and no add-new-trip
flag has to be passed down to the libdivecomputer glue code.
Moreover, since now the trip creation is done at the import step
rather than the download step, the latest status of the "add to
new trip" checkbox will be considered.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
If this flag is set, dives that are not assigned to a trip will
be assigned to a new trip. This flag is set if the user checked
"add to new trip" in the download dialog of the desktop version.
Currently this is a no-op as the dives will already have been
added to a new trip by the downloading code. This will be removed
in a subsequent commit.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Split the process_imported_dives() function in two:
1) process_imported_dives() processes the dives and generates
a list of dives and trips to be added and removed.
2) add_imported_dives() calls process_imported_dives() and
does the actual removal / addition of dives and trips.
The goal is to split preparation and actual work, to
make dive import undo-able.
The code adds extra checks to never merge into the same
dive twice, as this would lead to a double-free() bug.
This should in principle never happen, as dives that
compare equal according to is_same_dive() are merged
in the imported-dives list, but perhaps in some pathologival
corner-cases is_same_dive() turns out to be non-transitive.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
When importing log-files we generally want to merge trips. But
when downloading and the user chose "generate new trip", that
new trip should not be merged into existing trips.
Therefore, add a "merge_all_trips" parameter to process_imported_dives().
If false only autogenerated trips [via autogroup] will be merged.
In the future we might want to let the user choose if trips
should be merged when importing log-files.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
The old way of merging log-files was not well defined: Trips
were recognized as the same if and only if the first dives
started at the same instant. Later dives did not matter.
Change this to merge dives if they are overlapping.
Moreover, on parsing and download generate trips in a separate
trip-table.
This will be fundamental for undo of dive-import: Firstly, we
don't want to mix trips of imported and not-yet imported dives.
Secondly, by merging trip-wise, we can autogroup the dives
in the import-data to trips and merge these at once. This will
simplify the code to decide to which trip dives should be
autogrouped.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
To allow parsing into arbitrary trip_tables, add the corresponding
parameter to the parsing functions and the parser state. Currently,
all callers pass the global trip_table so there should be no change
in functionality. These arguments will be replaced in subsequent commits.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Currently trips are added to the global trip table. If we want to
make dive-import undoable, we should be able to parse trips of a
log-file into a distinct table. Therefore, add a trip_table
parameter to
- insert_trip()
- create_and_hookup_trip_from_dive()
- autogroup_dives()
- unregister_trip()
- remove_dive_from_trip()
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Currently, all trips are kept in a linked list. Replace the list
by a table in analogy to dive_table. Use this to keep the trip_table
sorted as suggested by dump_trip_list(). When inserting a trip into
the table do that after adding the dives, to avoid warnings coming
out of dump_trip_list().
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Rename
- dive_get_insertion_index() -> dive_table_get_insertion_index()
- unregister_dive_from_table() -> remove_from_dive_table()
- get_idx_in_table() -> get_idx_in_dive_table()
- sort_table() -> sort_dive_table()
This will make it more straight-forward to generate these functions
from macros.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Commit 68961a169e made it impossible
to edit a dive site on mobile if a dive-site was already set: If
divesite was non-null, no actions were taken. Remove the conditional.
Reported-by: Miika Turkia <miika.turkia@gmail.com>
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
For some devices the BT device name is different from the product name.
Make sure that name is available to the mobile UI. This helper fills it
in from the scan data (based on the device address).
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Apparently this does the trick of updating the dive list when pasting
data on mobile. Tnx janmulder.
Signed-off-by: Miika Turkia <miika.turkia@gmail.com>
DCDeviceData was using that weird pattern where the instance
variable was set in the constructor. There is no apparent
reason to do so, therefore convert to a "normal" singleton.
Access that directly in QMLManager instead of saving it in
a member variable first.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
The only remaining use of the tripflag was to mark dives that
were removed explicitly from a trip, i.e. shouldn't be autogrouped.
Therefore replace the enum by a simple boolean.
Currently, there is no way of unsetting the notrip flag. But this
shouldn't result in a user-visible change.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
The dives of each trip were kept in a list. Replace this by a
struct dive_table. This will make it significantly easier to
keep the dives of a trip in sorted state.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
There were two versions of the insert_trip() function: one
would merge trips if a trip with the same date already existed,
the other wouldn't. The latter was introduced with the dive-list
undo work.
The problem is that the "date" of a trip (i.e. the first dive)
seems ill-defined as this is a volatile value. Moreover in
the context of making dive-import undoable this is a very
dangerous notion, as the caller needs control over when the dives
are added to a trip.
Therefore, unify these two functions and never merge trips.
The decision on merging dives now has to made by the caller.
This will be implemented in a future commit.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Naturally the paste button should be disabled, if this is the case, but
it still makes sense to do this check.
Signed-off-by: Miika Turkia <miika.turkia@gmail.com>
Initial implementation/prototype of copy-paste support for
Subsurface-mobile. The UI part is really lacking; right now the copy
button is initially visible and paste is achieved by long press on a
dive and clicking the paste button when it appears. Delete is currently
not possible at all, as I just failed to layout the buttons properly
using QML. It just sounds so simple, to put all the copy-paste-delete
buttons next to each other...
The data to be copied is currently hard-coded. A dialog to choose
inteded fields would be nice, but it'll take quite a bit effort to get
used to QML enough to be able to hack something together.
Anyway, this seems to work, even though the UI is not always reflecting
the paste without switching dives (when testing on laptop).
Signed-off-by: Miika Turkia <miika.turkia@gmail.com>
While not something that many will use, editing a dive on
Subsurface-mobile should not result in data loss.
This makes the divemaster field behave in the same way as the buddy
field with regards to multiple entries.
Fixes#1853
Signed-off-by: Joakim Bygdell <j.bygdell@gmail.com>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Replace the UUID reference of struct dive by a pointer to dive_site.
This commit is rather large in lines, but nevertheless quite simple
since most of the UUID->pointer work was done in previous commits.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
This changes more of the dive-site interface to return pointers
instead of UUIDs. Currently, most call sites directly extract
UUIDs afterwards. Ultimately, the UUIDs will be generally replaced
by pointers, which will then simplify these callers.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
As a first step in removing dive-site uuids, change the interface
of the get_dive_site_*() functions to return pointers instead
of uuids. This makes code a bit more complicated in places where
the uuid is extracted afterwards (needed NULL check). Nevertheless,
these places should disappear once pointers instead of uuids are
stored in the dive-structures.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Instead of having people treat latitude and longitude as separate
things, just add a 'location_t' data structure that contains both.
Almost all cases want to always act on them together.
This is really just prep-work for adding a few more locations that we
track: I want to add a entry/exit location to each dive (independent of
the dive site) because of how the Garmin Descent gives us the
information (and hopefully, some day, other dive computers too).
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
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>
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>
This simple one-liner fixes an actual bug. On switching from
a no-cloud account to a actual cloud account, the dives from
the no-cloud are added to the actual cloud account. And indeed
the dives appear correctly. However, when exiting the app
right away, these added dives are not commited to the local
storage. Simply, the divelist needs to be marked dirty.
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
process_imported_dives() is more efficient for downloaded than for
imported (from a file) dives, because it checks only the divecomputer
of the first dive.
This condition is checked via the "downloaded" flag of the first
dive. Instead, pass an argument to process_imported_dives().
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Dives are now in all cases imported via distinct dive_tables.
Therefore the "preexisting" marker is useless. Remove.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
process_dives() is used to post-process the dive table after loading
or importing. The first parameter states whether this was after
load or import.
Especially in the light of undo, load and import are fundamentally
different things. Notably, that latter should be undo-able, whereas
the former is not. Therefore, as a first step to make import undo-able,
split the function in two versions and remove the first parameter.
It turns out the the load-version is very light. It only sets the
DC nicknames and sorts the dive-table. There seems to be no reason
to merge dives.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
We only store the address part of the connection name, so don't try to find an
exact match, try to find the sub-string.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
The unit types are set from system locale when the app starts.
We need to explicitly set the units to match the unit system that is saved in
the git repo. Or we end up with situations where the preferences and git say
"metric" but the units are "imperial".
Signed-off-by: Joakim Bygdell <j.bygdell@gmail.com>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Use qPrefLocationService::set_time_threshold and remove from
qmlprefs.cpp and qmlmanager.cpp
Remark: mobile UI shows time in minutes, while it is stored (and calculated)
in seconds. Therefore a /60 when reading and *60 when setting.
Signed-off-by: Jan Iversen <jani@apache.org>
Remove distanceThreshold from qmlprefs and use qPref instead
update qml
no user experience change
Signed-off-by: Jan Iversen <jani@apache.org>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Despite the fast that this code is sitting in core, its used mainly
from mobile. In 987e221f8e, the buttons to interact with the GPS
webservice were deleted from the UI. Now, delete all the code that
was used under these buttons.
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
When installing for the first time cloudCredentials needs to be added,
this commit a problem with updating them correctly
this problem was caused by
da6e8a4cd5
Signed-off-by: Jan Iversen <jani@apache.org>
To enable undo of divelog-importing it is crucial that parse_file()
can parse into arbitrary dive tables.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
A trivial cleanup: replace void by properly typed pointers in
cylinder_none() and weightsystem_none(). Moreover, remove the
unused function no_weightsystems().
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
remove use of SettingsObjectWrapper::
remove include of SettingsObjectWrapper.h
use qPrefFoo:: for setters and getters
replace prefs.foo with qPrefXYZ::foo() where feasible
(this expands to the same code, but gives us more control
over the variable).
Signed-off-by: Jan Iversen <jani@apache.org>
remove General from SettingsObjectWrapper and reference qPrefGeneral
update files using SettingsObjectWrapper/General to use qPrefGeneral
this activated qPrefGeneral and removed the similar class from
SettingsObjectWrapper.
Signed-off-by: Jan Iversen <jani@apache.org>
remove LocationService from SettingsObjectWrapper and reference qPrefLocationService
update files using SettingsObjectWrapper/LocationService to use qPrefLocationService
this activated qPrefLocationService and removed the similar class from
SettingsObjectWrapper.
Signed-off-by: Jan Iversen <jani@apache.org>
With the new setup we need to know which state we are coming from
when we are saving cylinder related info. When we are adding
a new dive we explicitly should save cylinder data to the first cylinder.
Signed-off-by: Joakim Bygdell <j.bygdell@gmail.com>
Same as for cylinder info, we need to make sure that the gasmixes gets saved to the correct cylinder.
Signed-off-by: Joakim Bygdell <j.bygdell@gmail.com>
Save the edited cylinder in the correct slot.
Since the cylinder number and the used cylinder number need not be
the same we first need to test if the cylinder are used.
Signed-off-by: Joakim Bygdell <j.bygdell@gmail.com>
Parse the device string and try to figure out what was plugged in.
In some cases we know exactly which vendor and product was plugged in,
in other cases we only know which vendor it was, in some cases we don't
even know that (if all we see is a generic FTDI cable).
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
And try to guess which one from the device string we get from the Intent.
The function is named to indicate its future use (because once the user
plugs in such a device, we should show the download page).
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
If the user plugs in a device on Android we get a device string that
should allow us to figure out which dive computer was plugged in. Make
that string available to the QML UI.
Right now all we do is log it.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
add enum to qPref and remove elsewhere
update source core to reference qPref.
the enum cannot be in pref.h because it is to be used in qml and Q_ENUM
need the enum to be defined as part of the class
Signed-off-by: Jan Iversen <jani@apache.org>
Use Q_ENUM instad of Q_ENUMS (which is depreciated) since it does the
meta registration for all Qt platforms.
Q_ENUM require the enum to be defined in the class and cannot refer to
a global class, therefore copied enum to class.
This commit is made to get the release to work, with minimal changes,
this class will be moved to qPref and the double definition solved
Signed-off-by: Jan Iversen <jani@apache.org>
Remove cloud_storage_status from qmlprefs.h.
usage to qPref::
enum cloud_storage_status is not used from C, but only from C++, and
having the same structure defined multiple times is a maintenance
challenge.
Signed-off-by: Jan Iversen <jani@apache.org>
The clipboard fails if we attempt to copy more than 1MB of data. But the
data buffer used is shared between all transactions 'in flight' and we
cannot tell what else is currently using that buffer. Limiting ourselves
to 500k of text for the logfiles seems reasonable and hopefully makes it
more likely that the transaction will succeed (sadly, Qt doesn't tell us
if it failed).
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
On some devices (e.g., a Sony Xperia phone) the GenericDataLocation is
not app writeable. Instead of just giving app, try a few other default
locations as well (and since all of these are actually string lists,
try all of the options that Qt gives us).
Reasonably, we should only set the libdivecomputer logfile name if we
found a writeable location.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
This way the user doesn't inadvertantly end up with information from a previous
run of Subsurface-mobile when they copy the logs to the clipboard.
Not sure we should do the same when building for desktop, so right now it's
only when building for a device.
Reported-by: Thomas Fänge <thomas.fange@gmail.com>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
We want to allow people to keep dives they collected without a cloud
account. The code was mostly there, we just got confused about the
existing status because we ran through this twice (no cloud -> unknown
-> verified). This way we explicitly remember this kind of transition.
Fixes#1404
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
add settings variables/functions to qmlprefs
remove settings variables/functions from qmlmanager
change manager. to prefs. in qml files for setting variables/functions
Signed-off-by: Jan Iversen <jani@apache.org>
Remove Q_OBJECT and qml properties from DCDeviceData class
Remove DCDeviceData register from mobile-helper.cpp
Change DCDeviceData constructor to be without parameters
Signed-off-by: Jan Iversen <jani@apache.org>
Subsurface-mobile has a long startup time; in order to isolate the problem(s) a
timer is added to see where time is "lost".
The collected startup times are added to the clipboard together with the other
logs, allowing test users to report back.
All this is only enabled when compiling with -DENABLE_STARTUP_TIMING
Closes#1340
[Dirk Hohndel: collapsed multiple commits and minor white space cleanups, added
missing QMutex variable]
Signed-off-by: Jan Iversen <jani@apache.org>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
The reply member variable was used to access the reply in the
handleSslErrors, handleError and retrieveUserid slots. This is a
very scary proposition in the light of multi-threading. Instead,
the reply can be accessed by using the QObject::sender() function.
Thus, we can remove the member variable.
The request member was just downright weird. This was only used
locally to describe a network request. Since QNetworkAccessManager::get()
copies the request, it can be destructed right away. Nevertheless,
the data was kept as a subobject. Remove member and make it function-local.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Read libdivecomputer.log file and append to clipboard
Remark, subsurface_open is not available in iOS so using
QFile instead.
Signed-off-by: Jan Iversen <jani@apache.org>
Users need a simple way to report download problems. Asking them
to go into settings, activate logging, and repeat the download
is too much.
libdivecomputer.log will always be generated, but overwritten, with every
download, so the latest log is ready.
The download is very slow due to BT, so there are no impact of having the log
active, neither in terms of size (a concern on small mobiles) or write time.
Signed-off-by: Jan Iversen <jani@apache.org>
on iOS it is practically impossible to copy the App log
to e.g. a mail! in iOS 11 the log file is stored within
the subsurface container and you first need to copy (actually
using the clipboard) out from there to the "normal" document
shared space, before it can be used.
At least iOS users (and I believe Android users) are not really
used to work with files, so the process is not easy to document
in an understandable way.
The alternative is to provide a button, which simply puts the
log on the general clipboard, allowing it to be pasted in a
multitud of applications.
Signed-off-by: Jan Iversen <jani@apache.org>
fake_dc() used to return a statically allocated dc with statically
allocated samples. This is of course a questionable practice in
the light of multi-threading / resource ownership. Once these
problems were recognized, the parameter "alloc" was added. If set
to true, the function would still return a statically allocated
dc, but heap-allocated samples, which could then be copied in
a different dc.
All in all an ownership nightmare and a recipie for disaster.
The returned static dc was only used as a pointer to the samples
anyway. There are four callers of fake_dc() and they all have access
to a dc-structure without samples. Therefore, change the semantics
of fake_dc() to fill out the passed in dc. If the caller does
not care about the samples, it can simply reset the sample number
to zero after work.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
So far we only wrote messages to subsurface.log on Android (since we couldn't
figure out how to make that file user accessible on iOS). Now that that's
fixed, we also need to actually write to the file in the first place.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
By creating it in the Documents path and setting the two magic keys,
iOS will make the log file available to the user.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
strdup(qPrintable(s)) and copy_string(qPrintable(s)) were such common
occurrences that they seem worthy of a short helper-function.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
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>
registerError() may be called from a different thread context. Passing
the message as a const-reference may lead to a dangling reference.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
As we need to update our new stringlist models on multiple
locations, just refactor them into a new function.
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
See also e6e1473e6. The construction of the locationlist
was not the same as the 3 previous lists, and it needs
the inclusion of a new model file (divelocationmodel.cpp)
in the mobile app. In addition, as the mobile app is mainly
interested in a simple stringList (model) to populate a HintsText
field (or maybe later a combobox), this stringlist is added
to the model, to easy interfacing with QML.
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
See e6e1473e6. Exact same commit but here for the
list of divemaster. The careful reader will spot a
small addition to the clearDetailsEdit() QML function.
Two more field are cleared.
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
This is the first of a set of commits that are (very) similar.
It appeared that a number of more or less static lists, which are
constructed by a loop over all dives in the logbook, were executed
when changing focus to a next dive. For example, the in this
commit addressed list of used dive suits.
What was wrong was that the suitList was linked to a dive. There
is only a need to construct the list of used suits when data is
changed (and obviously, once on startup of the app). Further, it
appeared that a lot of code was duplicated and that we can use
(in this case) the same code from the desktop completionmodels.cpp.
Basically, this commit involves the following changes:
- include completionmodels.cpp in mobile and desktop (so move
it from the desktop only category to the generic category).
- remove double code from DiveObjectHelper.cpp
- Do not differentiate in the init phase and the normal refresh
of the list
- the per dive logic is now only the getting of a previously
constructed list (in init or update of the divelist).
There are no visible changes in the UI, other than a better
performance when scrolling over dive details.
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
Remove the global error buffer and pass the error string directly
to the frontend. The frontend is then responsible for accumulating
errors.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Instead of manually logging errors after each potentially
error-producing function, use the error-callback. The error texts
are accumulated in the QMLManager object for further use.
The text is transported to the QMLManager object via a queued
connection. Thus, errors can be reported from other threads
without having to deal with manual locking.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
QMLManager::tryRetrieveDataFromBackend() was a one-liner calling
void QMLManager::checkCredentialsAndExecute() with a pointer-to-member.
The latter was never called with a different pointer, therefore
fold the latter into the former and remove the indirection.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
Since the timer will be started on first message anyway, we might
just start it at application startup.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
This was probably an oversight - the timer was only shown for very
specific messages. This was change to show the timer for all messages.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
These were assigned the member variables QMLManager::m_cloudPassword
and QMLManager::m_cloudPassword. Use the member variables directly.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
QMLManager was full of redundant getter functions of the type
bool QMLManager::locationServiceAvailable() const
{
return m_locationServiceAvailable;
}
These can be removed changing READ keywords to MEMBER keywords
in the QPROPERTY declarations.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
There are ca. 50 constructs of the kind
same_string(s, "")
to test for empty or null strings. Replace them by the new helper
function empty_string().
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
This is a somewhat hacky commit. For a very long time, the delete
from the divelist on mobile crashed. That is, not always for anyone,
but for me almost consistently. This commit tries to solve it.
I found that trying to save the delete immediately after removing
data from the underlying model seemed to cause the crash. Hacking
around, I found that a simple beginResetModel/endResetModel between
the delete of the underlying model data and actual save is
sufficient to solve the crash.
The big question is, why does this all work? I suspect some of race
condition between deleting model data, and giving the QML engine
the opportunity to do its thing.
This is also related to issue #311, but that is not implemented
here.
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
This is the only case where C-code sets the current file.
Remove this call for a better separation of C-backend and
C++-frontend parts.
There were four callers of clear_dive_file_data(). Two of them
would call set_filename() anyway. For the remaining two add an
explicit call to set_filename().
This commit fixes a bug introduced in commit b3901aa8f9:
The cloud-online menu entry was still enabled after "closing" the
cloud storage.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
In QMLManager::saveCloudCredentials(), clear_dive_file_data() was
called just before calling openLocalThenRemote(). The latter calls
the former anyway, so remove the former.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
See issue #949. In the dive list, dates are shown in short format, and
when we start to edit an existing dive, the date field on the edit
page is shown in sort format. However, when adding a new dive, the
initial date shows up in long (normal) format. This in not only
inconsistent, but also introduces the parsing problem, as described
in the mentioned issue.
This can be fixed by using short formatted dates in all cases in
the mobile app. As the screen real estate is precious on mobile,
this seems the most logical choice.
Fixed: #949
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
This commit fixes a very subtle bug. Probably there for ages,
but never noticed. When manually adding a dive, and canceling
the add right away, the divetable was extended by 1 dive (using
the core function add_single_dive), but canceling never removed
that dive (using delete_single_dive). This is corrected here.
Notice that commit ef543da5af claims to fix issue #950.
With this commit, however, that issue can be seen sometimes
again.
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
Instead use the application-wide instance. Creating a local
object defeats the whole purpose of these objects - nobody
can receive signals in case the settings changed.
No other cases of locally created SettingsObjectWrapper
objects were found.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
This fixes an annoying bug. When entering an existing divesite
in the mobile app, always a new divesite (with same name) was
created, resulting in multiple sites on the same gps location.
So, be more carefull in checking and creating a new site.
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
While the autocomplete function only works for the first entry
adding multiple comma separated buddies can still be done.
Fixes#608
Signed-off-by: Joakim Bygdell <j.bygdell@gmail.com>
Otherwise the divecomputer has to be in pairing mode
at app start time.
Unfortunately, this leaves less space for the progress message.
My time/qml knowledge does not suffice to move that to the next line
(when moving that out of the RowLayout it overlaps with
the buttons).
Signed-off-by: Robert C. Helling <helling@atdotde.de>
This enable use of current location on Add/Edit
Signed-off-by: Murillo Bernardes <mfbernardes@gmail.com>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
The second argument to set_filename() is a boolean, but was given
an empty string. Since an empty string evaluates to true, directly
pass true for clarity.
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
When the user entered a dive site using autocompletion, it
is a known site, of which we might have a GPS location already.
Just fill the known site coordinates.
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
Make all char * pointers in pref.h const to make it clear that these
strings are not mutable. This meant adding a number of (void *) casts
in calls to free(). Apart from being the right thing to do, this commit
makes the code more consistent, as many of the strings in pref.h were
already const.
While touching core/qthelper.cpp turn three instances of (void*) into
(void *).
Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>
First, obviously, I could have squashed this small commit into
the previous one, but I explicly decided not to. It shows
the fragility of all this credential processing code, and
the complex flow control troughout the code.
Testing on a brand new install, and immediately going for
a no cloud setup, the PIN screen was shown, instead of an
empty divelist. Looking at this small code change shows why.
In case of a no cloud situation, there is no PIN to verify
(or email and passwd rules checked). So just do not force us
there.
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
Before this change, there was only one way to create the local
no cloud repo on the device. The user needed to add at least
one dive to the no cloud account (so that there is something
to save). While this worked in some scenarios, it could also
get things in an inconsistent state: credential status = CS_NOCLOUD
but no local repo. This was a dead end.
In this commit, the creation of the no cloud repo is made more
explicit. When asking for no cloud mode, just create an (empty)
repo for it when it does not yet exist, and otherwise, just
open the existing (possibly empty) repo.
Now, a user can have no cloud repo, next to (any number of)
cloud accounts.
This leaves one functional aspect left: how does a user abandon
the no cloud repo, by merging his data into a true cloud
account. This is code for this, that tries to do this merge in
a smart way. This seems to be broken (too). To be clear: this
is no part of this commit.
Fixes: #667
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
This commit implements possible switching BT on and off during a session,
so not needing a restart of the app when the user forgot to switch
it on when starting the app.
For this, the following needed to be done: 1) create a handler that
reacts on local BT device status changes. 2) repopulate the connection
list in the download screen when a BT status change is detected.
Notice the subtile change of the Q_INVOKABLE btEnabled() function
to a Q_PROPERTY. This gives a nice dynamic behaviour when
switching BT on/off with the app open.
Fixes: #556
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
In certain places the '(int)' cast is used, while in other the
llrint() or lrint() functions. Make the conversation from degrees
in the 'double' form to the 'int' degrees_t consistent using lrint().
lrint() is the function which should give the best results,
because it accepts a 'double' and results in a 'long'
even if degrees_t is 'int'. If the truncation from 'long' to 'int'
is discarding some of the precision then the next step
would be to turn degrees_t into a 64bit signed integer type.
Possible fix for #625.
Signed-off-by: Lubomir I. Ivanov <neolit123@gmail.com>
Commit cf8e87545f implemented a way to cancel pin setup,
and this also has effects on an exit from the app after pressing
the android exit. The change button started with clearing the
email and passwd in order to get the credentail page(s) active
again. While this worked ok, it confuses users that exit the
app from the credential pages, resulting in the need to enter
the credentials again after a restart. It appears that clearing
the credential state is sufficient to get the pages active.
Notice that the android exit is still not working (it seems
a no-op), but the interaction with the buttons in the app
preserves the email/passwd.
Reported-by: Willem Ferguson <willemferguson@zoology.up.ac.za>
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
The old credential status should only be set when changing the
actual credential status using the setCredentialStatus function.
Setting it here is just wrong. It sets the old status to the
current, and than adds the current to the prefences,
obviously, resulting in old = current, which can not be right
and results in a wrong flow of control in the credential state
processing.
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
This commit tries to implement most of issue #515. It reworks the
one credential page, which its dynamic PIN part, into two pages.
Main driver of selecting one of the two pages is the showPin
boolean. Page 1 contains the email/passwd field (and the
option to use a no cloud setup). Page 2 only contains the PIN
part (and the option to cancel the process).
The Kirigami central button does not seem very handy here. We
need, for example, a cancel, sign-in and register, only register,
etc. buttons, which are not easy to handle in specific icons.
Therefore, normal pushbuttons are chosen to deal with user
interaction, and the Kirigami button is removed from these
pages.
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
This commit is just a precaution. It makes sure that the old
(aka previous) credential status is correctly set on all changes
of this status.
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
This is a very subtile bug. Testing/developing on the desktop for mobile,
with a normal logbook in a git repo, resulted in a surprising effect.
When swichting from a cloud account to a NOCLOUD situation, the no cloud
repo was (not always) reset to the NOCLOUD_LOCALSTORAGE, but to the
normal logbook. Resuling in commits in the wrong repo.
This can easily be solved by setting the filename to NOCLOUD_LOCALSTORAGE,
when switching to NOCLOUD mode.
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
In case the credential state is NOCLOUD, the saving of credentials
in the preferences was suppressed in case of invalid data in the
email/passwd fields.
There is no reason to check these fields for correct input, as they
are not used in case of NOCLOUD mode. A simple if statement is added.
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
As written in 8d9ad3cfea7e4c0875, the user needs to be able
to exit the PIN entry UI, in case no PIN can be received due
to a wrong email address.
The simplest way seems to just clear the cloud credential data,
and let the user try again. Obviously, we could argue if the
exact previous state of the 1st credentials screen could
be restored, but as it is only 2 simple fields, of which
it is higly likely that the email adress is misspelled (and
the password hidden), it seems overly complex to implement.
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
Makes sure that the CS_NEED_TO_VERIFY status is carried forward
and sets the showPin flag that triggers the 2 different states
in de QML UI.
The new credential UI will have one page for username/passwd,
used for setting up an account or switchting to a different
account, and a second page to enter a PIN only.
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
In case the credential status is UNKNOWN, and the cloud username
and password are empty, do not go automatically to NO_CLOUD status.
This is (again) preparation for future work on credential management.
For example, the user entered an email address with a spelling
error, so does not receive a PIN code email. This user needs a
way out, so there needs to be a <cancel> button on the PIN code
screen. And the most logic was of cancelling is emptying the
entered username/passwd and let the user try again.
Without this change, the user immediately gets into the (somewhat)
final NO_CLOUD state, which will result in (very) confused users.
With this change, there is exacly one way (left) to get into a
NO_CLOUD setup: hitting the proper button, so a deliberate user
action.
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
It appears that the onCompleted of the StartPage item is triggered
before the onCompleted of the rootItem. This is logical as the
Startpage is a child of the rootItem. And, yes, this does matter.
As the divelist also contains the logic for initial cloud
registration (and is the default page shown in a state where
the cloud credentials are valid (CS_VERIFIED state)), we need to
know the correct credential state at start of the app.
The move of this one line of code makes sure of that, in addition
to setting the credential state from the preferences. Now, the
setupActions function can reference correct credential data.
This is further preparation for a better cloud creation
process from mobile.
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
Having two different enums around with more or less the same
definition has lead to unclear code. After removing two not needed
states on the mobile end, the remaining step to one enum for the
credential state becomes almost is simple rename operation.
Unfortunately, I do not know a way to embed a plain C enum
from pref.h into the QMLManager object. So after this, there
are still 2 enums around, but now identical.
This commit is not changing any functionality.
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
And here, the removal of a second superfluous state from QMLManager.
This is true no-brainer. While this state was set once troughout the
entire mobile code, it was never tested for this state. Testing shows
that it is safe to change to the UNKNOWN state.
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
This is a no-brainer removal of the VALID_EMAIL state used in QMLManager.
All current usage of this state is "if state is VALID or VALID_EMAIL",
so there is no distinction between the two states.
It is even a little different. The comment suggests "when we can open
a local cloud storage, tied to a cloud account (so explicitly not
the no-cloud status), we have at least a valid email". While this
is formally true, this implies that there is also a cloud account
on the cloud server (ie. the cloud account is in a VERIFIED state).
In other words: currently, there can't exist a valid local storage
that is tied to a valid email adress, without valid cloud account
on the server.
Notice that this touches the discussion on GitHub for commit
e76f527fe5 (pull request #520). Can we implement the creation
of a valid cloud account without data link to the cloud server?
Currently, we need the server to confirm the email address (for
example for uniqueness reasons on server side). Obviously, we could
hack our way out of this, but we have a perfect solution already
in place. Create a no-cloud account, and transfer that later to
a true and valid cloud account.
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
When we want to go the NOCLOUD credential state, do not go the
incorrect INCOMPLETE state.
This is the first in a series of unraveling the preference parameter
prefs.cloud_verification_status and the data that is carried around
in the QMLManager::credentialStatus_t.
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
In commit e76f527fe5, the scenario of switching between 2 already
VERIFIED cloud accounts was identified, which was working poorly. It
needed a restart of the app to get the new account visible.
Reason for this, was the setting of the credentialStatus to the value
of an undefined (never set) old credentialStatus. This commit makes
sure we have a defined credentialStatus, just before changing it to
the new one.
A really mini step forward, as the behavior is still not perfect. Now,
the user has to select the dive list manually, after entering
credentials of the new clould account.
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
The PIN (and cloud account creation) is not limited to the
desktop (any more). Correct the string accordingly.
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
Independ of the settings, the threshold to reset the GPS data was
hard coded to 5 minutes. Now, honour the entered (and updated during
a session) time to refresh the GPS data in the location service.
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
Move the location services switch to the GPS menu, indicate both with icon and text if the service is active or not.
Signed-off-by: Joakim Bygdell <j.bygdell@gmail.com>
Add a checkbox to the preferences page to facilitate selective visibility of the developer menu. With the coresponding function in qmlmanager.
Signed-off-by: Joakim Bygdell <j.bygdell@gmail.com>
This commit adds the capability to cancel a running download from DC.
The actual cancel is fully handled in the underlying libdivecomputer
code. As the user may be interested in the dives downloaded up to
the moment of cancel, do not just close the download screen (as
it was before this commit). Now, the <quit> button changes to
<cancel> when the download is started, and pressing cancel, only
cancels the download and does not close the download screen, but
presents the so far downloaded data. When no download is running,
the <quit> button just quits the screen as before.
Fixes: #485
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
Trucate the logfile on open to remove any data from previous sessions.
This confused me very much, as the new logfile started from the
top, leaving old data on the bottom.
Also added system date for easy reference.
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
Email addresses are checked for a someone useful pattern, passwords are
letters, numbers, and +-_. only. Reject anything else.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
The trick is to pick a path that is accessible from other applications.
In theory QStandardPaths::GenericDataLocation should provide that.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
To be shared between C/C++ and QML code in order to show the updates
and potential error messages from libdivecomputer.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
This code is based on code from Marco Martin from the Kirigami Android
sample app. In order to simplify the QML code the QMLManager function is
there for all OSs, but it's a no-op on anything but Android.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Translate all of them, but also remove some redundant or possibly
misleading messages. These are now seen by users, not just developers
trying to debug the code.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
We have unified the debug output to the console and to the App Log, let's
use it consistently everywhere.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
The old system of cloud access updates with fake percentages just wasn't
helpful. Even worse, it hid a lot important information from the user.
This should be more useful (but it will require that we localize the
messages sent from the git progress notifications and make them more
'user ready').
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
One Andorid JNI include was missing. And removed the unused ones
from QMLManager.
[Dirk Hohndel: I had added the missing include elsewhere, so this commit
now only has the removal of the unused includes]
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
For reasons unknown to me, the DCDeviceData instance was freed way too early,
and used afterwards, obviously resulting in a SIGSEGV. This commit creates
the DCDeviceData as a direct child of the QMLManager instance, ensuring
it does not get freed prematurely.
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
This shouldn't be part of the UI (qmlmanager), but part of our
overall handling of dive computers and BT devices.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
appendTextToLog prepends the elapsed time which can be useful, so let's
not remove that but instead have it log to qDebug() instead.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
This adds a list of paired BT devices to the QMLManager class. In addition,
a very simple implementation is made of getting the paired BT devices on
Linux, so that we can test further processing of selecting the proper
devices, in a mobile-on-desktop situation.
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
It is very impolite to force BT on at start of the mobile app. We cannot
know if the user is going to import dives over BT.
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
As Qt is not able to pull the pairing data from a device, a lengthy
discovery process is needed to see what devices are paired. On
https://forum.qt.io/topic/46075/solved-bluetooth-list-paired-devices
user s.frings74 does, however, present a solution to this using JNI.
Currently, this code is taken "as is".
Currently, only for Android (so not mobile-on-desktop, or even desktop).
And only generating logging data in the logcat.
Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>
The naming scheme of OSTC dive computers doesn't match their product names,
but they all behave the same from a download perspective, so we assume that
any BT device that has a name starting with OSTC is an OSTC 3.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
We remember the offered service uuids as we detect the device and then
try the first one - likely this needs to be fixed / tuned to pick the
right one if multiple uuids are offered.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
If we find something that looks like a known BT dive computer, set
things up so that we can use it later. If multiple dive computers are
found, simply use the first.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
So far all this does is list all the BT devices that it finds
(and I worry if this will have negative battery implications
on a mobile device), but this should allow us to connect to
a standard BT dive computer (but that will of course require
more code to pick the right device).
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Once we re-inserted the dive in the list we aren't done, quite the
contrary - we now need to make sure that we handle any other changes
and mark the dive list as updated.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
We need to set the timeSpec after the QDateTime was parsed,
otherwise it gets converted to localtime again.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
We already have that for the other three fields where we offer auto
completion (buddy, divemaster, suit).
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
In order to make sure we don't render the initial profiles with the
wrong scale on devices, we need to seed the device pixel ratio with the
device default and then update it once the window has been created.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
For CCR dives we want to display the setpoint and pO2 information,
due to the limited screensize we have to remove the temperature graph or
the view will be to cluttered.
Signed-off-by: Joakim Bygdell <j.bygdell@gmail.com>
We were doing the right thing switching from metric to imperial, but in
order to swtich back you had to restart. Now it works both ways.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
There is no need to have two variables for the same purpose.
[Dirk Hohndel: changed to keep the two separate functions as otherwise
we no longer parse existing repos successfully]
Signed-off-by: Joakim Bygdell <j.bygdell@gmail.com>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
This allows the user to enter multiple buddies as a comma separated list,
the "Multiple Buddies" entry is still a special case as we can only populate
the combobox with a single name for each entry.
fixes#168
Signed-off-by: Joakim Bygdell <j.bygdell@gmail.com>
When editing adive in Subsurface-mobile we can only handle one buddy
due to the limitations of the combobox. To prevent loss of data when editing
a dive with more than one buddy we display "Multiple Buddies" in the buddy
field. This creates a special case where no changes are written to the buddy field
unless the user changes buddy for that dive.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
This adds the option to select a cylinder when adding or editing a dive.
Due to limited screen size we restrict the editing to the first cylinder only.
Signed-off-by: Joakim Bygdell <j.bygdell@gmail.com>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Subsurface uses "local time" which in particular means we never
display time zone information to the user. The user (and our file
format) only sees times like 5pm or 17:00. A better name than
local time (which could mean "local at the dive spot) would
be "watch time", the time displayed by the diver's watch when
she entered the water.
Internally, we store times as time_t, seconds since Jan 1 1970 0:00
UTC. Our convention for conversion between 5pm and time_t as always
been to treat 5pm as if it were UTC.
Then confusion arose since Qt's QDateTime (which is tied to UI elements
like QTimeEdit and similar) is time zone aware and by default assumes
the system time zone. So when we set a QDateTime to 5pm and then later
convert it to time_t we have to take care about the difference between
UTC and the system time zone.
This patch unifies our solution to this problem: With it, we set all
QDateTime's time zone to UTC. This means we don't have to correct for
a time zone anymore when converting to time_t (note, however, the
signedness issue: Qt's idea of time_t is broken since it assumes it
to be unsigned thus not allowing for dates before 1970. Better use the
millisecont variants).
We only need to be careful about time zones when using the current time.
With this convention, when assigning the current time to a QDateTime, we
need to shift for the time zone since its value in UTC should actually be
the watch time of the user who is most likely used to the system time zone.
Signed-off-by: Robert C. Helling <helling@atdotde.de>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
If we run the backend to verify credentials without waiting for it to
finish, the redirect might happen before we know if the credentials are
invalid, unverified or verified - which will cause us to give the wrong
information to the user.
Yes, this additional wait is annoying, but I can't come up with a better
way to do this and avoid incorrect information. At least the UI isn't hung
while we wait.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
In order to get autocomplete to work on manual dive add
we apparently need a separate init function.
Signed-off-by: Joakim Bygdell <j.bygdell@gmail.com>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
With offline the default now, we need to force a connection at least once
so that the repos are in sync. And then of course we need to return to the
correct state, regardless on whether this connection succeeded or failed.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Since saving checks if there are changes, we have to always mark the dive
list as changed before asking for the changes to be saved.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
This is rather simplistic, it just imports the local data into the remote
repository and therefore loses the git history of the local data - but I
wasn't able to make the git merge without shared base commit work, so I
went this much easier to implement route instead.
One thing we need to be careful about is that contacting the remote server
could fail. If we don't manage to merge the dives from cloud server and
local storage, we need to revery to no cloud status in order not to lose
the local data.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
This makes sure that if the user switches to offline mode they don't get
incorrect notifications that tell them that they are accessing cloud
storage.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
This exposes a locationServiceAvailable property to QML and keeps it in
sync with the corresponding state in the GpsLocation widget.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
I couldn't figure out how to break this down into small, useful commits.
Part of the problem is that I kept going while working on this and as you
can see from looking at the commit, diff tries so hard to find small code
fragments that moved around, that the diff overall becomes quite
unreadable and it seemed impossible to recreate the sequence of steps
after the fact.
It all started with adding the parsing for the GPS coordinates. But while
testing that code I found several issues with the rest of the function.
Most importantly it seemed ridiculous that we carefully tried to match the
texts that the DiveObjectHelper would create for the various fields,
instead of just using the DiveObjectHelper to do just that. And once I had
converted that I once again realized just how long and hard to understand
that function was getting and decided to break out some of the more
complex parts into their own helper functions.
But of course all this didn't happen in this logical, structured, ordered
way. Instead I did all of these things at the same time, testing,
rearranging, etc.
So in the end I went with one BIG commit that does all of this in one fell
swoop.
This adds four helper functions to deal with start time/date, duration,
location and gps coordinates, and depth of the dive.
To avoid mistakes when dealing with the GPS coordinates, there's another
helper to encapsulate the creation of the dive site and we switched to a
current GPS location.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
There were several logical flaws here. Ugh.
Don't save things if there are no unsaved changes, if we haven't
initialized this repository from the cloud or if we are already saving
things.
Then, once we decide that we should save, first always save to the local
cache and then check if we should save to the cloud and do so if
requested.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Let's not discard our GPS location when editing site name on mobile.
Fixes#1051
Signed-off-by: Miika Turkia <miika.turkia@gmail.com>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
The manager can now directly update the index of the selected dive, and
the UI tells the manager the timestamp of the currently selected dive.
This allows the manager to pick the best possible dive as selected dive
if things change (for example if the dive list gets reloaded because it
changed in cloud storage).
Fixes#1009
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
If the automatic sync is turned off we could be stuck in a state where we
always thought that we were already in the middle of a save.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Before it had the next dive still selected.
Fixes#1053
Signed-off-by: Robert C. Helling <helling@atdotde.de>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
On Android we can save locally right away, but we don't want to make the user
wait for a network sync. Sadly, on Android currently the saving in the
background doesn't work and the save will run when the user comes back.
Definitely not ideal.
On iOS the situation is different - a save to the local git cache takes
surprisingly long. Must be the shitty file system they use or something.
Because of that we only mark the dive list changed and instead save the next
time the app is not in the foreground (which works on iOS but not on Android -
go figure).
On all the other OSs (I guess that would be desktop builds of
Subsurface-mobile? But there may be other mobile OSs that people might want to
build it on) we save both locally and to the cloud right away.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
So when the user taps on the manual cloud sync, we always force access to
the cloud server. Otherwise we only access the cloud server if
git_local_only isn't set.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
This makes the code much cleaner and easier to understand and should allow
us to then switch back to doing at least the local save right after we make
any changes to the data.
This commit also tries to make sure that the accessingCloud status stays
correct and consistent throughout all the various success and error paths.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
If the user is on the credentials page, doesn't change the credentials
but simply taps on save, they now get back to the dive list.
Fixes#1047
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
We first check the sha to see if we want to load at all. But at that
point we already have the repository and the branch and we have synced
with the remote. So when we decide that we need to reload from storage,
we don't need to repeat those steps, instead we can go directly to the
git load.
For that to work we need to pass the repository pointer and the branch
name back to the caller so that we can directly call git_load_dives().
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
We first want to save any exiting unsaved changes to the local repository
(and ONLY to the local repository). After that we want to make sure that
we are syncing remotely, fetch the remote and then (possibly after a
merge) push the changes to the remote. In the end we reset the previous
"local git only" preference which we overwrote for this manual forced
sync.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
We had a redundant check of the server reply hidden in there, and the
logic which values were set where didn't really make much sense. This
seems clearer to me.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Most of the time we are adding all the dives, so do this in a single model
operation. This makes the case when adding a single dive (in the undo delete
function) slightly more complicated, but that seems totally worth it for the
speedup in the common case.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Having subsurface-core as a directory name really messes with
autocomplete and is obviously redundant. Simmilarly, qt-mobile caused an
autocomplete conflict and also was inconsistent with the desktop-widget
name for the directory containing the "other" UI.
And while cleaning up the resulting change in the path name for include
files, I decided to clean up those even more to make them consistent
overall.
This could have been handled in more commits, but since this requires a
make clean before the build, it seemed more sensible to do it all in one.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>