OpenCPN Version 4.4 Released

[RhythmDoctor]

Originally Posted by RhythmDoctor

OK, Dave, I finally got some time to test. I deleted opencpn.ini, opencpn.log, and CHRTLIST.DAT for a clean install free of most remnants of a previous version.

After some experimenting, I think I found a repeatable symptom - I'll let you determine root cause.

...

I hope you can recreate this issue on your machine based on this description. I've attached .ini and .log files for your reference.

Have you had a chance to replicate and/or diagnose this problem? I'm on vacation for the next week, and might have time to text and/or diagnose further, but I need guidance on what information you would need.

Thanks!

[rooiedirk]

4.5 beta: With OpenGL the prediction point of an AIS target is drawn on the screen edge, when it is in reality outside the screen.
OpenGL off works OK.
Made a Pull Request to correct this.

[reflectionsv37]

Hi Dave!

I finally upgraded from 4.0 to 4.4 and found a problem. The Furuno GPS Upload is failing again.

I'm not sure when it started but it was working fine in 4.0. Looking at the NMEA upload, it looks like the beginning of the sentences for waypoints and routes has changed. I did an upgrade to 4.4 so I don't have an example of 4.0 output

Here is an old file we were once testing with. This shows the correct beginnings for each line according to the manual but other things in the sentence have changed since then.
$PFEC,GPwpl,1302.2565,N,06145.4622,W,WP0001,3,@q12 :57 14DEC10,A,,,,
$PFEC,GPwpl,1402.2570,N,06145.4617,W,WP0002,3,@q12 :57 14DEC10,A,,,,
$PFEC,GPwpl,1502.2569,N,06145.4618,W,WP0003,3,@q12 :57 14DEC10,A,,,,
$PFEC,GPwpl,1602.2568,N,06145.4618,W,WP0004,3,@q12 :58 14DEC10,A,,,,
$GPRTE,1,1,C,1, WP0001, WP0002, WP0003, WP0004
$PFEC,GPxfr,CTL,E

The 4.4 upload begins waypoint sentences begin with "$EC" without a comma vs. "$PFEC," which includes a comma.

The route sentence begins with "$ECRTE" vs "GPRTE".

The GPS beeps and displays that the upload completed, but there are no waypoints or route loaded.

I've attached a JPEG of the NMEA output screen of the upload.

Gene

[reflectionsv37]

A little more info on the Furuno GPS Upload.

I tried turning off the Furuno Upload format button and compared the Furuno upload to the non-Furuno upload. It looks like all the formatting for the Furuno upload sentences are correct, but it looks like it is using the sentence prefix from the non-Furuno upload.

Gene

[bdbcat]

Gene...

Well, that's a regression bug alright.

There is a workaround.
Add the following to your config file:

TalkerIdText=PFEC,

Don't forget the comma.

Let me know how this works, and if you have time, please add as a Flyspray bug so that it doesn't get lost.

Thanks
Dave

[RhythmDoctor]

Originally Posted by RhythmDoctor

OK, Dave, I finally got some time to test. I deleted opencpn.ini, opencpn.log, and CHRTLIST.DAT for a clean install free of most remnants of a previous version.

After some experimenting, I think I found a repeatable symptom - I'll let you determine root cause.

After my first launch of the portable version, the new .ini file is created with a default chart directory pointing to a subdirectory of the portable folder (E:\Program Files\OpenCPN-Test\charts). That entry is in the directory list even if there is no such folder on my drive.

So first I created a folder in that location, put some charts there, and forced a rebuild of the chart directory. OpenGL worked fine under this situation.

However, since I have many charts with limited space on the SD card, and I like to run two instances of OpenCPN simultaneously, I have a separate unified chart directory that is shared by multiple instances (E:\Sailboat\Charts). So I deleted the default path and added a new one pointing to that unified chart directory. After launching OpenGL worked momentarily on ENC, but upon selecting a RNC chart the window froze, and continued to be frozen even if I switched back to ENC.

So I deleted this alternate chart directory and added back in the original default location (E:\Program Files\OpenCPN-Test\charts). However, when adding the folder within the settings dialog, the dialog displays a relative path to the directory, not a full path ("charts" instead of "E:\Program Files\OpenCPN-Test\charts"). This time, OpenGL froze. When I pointed OpenCPN to the different path, OpenCPN saved a relative reference for that as well (..\..\Sailboat\Charts instead of E:\Program Files\OpenCPN-Test\charts). Note that in this case, the relative path backs up a couple levels first.

So it appears that the reason OpenGL is crashing for me is related to the fact that when OpenCPN adds a new chart directory, it is stored as a relative reference, not an absolute reference. Under prior versions of OpenCPN, the program would store the chart directories as a full path.

Of course, it's always more complicated with me. For running portable mode on Windows, relative paths are a more universal way of saving chart directories because when you insert an SD card or flash drive, you sometimes get a different drive letter which breaks everything in the .ini file. Under v4.0 and prior, I would actually go into the .ini file and edit the chart paths, SENC path, etc. to convert them from absolute path to relative path so that it could run under any drive letter (obviously the .ini file requires double backslash \\ in the path instead of single backslash \). This worked fine for me under v4.0 and made portable mode far more bullet-proof.

So the irony here is that v4.4 saves the relative path natively (which I like), but doing so seems to crash OpenGL.

Obviously, this is a correlation with no proof of causation. I'll have to leave that to you.

I hope you can recreate this issue on your machine based on this description. I've attached .ini and .log files for your reference.

Has anyone been able to replicate this problem? It seems pretty simple to test - OpenGL breaks if if you select a chart path that is different from OpenCPN's default.

[bdbcat]

RD...

I have tested and can reproduce the problem. O4 works fine, and O42 does not. So this is a definite regression.

It will take some time to determine the root cause and post a correction. I'm a bit backed up at the moment.

Meanwhile, there is a workaround which worked for me. Simply edit the config file and change the ChartDir path(s) in there into a full, absolute path. Start O, Rebuild the database, and restart O. For me, this allows GL to work as normal.

Of course, in your world, it may not be so simple. You may need relative paths to accommodate relocated disk images. If so, this workaround might not help you as much.

Despite the workaround, this is a real bug. So please, if you have time, capture this in Flyspray so that it won't be lost. We do need to figure out the OpenGL dependency as the root cause.

Thanks, and let me know how you get on with the workaround.
Dave

[reflectionsv37]

Originally Posted by bdbcat

Gene...

Well, that's a regression bug alright.

There is a workaround.
Add the following to your config file:

TalkerIdText=PFEC,

Don't forget the comma.

Let me know how this works, and if you have time, please add as a Flyspray bug so that it doesn't get lost.

Thanks
Dave

Hi Dave!

Update on the Furuno GPS upload issue.

I changed the TalkerIdText in the config file from EC to PFEC,

It corrected the waypoint sentences and they now get uploaded successfully to the GPS. Unfortunately, it also changed the Route sentence to $PFEC,RTE so the Furuno ignored the route sentence and no route info was uploaded. The route sentence should be $GPRTE and 4.4 changed it to $ECRTE

Fortunately, I'm not going anywhere soon so maybe you can get it fixed before I get underway again.

I'll add it to the Flytracker!

Thanks!
Gene

[reflectionsv37]

Added the bug report to v4.2 Stable in Flyspray. v4.4 Stable wouldn't let me add a report??

[RhythmDoctor]

Originally Posted by bdbcat

RD...

I have tested and can reproduce the problem. O4 works fine, and O42 does not. So this is a definite regression.

It will take some time to determine the root cause and post a correction. I'm a bit backed up at the moment.

Meanwhile, there is a workaround which worked for me. Simply edit the config file and change the ChartDir path(s) in there into a full, absolute path. Start O, Rebuild the database, and restart O. For me, this allows GL to work as normal.

Of course, in your world, it may not be so simple. You may need relative paths to accommodate relocated disk images. If so, this workaround might not help you as much.

Despite the workaround, this is a real bug. So please, if you have time, capture this in Flyspray so that it won't be lost. We do need to figure out the OpenGL dependency as the root cause.

Thanks, and let me know how you get on with the workaround.
Dave

Dave,

Thanks for letting me know. It's good to know it's not just me. I can understand that a fix may take some time, and the various workaround options are perfectly satisfactory in the meantime. I've done a lot of manual editing of the .ini files in the past, so I'm comfortable doing that. I'll try to enter it into Flyspray.

It sounds like you understand my situation with running in portable mode. Basically for me OpenCPN is something that travels in a microSD card in my wallet (since my chart files are a little big to synchronize with the cloud), and I stick the card into one of several computers as needed. So use of relative paths in OpenCPN is very powerful and helpful. It seems that you could have a couple ways that you could fix this problem:

1. OpenCPN configuration dialog writes full pathname to the .ini file (in Windows would be C:/etc...), just like it did for v4.0
2. OpenCPN configuration dialog writes relative path to .ini file (like it does in v4.2/4.4) but you fix another part of the code so that this does not break OpenGL

Of those choices, you can probably guess that I would prefer #2, as it give a more universal configuration that works better in a portable environment.

[ccgarnaal]

AIS MOB device course and speed are not displayed.

When I drop a AIS MOB device such as: AIS MOB Devices

Opencpn gives a warning to show it's location. The field course and speed remain empty. This worked correctly in 4.2
When I open the ais target list it is the same result.

Opencpn does draw a track and shows a speedvector of the target thus it receives the neccesary data. Is there a workaround? Or should I go back to 4.2?

[nohal]

Please record the NMEA data stream and post here.

Pavel


Sent from my iPhone using Cruisers Sailing Forum

[cagney]

Multiple CM93 instances regression?

I'm not able to load multiple instances of CM93.
I have 3 directories loaded, each with a separate version of cm93, and they are also in a chartgroup each. 2 of the chartgroups are empty. Adding a cm93 directory to the chartgroups works, after "Apply" + "OK", the directory is still there, but after closing "Options" and going back to chartgroups, two out of three cm93 directories are once again empty. Loading "All active charts" only shows one yellow rectangle instead of three.
Looking at the log, all appears normal. The " CM93Composite Chart Root" is found for all three cm93 directories.
I'm sure this worked in 4.0. I'm not sure about 4.2.
The same problem exists in todays git master.

Anyone else with the same problem?

/Thomas

[P_Dub]

Well I have an issue for those who program OpenCpn:-
The tide for New Zealand should indicate NZST during the northern summer months and NZDT during the winter of the northern hemisphere.
It does exactly the opposite. Could someone take a look and see what might be wrong?
Best Regards and good luck to find the problem.
The time zone works Ok in an other tide program.

[nohal]

P_Dub...
As far as I can tell Auckland is in New Zealand, it is now approximately 11:45 there, we show the same times as MetService (+/- few minutes as they have better data) it is in UTC+12:00 timezone. What exactly is wrong, please?


Pavel