Dirk Hohndel [Wed, 5 Oct 2011 18:36:15 +0000 (11:36 -0700)]
Add XML file import back and treat open and import differently
Open (or adding a file name on the command line) means that this is just
one of the files that you consider part of your dive history. So dives
don't get automagically numbered and the dive_list is not considered
"changed" just because another file was opened.
Import (or adding a file on the command line after --import) means that
you are importing the content of this file to your dive history. So if the
imported file has un-numbered dives that are newer than everything else,
those get correctly renumbered. And importing marks the dive_list as
changed.
Dirk Hohndel [Wed, 5 Oct 2011 18:36:15 +0000 (11:36 -0700)]
Add an '--import' command line option
This option indicates that all files that come AFTER it on the command
line are being added to our divelist. The dives in these files should
receive numbers (assuming they are un-numbered and are all newer then the
dives in the files before the --import option, and assuming those dives
are numbered).
This also marks the dive_list changed after the new dives are added.
Using this option gives us a reasonable user experience in the case where
a user has one file with all their dives and wants to add newer dives
after this (after extracting them from a dive computer - as in the case of
a uemis owner where there is no direct import from the dive computer,
yet). Something like
subsurface MyDives.xml --import NewDives.SDA
It also doesn't break Linus' vision where the user has many files on the
command line which don't imply a changed dive_list.
Dirk Hohndel [Wed, 5 Oct 2011 16:24:52 +0000 (09:24 -0700)]
Mark divelist changed when renumbering or adding dives
The behavior is not yet consistent when calling with multiple file names
on the command line (as we don't add number to the later ones in this
case), but at least it catches the case if you manually renumber the dives
or if you import new dives that get added at the end - which are the two
most typical cases.
Linus Torvalds [Wed, 5 Oct 2011 15:37:14 +0000 (08:37 -0700)]
For a manual renumber, default to the existing first dive number
If renumbering a list of dives, default the start number to the existing
first dive number. That way, if you do need to renumber (overlapping
import or whatever), but your at least had your really old dives already
numbered, we start off with a sane default.
Linus Torvalds [Wed, 5 Oct 2011 15:31:31 +0000 (08:31 -0700)]
Automatically renumber new dives when they are "obvious".
When importing (or reading xml from files) new dives, we now renumber
them based on preexisting dive data, *if* such re-numbering is obvious.
NOTE! In order to be "obvious", there can be no overlap between old and
new dives: all the new dives have to come at the end. That's what
happens with a normal libdivecomputer import, since we cut the import
short when we find a preexisting dive.
But if any of the new dives overlap the old dives in any way, or already
have been numbered separately, the automatic renumbering is not done,
and you need to do a manual renumber.
Linus Torvalds [Wed, 5 Oct 2011 15:06:48 +0000 (08:06 -0700)]
Move 'dive_list_update_dives()' call into 'report_dives()'
All the callers were always calling report_dives first, followed by
dive_list_update_dives(). And there really was no reason to have the
callers call two separate functions for the "I've added new dives" case.
So just call dive_list_update_dives() directly from report_dives(), and
remove it from the callers.
Dirk Hohndel [Wed, 5 Oct 2011 03:33:17 +0000 (20:33 -0700)]
Make the dive merging code more tolerant
Depending on the tool used to import a dive from the uemis Zurich we end
up with different time stamps for the dive - just by a few seconds, but
the existing code insisted on an exact match.
We now allow for up to 60 seconds in difference and still consider two
dives as the same.
Dirk Hohndel [Tue, 4 Oct 2011 19:27:55 +0000 (12:27 -0700)]
Replace event text with small red triangle and tooltip
We draw a little red triangle (of hardcoded size - not sure if this SHOULD
scale with the size of the plot... I like it better if it doesn't) to the
left of an event.
We then maintain an array of rectangles that each circumscribe one of
those event triangles and if the mouse pointer enters one of these
rectangles then we display (after a short delay) a tooltip with the event
text.
Manually creating these rectangles, maintaining the coordinate offset,
checking if we are inside one of these rectangles and then showing a
tooltip... this all seems like there should be gtk functions to do this by
default... but if there are then I failed to find them. So instead I
manually implemented the necessary logic.
Dirk Hohndel [Tue, 4 Oct 2011 19:14:26 +0000 (12:14 -0700)]
Change plot routine to take a drawing_area as argument
Previously we passed in width and height and the routine itself decided to
keep 5% margin around each edge - oddly doing this with double precision,
even though this is all integer coordinates.
Instead we are now passing in a drawing_area. We are kind of abusing the
cairo_rectangle_int_t data type here - but it seemed silly to redefine a
new data type for this.
Width and height give the size of the TOTAL drawing area (as before).
x and y give the offset from the edges - so the EFFECTIVE drawing area is
width-2x and height-2y
This is in preparation for adding tooltips - those need to know the
coordinate offsets from the edges - so having this hard coded inside the
plot function didn't make sense anymore.
Linus Torvalds [Mon, 3 Oct 2011 20:19:23 +0000 (13:19 -0700)]
Merge branch 'uemis-integration' of git://github.com/dirkhh/subsurface
* 'uemis-integration' of git://github.com/dirkhh/subsurface:
Much nicer implementation of uemis sample parsing - and add events, too
Add working pressure to uemis tank data
Dirk Hohndel [Mon, 3 Oct 2011 19:27:14 +0000 (12:27 -0700)]
Much nicer implementation of uemis sample parsing - and add events, too
This is something I wanted to do for a while. Every uemis sample is simply
a packed structure with no padding. Instead of grabbing random bytes from
the middle of an unstructured data blob let's just define the structure
and access its members.
And while we do that, add support for the more useful uemis events as
well.
A couple of the warnings are disabled by default (compile time flag) as
they are just crazy - any normal dive will give you dozens and dozens of
speed warnings. Same goes for the PO2 green warning (I haven't looked but
this seems to trigger on a PO2 over 1.0 or something). Completely useless
and just hides actually useful info.
I still want to redo the way we visualize events in general - just
printing the text ontop of the profile really is suboptimal. Especially as
the uemis really seems to love to repeat several of the warnings quite
frequently.
Linus Torvalds [Mon, 3 Oct 2011 19:13:54 +0000 (12:13 -0700)]
Merge branch 'uemis-integration' of git://github.com/dirkhh/subsurface
* 'uemis-integration' of git://github.com/dirkhh/subsurface:
Remove the ability to 'Import' .SDA files
Integrate loading of uemis SDA files into the regular xml parsing
First steps towards integrating SDA files into the default XML loading
Dirk Hohndel [Mon, 3 Oct 2011 16:55:17 +0000 (09:55 -0700)]
Add working pressure to uemis tank data
Turns out they use 202.6bar as default working pressure. WTF?
Also I had misunderstood the way I should record the pressure internally
(which happened to work since I didn't set the working pressure). This is
now fixed as well.
Dirk Hohndel [Mon, 3 Oct 2011 16:55:17 +0000 (09:55 -0700)]
Add working pressure to uemis tank data
Turns out they use 202.6bar as default working pressure. WTF?
Also I had misunderstood the way I should record the pressure internally
(which happened to work since I didn't set the working pressure). This is
now fixed as well.
Dirk Hohndel [Mon, 3 Oct 2011 15:27:36 +0000 (08:27 -0700)]
Remove the ability to 'Import' .SDA files
We can instead 'Open' these files as they are just bastardized XML files.
This gets us back to a more consistent point where 'Import' gets data
directly from the dive computer (and hopefully soon we will add the
ability to load a dive directly from a uemis SDA to libdivecomputer),
and 'Open' loads a file from the filesystem of the computer we are
running on (this last sentence phrased so awkwardly as the uemis Zurich
SDA is a computer and presents a file system when connected via USB - it
just doesn't have the dive data in an accessible format in that file
system).
As a bonus we get to throw away quite a bit of code (the uemis specific
file handling, mini-XML parser with helper functions, the file open dialog
in the importer). Yay!
Dirk Hohndel [Mon, 3 Oct 2011 04:59:54 +0000 (21:59 -0700)]
Integrate loading of uemis SDA files into the regular xml parsing
There are a few interesting issues with this:
- this requires a change to the SDA file format; thankfully I control that
format, too (the default files are not valid XML files)
- once again, the fact that adding samples can change the dive pointer
messes with me - I decided to change the interface of ALL of the
XXX_dive_match functions to take a struct dive**
I know this is not ideal as all the other functions don't need that -
but I would have hated the inconsistency
- there is the issue that we now overload two _different_ uemis formats in
the same function - that's certainly a potential point of confusion
- a minor detail is the problem that the SDA format is kinda odd to parse
and that we trigger on the duration field by it being the only float.
Yeah, that's not ideal - but again, I control the format, so I _know_
this is true.
Linus Torvalds [Mon, 3 Oct 2011 02:14:00 +0000 (19:14 -0700)]
Remove cylinder index from cylinder list model
Instead of having to keep the index up-to-date as we edit entries
around, just figure out the entry index from the model itself. Gtk
seems to make it unnecessarily hard, but what else is new?
Linus Torvalds [Mon, 3 Oct 2011 00:16:50 +0000 (17:16 -0700)]
Change calling convention of 'edit_cylinder_info'
Instead of passing it the model and iterator (which requires that we
create the new entry for an 'add' event even if we then cancel the
operation), just make the caller do the final cylinder list update.
This way we can make 'add' work more sanely: if you cancel the add, we
now do not create an empty cylinder entry at the end.
Linus Torvalds [Sun, 2 Oct 2011 23:41:17 +0000 (16:41 -0700)]
First cut at working cylinder editing dialog
This currently only does the same old things we used to do (so still no
start/end pressure or trimix support), but despite that this is already
more flexible than the old model:
- we can now add new cylinders, rather than just edit the information of
the first two cylinders of the dive
- because the cylinder editing is being done in a edit dialog, it is
now much more reasonable to use multiple lines and expand all the
things we can edit.
But to actually make this fully fledged, we'll need to add all the other
info to the cylinder edit dialog, and probably add a confirmation dialog
for the "delete cylinder" case too.
Oh, and right now deleting a cylinder doesn't mark the dive info changed.
Linus Torvalds [Sun, 2 Oct 2011 20:42:51 +0000 (13:42 -0700)]
Add the ability to add new cylinders
This is totally useless since you cannot actually *edit* the resulting
new dive yet, but we'll get there. And this already conceptually shows
a capability that we didn't use to have with the old interface.
Linus Torvalds [Sun, 2 Oct 2011 20:13:27 +0000 (13:13 -0700)]
Start re-organizing the cylinder entry in equipment.c
This leaves the actual editing code unconnected, so now you can only see
the cylinder information, not actually edit it. However, with the big
re-organization I really do want to have this as a half-way point where
I have created the new cylinder tree-view.
I now need to connect the "add/edit" buttons to dialogs that then use
the editing widgets - so I've left that widget code around, because I'll
be able to reuse a lot of it. Not all, but the cylinder type model code
in particular will be re-used pretty much as-is.
Linus Torvalds [Sun, 2 Oct 2011 20:05:12 +0000 (13:05 -0700)]
Split up generic code to generate a gtk tree view column
We used to do this just for the dive list, but the new cylinder view
will want to do a lot of the same boilerplate gtk stuff, so make it a
bit more generic and move it to gtk-gui.c.
Linus Torvalds [Sat, 1 Oct 2011 04:55:51 +0000 (21:55 -0700)]
We forgot to pick up the 'value' field of a dive event
Just missed that one entirely in the xml parser for some reason.
Probably because the fields don't have much semantic meaning, so I
didn't even realize that I had missed one of the random integer values
in an event.
On my suunto, the 'value' field seems to contain things like the new
Oxygen percentage of a gas change event etc.
Distinguish internally between min pressure and end pressure
And don't artificially end dives on min pressure
This may be a problem for dive computers like Linus' Suunto Vyper Air
where the failure mode seems to be _high_ pressure readings (that's scary,
btw). If the transmitter fails at the end of the dive the pressure plot
ends with incorrect high pressure. But that's simply a bug with the dive
computer and not something that subsurface should hack around. Maybe we
should offer a way to edit the incorrect data points instead.
Always ending on the minimum pressure is definitely wrong as it causes
bogus plots when you do a valve shutdown during the dive (which means that
valid data gets plotted incorrectly).
We were missing the last sample (which is usually a fast ascent).
Also, reduced the velocity smoothing to 15 seconds as the 30 seconds were
hiding too much valid information
Correctly parse the braindamaged tank size information from uemis
Admittedly the cuft ratings are stupid, but still, it's not that hard.
In order to correctly describe a tank based on the cuft system you need to
know the cuft AND the working pressure. But the uemis Zurich always
assumes that the working pressure is 200bar. That's pretty close to
3000psi and therefore works "good enough" for Aluminum tanks - but in
general this will of course fail (e.g. for HP or LP tanks).
Yes Linus, gas pressure can indeed go up during a dive
At first glance it seems logical to make the ending pressure be the
lowest pressure observed during a dive. But if you do valve shut down
drills with a tech setup (where you have a fully redundant double
tank setup with two valves, two regulators and a manifold in between),
then you continue to breath from what is indeed the same "tank", but still
the valve on which your air pressure transmitter sits does get shut down
and de-pressurized. So your pressure goes down by quite a bit, and then
comes back up when the valve is turned back on.
And the ending pressure of the dive (which is used for things like the SAC
calculation) is indeed potentially higher than the lowest pressure
observed during a dive.
Allow larger tanks (change maximum from 200 to 300 cuft)
We don't handle doubles any different than single tanks - so while
200 cuft was a sane maximum size for a tank, once you dive with
doubles this logic fails.
We may or may not decide to implement special handling for doubles at some
point, but for now simply allow for tanks all the way up to double-150.
Merge branch 'ui' of git://github.com/dirkhh/subsurface
* 'ui' of git://github.com/dirkhh/subsurface:
The notebook pages can only be dropped back into the main notebook
Linus would like to be less on the bleeding edge of Gtk+
Use the correct signal to avoid Gtk-CRITICAL error message
Clean up the drag and drop code and allow ripping off the Dive Profile
Use the correct signal to avoid Gtk-CRITICAL error message
We used the wrong signal - "data-drag-received" is intended to check
whether the target will accept the drop. What we want is the "drag-drop"
signal which tells the widget that something was dropped on it.
Also fix an embarrassing lack of NULL pointer checks in my string
comparisons...
Clean up the drag and drop code and allow ripping off the Dive Profile
Linus had used some deprecated interfcase and didn't correctly untangle
the new window that he created (hiding it the window... very nifty).
I think I'm closer to the real solution with a data structure that keeps
track of the components of the new top level window that I need to be able
to untangle (and eventually, destroy) at the end.
The one error I also can't seem to get rid of is the
Clean up the drag and drop code and allow ripping of the Dive Profile
Add note on dive computers using the same import engine
Lots of dive computers are just variations on a theme, or sometimes even
just rebadged copies of each others with different manufacturer and
model names. The import dialog may not mention your exact dive computer
by name, but that doesn't necessarily mean that you cannot import data
from it.
Make that clearer in the README, and list the rough list of dive
computers supported by libdivecomputer.
Add drag-n-drop support to be able to re-integrate the dive list
This is somewhat hacky, and there is clearly something I still don't
understand about gtk selections and drag-n-drop. Dropping it back
works, but I get a nasty error when I do it:
even though I actually never set any selection at all directly. So
there must be some internal gtk rule that I am violating, but I can't
see what it is.
I probably shouldn't commit it with a known ugly wart like that, but I
really have no clue. Maybe somebody else can figure out what is up.
That also makes it always stay in front of the other window, which is
just annoying. I only did it because I wanted to make sure it dies when
the main window does, but since we just kill the main loop when closing
either window, that just isn't an issue.
Merge branch 'otu-tracking-v2' of git://github.com/dirkhh/subsurface
* 'otu-tracking-v2' of git://github.com/dirkhh/subsurface:
Store options in gconf
Add preference option to chose if SAC and/or OTU should be in divelist
Fix up trivial conflicts in gtk-gui.c (cleanup in gtk dialog wrt
gtk_dialog_get_content_area() having introduced a new 'vbox' widget)
While it's not the most elegant way to do this I opted to store the
options with "inverted polarity" - i.e., the options that are supposed to
default to "True" are stored inverted since gconf reports an unset option
(first time the user runs the program) as "False".
I've been wondering how to make 'subsurface' work better on a small
screen (I used to travel with a crappy netbook - I may have upgraded my
laptop since, but it is still a design goal of mine to make sure it all
works fine in that kind of environment).
And ever since the dive list was made much wider and moved below the
notebook, it's annoyed me how much room it all takes if I want to have
both a reasonable plot window and several dives visible at the same
time.
The solution seems to be to just make the dive list be a notebook page.
That makes the default layout very dense.
At the same time, when you have the pixels, it's horrible, because you
would want to see the dive list and move between dives while at the same
time also seeing the dive profile change. But that is solvable by
simply making the dive list notebook page be detachable, so if you have
a nice big screen, just detach the dive list page and now you have
independent windows for the dive list and the dive info.
NOTE! I don't have any way to re-attach the dang thing. I think I'd
need to learn about drag-and-drop targets etc. So once you've detached
the dive list, it stays detached.
Update Mares IconHd parsing to current libdivecomputer interface
The libdivecomputer interfaces are pure crap. There are no generic
"open the dive computer" or "create a parser for the dive computer"
interfaces, instead each dive computer you support has its own open and
parser generator interface.
And they change. Happily fairly seldom, but they change. And two days
ago, Jef changed the interface for the Mares Icon HD computer in order
to support the newer HD Net Ready variant.
I've asked Jef to make a sane interface for "open the dive computer" and
"just create the parser" for libdivecomputer, but he claims that he
cannot just track the device model details internally. Which is
obviously a completely bogus claim, since the way *we* track the model
details is to just feed it back from the silly event.
libdivecomputer should just do that internally and not bother us with
its crazy internal model numbers. But whatever.
In the meantime, work around this braindamage, and hope that
libdivecomputer comes to its senses some day.
Stop libdivecomputer import when we start seeing old dives
I don't know about other dive computers, but the Suunto Vyper Air is
slow as hell to import all the dives from. And libdivecomputer seems to
be importing dives "most recent first", so this just makes it stop
importing dives when it finds a dive that we've already seen.
Caveat: libdivecomputer has this fancy notion of "dive fingerprints",
and claims that's the way to do things. That seems to be overly
complicated, and not worth the bother.
If you worry about the import finishing early due to already having some
dives with the same date in your dive list, just import starting from an
empty state, and thus get a pure "dive computer only" state with no
early out. Then you can just load the old dives afterwards, and depend
on subsurface merging any duplicates.
But for normal operation, when you just want to import a couple of new
dives from your dive computer, the "exit import early when you see a
duplicate" is the right thing to do.
Add a GtkEntry to allow editing of the device name
Ok, so some file chooser widget with a popup dialog would have been more
professional, but I'm lazy. Plus I suspect the popup would look
horrible when populated with /dev entries, and I don't think there is
any sane filter function.
So this works, and means that you don't *have* to recompile the whole
program just because you have your dive computer on something else than
a USB serial line.
I suspect I should save the default name as a config variable too.
Maybe a setting in the preferences dialog.
It's really just about the logo, but whatever. Dirk tells me I need one
of these in order to call it 1.0. And I'm not going to fall into the
trap of thinking that 1.0 needs to be something polished, it just needs
to be working well enough..
This is *really* ugly. We really should just create some kind of widget
that when moused over will show the event. Or something. Rather than
putting text on top of other text: the events - when they happen - are
usually bunched together (PO2 warnings, max depth, fast ascent leading
to mandatory safety stop, you name it).
But at least this way we see that the data is there, even if we see it
in ugly ways.
Drop surface events when reading from an XML file too
Remember those useless surface events that we ignore when we import a
dive from a dive computer? Yeah, they exist in the libdivelog xml files
too. So ignore them when we see them there too.
As reported by Mauro Dreissig, the progress bar doesn't work and causes
a SIGSEGV due to a missing allocation. The code broke when Dirk
separated out the GUI from the core code, and I hadn't tried
divecomputer downloads since.
The calculation assumes that the cylinderindex in each sample tells us
which PO2 the dive was breathing at that time. This needs to be verified
with dives where there is an actual gas switch.
No idea where to display them, yet. Far fewer people will care about this
than care about SAC - does this still rate a spot in the dive_list?
I guess I could make it part of the dive_info - but it's not editable.
It doesn't seem to fit with the equipment page (even though this is the
one editable field that is related - nitrox %)
We should always strive to have a dive selected, so pick the first one
(that was how the dive list logic worked anyway, it just wasn't truly
selected at the tree-view level, so it wasn't *visibly* the selected
dive).
It used to be "index 0" which originally was the date string, but not
only has that changed (it's now just the dive index), it's kind of
pointless to search for a date string.
Show "m" or "ft" instead of "max/m" vs "max/ft". The column really
doesn't want to be that wide. The column header is already the widest
part of it even with this short name (due to the sort order arrow
thing).
Merge branch 'quit-handling' of git://github.com/dirkhh/subsurface
* 'quit-handling' of git://github.com/dirkhh/subsurface:
Use the last (or only) filename on command line as default for saving
Show the "save changes" dialog before the main window is destroyed
Check for changes at regular 'quit' events as well
Catch changes to the info of the current dive when quitting
Tracking changes to tanks is trivial
Simplistic first attempt to get changes saved when quitting subsurface
It's getting to the point where I'm happy with this. This just makes
the spacing between the location and the notes a bit bigger to visually
separate them more, and adds units ("min") to the dive duration (and
removes the seconds, that really didn't make any sense at an overview
level).
For really long dive locations, we now limit the width to the same size
as the date and time, and force it to a single line - with an ellipsis
if it ends up being too big.
Also, since we no longer use any markup anywhere, we migth as well show
the dive buddy information too, as we don't need no stinking quoting.
Show the "save changes" dialog before the main window is destroyed
By using the delete-event callback instead of the destroy callback we are
able to display our dialog (and the file-save dialog) while the program
window is still being displayed. Much nicer this way.
Catch changes to the info of the current dive when quitting
As the application shuts down we do one more check to see if the dive that
is currently being displayed has been modified (we previously just checked
as we switch dives)
This way we can avoid the need for quoting, since we can just use text
rendering instead of markup for the free-form fields. And we will want
to make the pango layout width different for the date and location,
since we want to fit the depth/duration to the right of them.
I still haven't set the different width for the date/location, but this
at least is going in the rigth direction.
The default cairo font seems to be sans, but the default pango font is
serif. Maybe it has something to do with my system font settings, but I
doubt it: my desktop font settings are all sans-serif. So I think pango
is just showing bad taste.
Anyway, this just hardcodes the font to "Sans". Maybe somebody wants to
make this all part of preferences some day, or pick it from their
desktop font preferences. In the meantime, just fix the pango brain-damage.
projects / ext / subsurface.git / log