robertl [Tue, 15 Jun 2010 22:31:00 +0000 (22:31 +0000)]
Makefile.in: Changes to build with alternate build systems.
cet_util.h: Likewise.
defs.h: Changes for alternate build systems. Merge from trunk for
trkseg support. Add global_traits. Add power.
configure.in: Bump version number to -beta20090708b.
configure: Regenerated.
gbversion.h: Regenerated.
exif.c: Changes for alternate build system.
gbfile.h: Likewise
gbsleep.c: Likewise
vecs.c: Move track to minimal filters.
garmin.c: Merge from trunk for trkseg support.
gbser_win.c: Explictly use CreateFileA.
gpx.c: Merge from trunk for trkseg support.
gtrnctr.c: Add support for power
kml.c: many new features for Earth 5.2: trkseg in LineString tracks, Power,
extended data, gx:Track, geocaching updates
testo: update with new KML test cases.
trackfilter.c: Add "segment" option.
queue.h: Merge from trunk for trkseg.
waypt.c: Add traits handling.
Enable drag & drop of files into main window.
Smarter handling of formats with unknown extensions.
Allow user to edit filenames in the main dialog instead of just using the
file picker.
robertl [Fri, 26 Mar 2010 03:33:18 +0000 (03:33 +0000)]
Martin Buck delivers "track" upload for Garmin fitness devices.
Changes:
* If tracks and waypoints should be uploaded to a device supporting course
upload, combine them to courses and course points and upload them together
* Split track/waypoint preparation and upload into separate functions in
garmin.c since both preparation functions are needed together for
course+course point upload
* Merged GPS_A301_Get()/GPS_A302_Get, GPS_A301_Send()/GPS_A302_Send()
* Properly handle track segment markers for fitness devices consisting of 2
consecutive invalid track points
* Calculate lap data totals when creating a course. Also create time stamps
(currently using a hardcoded speed of 10km/h) for track points that lack
them. This is required so that course points can refer to track points and
identify them uniquely.
* Fixed course garbage collection to remove unused track points properly. So
far, we compared the track index with the course index instead of the
course track index which inadvertently removed tracks that were
refereneced and left others that weren't referenced.
* Remove duplicate course points (same course index and track point time
stamp) because the protocol spec requires these two values to be unique.
* Create course points from waypoints by mapping them to the nearest track
point.
* Use xrealloc() consistently
robertl [Sun, 7 Mar 2010 03:49:57 +0000 (03:49 +0000)]
From Martin:
* Initialize all gps_*_type and gps_*_transfer variables to -1 exactly once
in GPS_A000() instead of some in GPS_A000(), some in GPS_A001(), some in
both and some not at all
* Bug fix (?): If device reports device command protocol A011, set
gps_device_command to 1 (=A011) instead of 0 (=A010). Changes behaviour
for aviation devices that report capabilities via A001 (if there are any).
If this bugfix is wrong, there should be at least a comment stating
that/why using the "wrong" command protocol is intentional
* Minor indentation/comment fixes
* Converted mixed hex/decimal constants for COMMAND_ID and LINK_ID to
all-decimal. The rest of jeeps and the Garmin docs all use decimal, so
it's easier to read/compare this way
* Bug fix: Set course limits transfer protocol Cmnd_Transfer_Course_Limits
to 565 (course limits transfer) instead of 564 (course tracks transfer)
for devices using device command protocol 1 (A010). Shouldn't make a
difference for gpsbabel, since this value isn't used (yet)
* Bug fix: Set proximity transfer protocol Cmnd_Transfer_Prx to 17 instead
of 0 (unsupported) for devices using device command protocol 2 (A011).
Shouldn't make a difference for gpsbabel, since the functions using this
value are never called
robertl [Sun, 14 Feb 2010 05:33:36 +0000 (05:33 +0000)]
Correct statistics reports by zeroing counters on successful checkin. Lay groundwork for future UI work like disabling formats in menus and disabling update check.