Make search panel floating so that web view does not resize.
This commit also addresses following issues:
- search now wraps around document
- application icon and window title are now back
Signed-off-by: Sergey Starosek <sergey.starosek@gmail.com>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
The shortcut may be used elsewhere, e.g. in user manual
Signed-off-by: Sergey Starosek <sergey.starosek@gmail.com>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
We now never remove units or percentage signs, and always just compare
the string data, so we should remove the hacky arguments that are no
longer used.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
This matches the pattern of unit conversion, and will allow us to remove
all the code that uses the old complex "CHANGED()" macro that tries to
remove units or percent signs.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Whittling down on the string parsing that doesn't check user-specified
units. Still need to handle temperatures (and will do percentages to
match the pattern too), but this is getting us closer to always honoring
user-specified units.
With this you can say that you have a "10l" cylinder at "3000psi", and
it will do the right thing (it's basically a 72 cuft cylinder in
imperial measurements).
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
We now never remove units or percentage signs, and always just compare
the string data, so we should remove the hacky arguments that are no
longer used.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
This matches the pattern of unit conversion, and will allow us to remove
all the code that uses the old complex "CHANGED()" macro that tries to
remove units or percent signs.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Whittling down on the string parsing that doesn't check user-specified
units. Still need to handle temperatures (and will do percentages to
match the pattern too), but this is getting us closer to always honoring
user-specified units.
With this you can say that you have a "10l" cylinder at "3000psi", and
it will do the right thing (it's basically a 72 cuft cylinder in
imperial measurements).
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
This just adds (and uses) a string_to_pressure() to parse pressure units
correctly when filling in cylinder pressures.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
If no average depth is known the current fake profile behavior is rather
unintuitive (we make up an average depth).
Instead we should assume that this is a PADI style dive log and give the
user a "rectangular" profile (actually, it's a trapecoid as we at least
try to enforce a sane ascent / descent speed). If the dive is somewhat
longer or deeper (10 min / 10 m) we even add a 3m safety stop at 5m.
Added a new dives/test0b that tries to capture the typical cases to test
this.
Fixes#398
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
It's currently only used for the setting of the cylinder switching
depth, but now that one should work with user-specified units (so you
can set a max depth in feet even if you use metric, and vice versa).
In the future, if we also make the unit preferences something you can
pass in (with user preferences as a default argument value), we might
want to use this for parsing the XML too, so that we'd honor explicit
units in the XML strings. But the XML input unit preferences are not
necessarily at all the same as the user preferences, so that does
require us to extend the conversion functions to do possibly explicit
unit preference selection.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
The C library doesn't use const char pointers for legacy reasons (and
because you *can* modify the string the end pointer points to), but
let's do it in our internal implementation just because it's a nice
guarantee to have.
We actually used to have a non-const end pointer and replace a decimal
comma with a decimal dot, but that was because we didn't have the fancy
"allow commas" flags. So by using our own strtod_flags() function, we
can now keep all the strings we parse read-only rather than modify them
as we parse them.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Make it use 'weight_t' and hide the "grams" part inside the type. That
was the whole point of the weight_t type, after all. Returning a
"double" was always bogus, since we internally always do integer grams
(and the function actually used "rint()" to get all the rounding right
anyway).
As a result, it's now called "string_to_weight()".
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
For some devices, the event data contains important data that is
required for parsing the dives, but which is not present in the full
memory dump.
Signed-off-by: Jef Driesen <jefdriesen@telenet.be>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
If the 'logfile_name' and 'dumpfile_name' were NULL we can simply
strdup() them with a new value, but if there was a previous value
we need to free() first.
C99 6.7.8 allows us to keep said variables without the
explicit NULL initialiazation.
Signed-off-by: Lubomir I. Ivanov <neolit123@gmail.com>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
C99 7.1.4, says nothing about passing NULL to fopen(),
which means that it isn't portable and there are no guaranties
that the return will be a NULL pointer or that that a library
implementation will not assert or SYSSEGV in the middle of the
fopen() branch.
libdivecomputer.c's 'dumpfile_name' and 'logfile_name' could
cause problems in that regard.
A possible fix for #411
Signed-off-by: Lubomir I. Ivanov <neolit123@gmail.com>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
On Win32 subsurface_fopen() can reach an assert in windows.c:
utf8_to_utf16(), if NULL is passed for 'path'.
Let's return NULL/-1 for some of the *open() functions in there.
Signed-off-by: Lubomir I. Ivanov <neolit123@gmail.com>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Restoring the selection was not setting the selected dive as current, and
thus, breaking keyboard navigation.
Fixes#402
Signed-off-by: Tomaz Canabrava <tcanabrava@kde.org>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Just one of the completers had Qt::CaseInsentitive set,
setting for all of them.
Fixes#400
Signed-off-by: Tomaz Canabrava <tcanabrava@kde.org>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
In commit 528d0ea0e7 ("Print numerical value of mean depth") Miika once
again forgot the three non-metric countries on this planet... :-)
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
This will print the numerical value of mean depth to the profile graph.
Fixes#405
Signed-off-by: Miika Turkia <miika.turkia@gmail.com>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
In commit 3fd39a7a87 ("Remove some constants and use helpers instead")
Anton missed the fact that the Uemis gives temperatures in the handy unit
of "centi degree C". Now things work again.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
We have the wonderful Qt string functions. Let's use them to make the code
simpler and easier to read.
Suggested-by: Lubomir I. Ivanov <neolit123@gmail.com>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
When a dive contains no cylinders, clicking the '+'
button could SIGSEGV if current_dive->dc.model is NULL.
Signed-off-by: Lubomir I. Ivanov <neolit123@gmail.com>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
This just adds (and uses) a string_to_pressure() to parse pressure units
correctly when filling in cylinder pressures.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Interpolated pressure should always be set, even it we didn't
calculate a new one. In this case we should just use the last one.
Signed-off-by:Rodrigo Severo <rodrigo@fabricadeideias.com>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
If no average depth is known the current fake profile behavior is rather
unintuitive (we make up an average depth).
Instead we should assume that this is a PADI style dive log and give the
user a "rectangular" profile (actually, it's a trapecoid as we at least
try to enforce a sane ascent / descent speed). If the dive is somewhat
longer or deeper (10 min / 10 m) we even add a 3m safety stop at 5m.
Added a new dives/test0b that tries to capture the typical cases to test
this.
Fixes#398
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
When an interpolated segment is on the surface we get no pressure time -
and try to divide by zero.
(plus a small whitespace issue that slipped throught the cracks earlier)
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
The small straight parts at the end of tank pressure lines are more of a
aesthetic issue, not causing real harm so it is no reason to remove the
SURFACE_THRESHOLD test from pressure_time function only because of this.
Also improved interpolate data debuging, rearranged
get_pr_interpolate_data and removed an unused variable from
get_pr_interpolate_data. No real change here, just trying to make the code
clearer.
[Dirk Hohndel: clean up whitespace damage from this and the previous
commit]
Signed-off-by: Rodrigo Severo <rodrigo@fabricadeideias.com>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Interpolated lines for tank pressures were presenting discontinuities
(sudden drops) and stalls (horizontal lines) with certain dive profiles.
The main reason seems to be that the discrete interpolation of tank
pressure was adding small pressure increments that could be rounded down
or up repeatedly generating cumulative rounding errors that would mean
either a delay on pressure drop that would be drawn as a sudden drop or as
a premature pressure drop that would result in a flat line.
This patch changes the way the discrete interpolation is done, so that we
don't have cumulative rounding errors distorting tank pressure lines.
To calculate accumulated pressure_time values the get_pr_interpolate_data
function was created. The fact that get_pr_interpolate_data transverses
the beginning of the plot_info entry list for each entry that needs
interpolated tank pressure isn't optimal at all. There might be a way to
properly track the data necessary to interpolate tank pressures from
inside the main pi->entry loop in fill_missing_tank_pressures.
Unfortunately I didn't manage to do it inside fill_missing_tank_pressures
so we have get_pr_interpolate_data.
The SURFACE_THRESHOLD test from pressure_time function was also removed as
no matter how shallow the diver is, if he is using the cylinder to breathe
the cylinders tank pressure should be affected.
Signed-off-by: Rodrigo Severo <rodrigo@fabricadeideias.com>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
It's currently only used for the setting of the cylinder switching
depth, but now that one should work with user-specified units (so you
can set a max depth in feet even if you use metric, and vice versa).
In the future, if we also make the unit preferences something you can
pass in (with user preferences as a default argument value), we might
want to use this for parsing the XML too, so that we'd honor explicit
units in the XML strings. But the XML input unit preferences are not
necessarily at all the same as the user preferences, so that does
require us to extend the conversion functions to do possibly explicit
unit preference selection.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
The C library doesn't use const char pointers for legacy reasons (and
because you *can* modify the string the end pointer points to), but
let's do it in our internal implementation just because it's a nice
guarantee to have.
We actually used to have a non-const end pointer and replace a decimal
comma with a decimal dot, but that was because we didn't have the fancy
"allow commas" flags. So by using our own strtod_flags() function, we
can now keep all the strings we parse read-only rather than modify them
as we parse them.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Make it use 'weight_t' and hide the "grams" part inside the type. That
was the whole point of the weight_t type, after all. Returning a
"double" was always bogus, since we internally always do integer grams
(and the function actually used "rint()" to get all the rounding right
anyway).
As a result, it's now called "string_to_weight()".
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
We do not need any import dialog when importing normal XML based
divelogs. With this they are imported directly after file selection
dialog. However, when CSV files are detected in the import list, the
configuration dialog is displayed and applied for them. (CSV files are
detected by file extension.)
Signed-off-by: Miika Turkia <miika.turkia@gmail.com>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
For some devices, the event data contains important data that is
required for parsing the dives, but which is not present in the full
memory dump.
Signed-off-by: Jef Driesen <jefdriesen@telenet.be>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
If the 'logfile_name' and 'dumpfile_name' were NULL we can simply
strdup() them with a new value, but if there was a previous value
we need to free() first.
C99 6.7.8 allows us to keep said variables without the
explicit NULL initialiazation.
Signed-off-by: Lubomir I. Ivanov <neolit123@gmail.com>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
C99 7.1.4, says nothing about passing NULL to fopen(),
which means that it isn't portable and there are no guaranties
that the return will be a NULL pointer or that that a library
implementation will not assert or SYSSEGV in the middle of the
fopen() branch.
libdivecomputer.c's 'dumpfile_name' and 'logfile_name' could
cause problems in that regard.
A possible fix for #411
Signed-off-by: Lubomir I. Ivanov <neolit123@gmail.com>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
On Win32 subsurface_fopen() can reach an assert in windows.c:
utf8_to_utf16(), if NULL is passed for 'path'.
Let's return NULL/-1 for some of the *open() functions in there.
Signed-off-by: Lubomir I. Ivanov <neolit123@gmail.com>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
In commit c3fe1a9e9f ("Get rid of pointers to dive structures in the
UI") I was a bit too aggressive moving away from pointers to dives.
This is only needed for pointers that are held across operations that
could change the dive_table. I figured that it wouldn't hurt to get rid of
some more pointers as well, but it turns out I was wrong. The current dive
that we store in the Cylinder and Weight models can be a dive that isn't
in the dive_table at all: the multiEditEquipmentPlaceholder. And when
using the diveId we end up finding the original dive in the dive_table and
therefore modify the wrong structure.
This undoes two thirds of the above mentioned commit.
Reported-and-analyzed-by: Patrick Valsecchi <patrick@thus.ch>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Turns out this function can be called when there is no valid dive - so
only protect those parts that dereference the dive.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
The assumption that the pointer will keep pointing to a valid structure is
fundamentally flawed. And even if that is true today, it might change in
the future - just don't do it. Use the diveId instead.
The exception is when you own the structure and use it within one UI
interaction during which any way to change the dive_table is disabled
(e.g., while adding / editing a dive).
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
We try to do the most reasonable thing. If you have an existing dive and
merge a just downloaded dive with it - the you want the existing id.
If you merge two consecutive dives, then the start of the merged dive is
the earlier of those two dives so we keep that id.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
This id is just held in memory. It's not supposed to be used for anything
but having a unique handle that represents a dive. Whenever you need to
remember a dive across an operation that might change the dive_table, this
is what you should hold on to, not a dive number, a dive pointer, or
anything like that.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Restoring the selection was not setting the selected dive as current, and
thus, breaking keyboard navigation.
Fixes#402
Signed-off-by: Tomaz Canabrava <tcanabrava@kde.org>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
