With 2.5.0 I have recently ran into a number of occasions where on program start there was an (unexpected) Anchor Watch alarm. Moreover, it was sometimes set on points which I thought are "impossible" - tracks or routes.
A quick look shows that probably the
Code:
AnchorWatch1GUID=
AnchorWatch2GUID=
values in .ini are saved incorrectly (should be empty, but are not). I suspect that
Code:
g_AW1GUID.Clear();
g_AW2GUID.Clear();
is missing in some contexts (deleting a WP?) or perhaps the backup mechanism is interfering here.
However, this would explain only part of the problem. Even with GUIDs erroneously saved, there should be no (or: infinitely small) possibility of assuming the watch on another random object...
I attach relevant part of opencpn.ini and .navobj.xml for one of the examples. A number of routepoints there suspiciously shares the same GUID.
I would like now to rewrite the Anchor Watch code to allow for more Watch Points, but I think some checks or fixes are perhaps needed before on the GUID issue ...
I've noticed a similar problem with anchor watch, though not on start-up. I deleted the mark with the anchor watch still active, then could not stop the anchor watch alarm. Setting another unrelated mark elsewhere then "captured" the anchor watch, much as you describe, and the alarm could then be disabled.
malloc(): smallbin double linked list corrupted: 0x0a
Hi,
I'd like to report a segfault with 2.5.0 on Debian/stable (DebianGIS build rules). Usually I'd let it slide but this happened when I pressed the MOB button. I was just using it to make an instant waypoint where the boat was, but if it had happened when someone had actually gone overboard... :-( displaying merc BSBcharts with the dashboard running, getting nmea from gpsd (libgps). I did have two days worth of gps tracks on screen (logging every 10 sec) which was perhaps a bit unwieldy, but not unreasonable for a long transit.
Re: malloc(): smallbin double linked list corrupted: 0x0a
Quote:
Originally Posted by HamishB
Hi,
I'd like to report a segfault with 2.5.0 on Debian/stable (DebianGIS build rules). Usually I'd let it slide but this happened when I pressed the MOB button.
opencpn.log for the crash and manual program restart:
Code:
...
05:30:00 NZDT: LOGBOOK: 2012-01-23 16:30:00 UTC GPS Lat -45.72505 Lon 170.66605 COG 23.00000 SOG 6.50
06:00:00 NZDT: LOGBOOK: 2012-01-23 17:00:00 UTC GPS Lat -45.66522 Lon 170.71702 COG 34.00000 SOG 9.10
06:30:00 NZDT: LOGBOOK: 2012-01-23 17:30:00 UTC GPS Lat -45.60067 Lon 170.78068 COG 35.00000 SOG 9.70
06:58:39 NZDT: MAN OVERBOARD Time: Tue 24 Jan 2012 06:58:39 NZDT Position: 045 33.8980 S 170 49.2320 E
06:58:43 NZDT:
06:58:43 NZDT: 2012-01-24
06:58:43 NZDT: -------Starting opencpn-------
06:58:43 NZDT: Version 2.5.0 Build 2011-07-26
06:58:43 NZDT: MemoryStatus: mem_total: 1000 mb, mem_initial: 27 mb
...
I have been using OpenCPN as a secondary Nav system on my cruising yacht for a year or so, and love it. I thought it was payback time, so I installed the Beta 2.6.1511 so I could give some beta site feedback. Of course, I should have taken a backup before starting this!
The installation process was a little fraught, trying the upgrade from 2.5.0 to 2.6.1511 resulted in an app that wouldn't start. So uninstall, and reinstall from scratch. I've now lost my Track and Waypoints (small disaster.) Sorry to be a drain on your resources, when I was trying to be a helper!
Any ideas if I can recover this info - I should have backed up to GPX before installation, I now realise. Any other tips for beta testers on real boats, that I should take notice of!?
Thanks for your help and a great piece of software! Andy
Ok, cancel that question. I did some reverse engineering and restored the 4 files in C:\Program Data\OpenCPN using Windows 7's Previous Versions Restore options (right click / Previous Versions tab) then restarted OpenCPN 2.5.0 and tracks and waypoints are all present! Nice!!
This time, I'll send to a GPX for a backup!
Hopefully I'll be able to send some useful feedback at some stage !!!!
Andy...
Could you also try to dig out the logfile from the crashing runs of the 2.6 beta?
I will try to produce a, incomplete, list of causes for crashes during beta testing I encountered so far and some general suggestions on how to handle them:
plugins - likely to cause startup crashes - purge all but grib_pi.dll and dashboard_pi.dll from <OpenCPN install dir>\plugins - especially while transiting from 2.5 to 2.6 where we build against different runtime libs, this is highly probable cause of some troubles - Pretty easy to confirm from the end of the logfile - the last recorded message has something to do with plugin loading, init or deinit
ini settings - likely to cause startup crashes - remove openpcn.ini from the config dir to start with the default values. Please, make a backup of the INI file before and if it solves your problem, give us the old one causing problems for examination. The end of the logfile from the last crashing run might be interesting as well.
charts - once you end up positioned on a problematic chart likely to cause crashes that look like startup crashes - They are not all the same and the devs don't have access to all of them - try to isolate an exact scenario that is reproducible, ideally try to find one chart that causes the problem, the more exact you are, the higher the chance that we will find the cause of the problem. Especially if it's a commercial chart, expect that we will need you to do quite some testing as we simply can't get access to that chart. The long time not seen cause for problems could be corrupted chart database (CHRTLIST.DAT) in the config folder - you may try to delete it (keep backup so we can have a look). Purging the SENC folder also doesn't hurt (all your ENCcharts will be ingested again, so it will cost you some time)
something else - Are you sure that it's not something from the causes mentioned above? If so, try to isolate an exact scenario to reproduce the problem and give us as much detail as you can: OS version, machine specs, logfile from the run that crashed, your INI file etc. The more info we get the more we can figure out from it.
The common Windows practice to try to solve everything with continuous reboots and reinstalls is nice, but it's really a last resort solution and actually has close to zero chance to do anything you haven't done in a controlled manner by hand going through the steps above.
Do not use development versions on production systems if you don't realize the risks - they are meant for testing and can be, and sometimes are, broken.
If you don't want to lose your data, keep regular backups.
Andy...
Could you also try to dig out the logfile from the crashing runs of the 2.6 beta?
I will try to produce a, incomplete, list of causes for crashes during beta testing I encountered so far and some general suggestions on how to handle them:
plugins - likely to cause startup crashes - purge all but grib_pi.dll and dashboard_pi.dll from <OpenCPN install dir>\plugins - especially while transiting from 2.5 to 2.6 where we build against different runtime libs, this is highly probable cause of some troubles - Pretty easy to confirm from the end of the logfile - the last recorded message has something to do with plugin loading, init or deinit
ini settings - likely to cause startup crashes - remove openpcn.ini from the config dir to start with the default values. Please, make a backup of the INI file before and if it solves your problem, give us the old one causing problems for examination. The end of the logfile from the last crashing run might be interesting as well.
charts - once you end up positioned on a problematic chart likely to cause crashes that look like startup crashes - They are not all the same and the devs don't have access to all of them - try to isolate an exact scenario that is reproducible, ideally try to find one chart that causes the problem, the more exact you are, the higher the chance that we will find the cause of the problem. Especially if it's a commercial chart, expect that we will need you to do quite some testing as we simply can't get access to that chart. The long time not seen cause for problems could be corrupted chart database (CHRTLIST.DAT) in the config folder - you may try to delete it (keep backup so we can have a look). Purging the SENC folder also doesn't hurt (all your ENC charts will be ingested again, so it will cost you some time)
something else - Are you sure that it's not something from the causes mentioned above? If so, try to isolate an exact scenario to reproduce the problem and give us as much detail as you can: OS version, machine specs, logfile from the run that crashed, your INI file etc. The more info we get the more we can figure out from it.
The common Windows practice to try to solve everything with continuous reboots and reinstalls is nice, but it's really a last resort solution and actually has close to zero chance to do anything you haven't done in a controlled manner by hand going through the steps above.
Do not use development versions on production systems if you don't realize the risks - they are meant for testing and can be, and sometimes are, broken.
If you don't want to lose your data, keep regular backups.
Pavel
1. Plugins are not enabled, so they are not an issue.
2. To solve the startup crash I can remove the INI file - the one that causes the crash is attached (removed .doc file extension.)
To get my Track and Waypoints back, I simply copied over the NAVOBJ.XML file from 2.5.0 and all appears perfectly. I exported from 2.6.1511 to GPX as well just to check out data integrity and to get a backup.
I am getting a crash every time I close OpenCPN now, not sure what is causing that yet...
I hear what you're saying about not using Beta software in a real situation, but I have a Raymarine C80 in the cockpit doing the real nav, and MaxSea running in parallel in WinXP mode parallel, so I have a bit of insurance. And since I have B+G instruments, autopilot, GPS and AISNMEA data all transmitting through Bluetooth into GPSgate. This then multiplexes into OpenCPN, GoogleEarth, PolarCOM and MaxSea. I'm cruising the Philippines, so CM93 charts but mainly scanned charts in KAP format. So, I reckon an interesting test platform for OpenCPN?
[Win 7 Pro x64 SP1 on Lenovo X200 Tablet with 4G RAM.]
Some more experimentation, I notice my Track & Waypoints data file is much smaller in 2.6.1511 (2701 KB) versus the same data in 2.5.0 (9835 KB) This is true whether the data is in C:\Program Data\opencpn\navobj.xml or an exported GPX file of my chosen name. I take it that you are compressing the data in this latest release (or is something being lost?)
I can ZIP and send the files if you'd like, but they are too large to attach here.
Andy...
This is intended. There's a change in the way we handle tracks in 2.6 - we removed a bit of abstraction that made the trackpoints hold the same internal identification data as routepoints and so are dumping this info from the files (no part of the data important to the user is lost) - you should observe a dramatic performance improvement with very long tracks and the files should be portable pretty painlessly in both directions between 2.5 and 2.6.
Can't say per Win7 BUT on xp..
Running the 2.6 Betas as PORTABLE versions can be is a good way to run really parallel installations-without interfering with 2.5 Stable...
...use 7Zip (also available as a Portable!) to extract the files from the opencpn beta installer "exe"
and read this... Portable OpenCPN | Official OpenCPN Homepage
you can run it at the same time with other versions...all your stuff is in the one folder and the folder can be sitting on your desktop if you like.
.Is this a bug?..one thing that continues to seem odd to me is losing "window focus" after fiddling in the toolbox....that is, upon closing the tool box, I can't scroll, zoom, or use keyboard shortcuts until I "revive" the focus by left-clicking in the chart window....
This also seems to happen when the window is "restored" from the taskbar, from the minimised condition, that is.
My OS is XP....This applies to 2.5 and 2.6betas that I know of.
This is not the normal case in most Windows apps?....When a Preference dialogue is opened, used and closed ("ok" or "apply"), then the main window is focussed and can be used without left-clicking in it...