summaryrefslogtreecommitdiff
path: root/gpsd.c
Commit message (Collapse)AuthorAgeFilesLines
* Process mode arguments a bit more forgivingly.Eric S. Raymond2009-03-021-1/+1
|
* Full support for setting parity and stop bits.Eric S. Raymond2009-03-021-9/+29
| | | | | Needs to be tested for non-8N1 modes on the devices that can support it: SiRF, UBX, Zodiac.
* Create infrastructure for allowing changes of parity and stop bits.Eric S. Raymond2009-02-281-1/+4
| | | | | Mainly useful for debugging odd stuff like Trimbles, most things run 8N1 these days. Don't expose it to the user yet.
* New channel allocation policy: seek fix quality.Eric S. Raymond2009-02-261-4/+17
|
* Refactor channel allocation policy.Eric S. Raymond2009-02-261-16/+13
| | | | We're going to try to make it pick devices with better fixes, next.
* Knock the rust off the D-Bus code.Eric S. Raymond2009-02-241-2/+3
|
* Magic-number elimination and compiler hushing.Eric S. Raymond2009-02-241-4/+4
|
* Major and minor API revision numbers need to be returned by L command.Eric S. Raymond2009-02-121-1/+1
|
* All driver method tables are now const.Eric S. Raymond2009-02-121-5/+5
| | | | | | | This required moving one field, the modifiable cycle time, into the session structure; the driver type field is now the default value set on switching to that driver, if the field has not already been set during a previous activation.
* Address tracker bug #14196.Eric S. Raymond2009-02-111-5/+19
| | | | | | Fix the obsolete 'e' command to work with the buffering logic. We still don't report epe, as doing so would require some structure changes in gps.h and a major library revision.
* Typo fix.Eric S. Raymond2009-02-111-1/+1
|
* sirfmon can read from gpsd again.Eric S. Raymond2009-02-051-45/+29
| | | | | | | | | | | The culprit was excessively complicated logic in throttled_write(), introduced in my throttled_write() change of 12 Jan, that aimed to make each NMEA packet in a multiple-sentence transmission of faked NMEA an atomic write. This was intented to solve the regression-test timing problem, but didn't. It turned out to screw up super-raw mode instead. Regression tests pass.
* The 'g' mode switch command now requires, and returns, 'rtcm104v2'...Eric S. Raymond2009-01-301-5/+5
| | | | | ...rather than 'rtcm104'; this is design forward for when RTCM104v2 is fully working.
* Splint cleanup and removal of some unneeded headers.Eric S. Raymond2009-01-211-0/+2
|
* Splint cleanup.Eric S. Raymond2009-01-211-2/+2
|
* fix reporting of a device going offline. from Michael LedfordChris Kuethe2009-01-151-2/+0
|
* hushChris Kuethe2009-01-131-1/+1
|
* Transfer mask dumping is not very useful since we solved the J problem.Eric S. Raymond2009-01-131-2/+2
|
* Improve the packet instrumentation slightly.Eric S. Raymond2009-01-121-1/+0
|
* Fix instrumentation so it reports atomic chunks as well as writing them.Eric S. Raymond2009-01-121-4/+5
|
* Change the throttled_write() logic so each sentence shipped is an atomic write.Eric S. Raymond2009-01-121-29/+44
|
* Be explicit about which write returns we're ignoring.Eric S. Raymond2009-01-091-18/+21
|
* Experimental code for sending hexified binary over the control channel.Eric S. Raymond2009-01-091-0/+21
|
* Prevent shipping of stale data as the first fix when a device is reallocated.Eric S. Raymond2009-01-081-0/+2
|
* fix format and sign comparison warningsChris Kuethe2009-01-061-9/+13
|
* This diff goes most of the way to getting us splint-clean again.Eric S. Raymond2009-01-051-4/+13
|
* crank up the precision of fixes to 9 decimal places.Chris Kuethe2009-01-051-9/+9
| | | | | while consumer receivers are unlikely to have this sort of resolution, but survey receivers are capable of it.
* My change to the scaling of O reports four days ago was backwards.Eric S. Raymond2009-01-031-5/+5
| | | | | | | It was the E report that was wrong. The confidence-interval unit has to be tweaked to 95% at driver level, where we actually have some clue what the device type reports, not messed with in the gpsd.c upper-level code.
* Scale the O reports of horizonal and vertical errors properly.Eric S. Raymond2008-12-311-2/+2
|
* Throw an assertion on failed writes. Eric S. Raymond2008-12-301-15/+15
| | | | Pacifies GCC 4.0's unused-return-value checks.
* GCC now gets huffy about unchecked write(2) returns. Fix some of these.Eric S. Raymond2008-12-291-2/+11
|
* Add a wrapper function around gpsd_hexdump to avoid hexdumping buffers...Chris Kuethe2008-12-271-0/+3
| | | | | | | | ...and copying ascii strings around when they're not going to be printed. This saves quite a lot of CPU. I processed a 50MB ubx binary file. With no "-D" options, this saved nearly 2.2M calls to gpsd_hexdump and the processing time for this file went from 84 seconds to 35 seconds.
* Greg Troxel's patch to allow the user to decide whether to listen on theChris Kuethe2008-12-261-2/+9
| | | | | | loopback or to any address, making the gpsd instance world-readable. A bit more manpage fluff from me. We will probably change this in the future to accomodate ipv6 and multiple listener addresses.
* don't truncate "ERROR\n". from Jan-Benedict GlawChris Kuethe2008-10-191-1/+1
|
* inverting the behaviour of a flag is bad... no cookie for me. Chris Kuethe2008-10-161-1/+1
| | | | | instead, make the "-n" behaviour the One True Way. The rest of the "-n" code can be cleaned later.
* invert the sense of "-n"; Chris Kuethe2008-10-151-2/+2
| | | | | | | gpsd will camp on devices by default unless you use the "-n" flag to try save power. i have no evidence that this attempt at power saving actually does any good in the real world, but i have seen lots of complaints and confusion about it.
* put the listen address in network byte order. Chris Kuethe2008-10-031-1/+1
| | | | | not required for INADDR_ANY but a good reminder if we ever listen on something else.
* validate sub->fd. From Bartos-Elekes Zsolt on gpsd-devChris Kuethe2008-09-101-1/+4
|
* We are now splint clean.Eric S. Raymond2008-07-101-1/+1
|
* Splint got more capable, or something. Remove some overrides.Eric S. Raymond2008-07-101-1/+1
|
* Add a CRC-24Q implementation. It's wrong, but it's a start.Eric S. Raymond2008-06-301-1/+1
|
* RTCM104v3 packet detection. Eric S. Raymond2008-06-301-1/+1
| | | | | This is without checksum support, and the daemon code does not do anything with the RTCM data yet.
* More preparation for RTCM 3. No behavior changes. All regression tests pass.Eric S. Raymond2008-06-271-4/+4
|
* 'typename' is a reserved word in C++. Gary E. Miller2008-01-191-2/+2
| | | | | Change to 'type_name' so gpsd pieces can be used with C++. Reported by Matt Roberds <mattroberds@cox.net>.
* Make the -h output include a list of compiled-in drivers.Eric S. Raymond2007-12-271-2/+8
|
* Cosmetic fix.Eric S. Raymond2007-12-131-6/+6
|
* Comment fixes.Eric S. Raymond2007-12-131-7/+2
|
* Fix to time conversion by Petr Slansky; passes all stable regressions.Eric S. Raymond2007-12-111-1/+1
| | | | Also, change a formal from int to size_t appropriately.
* Improved error and rpgress reporting, plus a typo fix.Eric S. Raymond2007-12-101-1/+1
|
* Log the transfer mask describing what data gets moved from devices... Eric S. Raymond2007-12-101-0/+2
| | | | | ...to user fixbuffers, this is useful in pinpointing data-management problems.
View Lorry Controller Status