OpenCPN

[bdbcat]

Philip....

Hmmmm. I'll try that myself and see if I can reproduce it.

Meanwhile, to get going again, you may edit the configuration file

/home/philip/.opencpn/opencpn.conf

You will see a line like this:

VPLatLon=" 22.1948, -78.1372"

This is the starting lat/lon on program load.
Edit that line to something reasonable for your loaded charts. Save. Restart opencpn.

If that doesn't work, then simply delete the .conf file. It will be recreated generically on the next start of opencpn.

By the way, I would like to see a copy of the .conf file before editing, by email, if possible.

Dave

[FuzzyFeat]

Deleted the (hidden) .conf file and now it works again.
Thanks again Dave.

[gjorgensen]

Dave,

What version of CM93 have you tested against? I snagged a copy of CM93-2001 (cd2 only at the moment) and I'm not having any luck with it on Windows Vista. Adding the top of the tree to the chart folders doesn't seem to do anything. I'm wondering if this version of CM93 is encrypted.... Is there a log file some place or some way to see what it's doing?

[gjorgensen]

Well, never mind, I answered my own question. Once I loaded up CD1 in the same tree CM93-2001 magically came to life. Obviously there must be something in the CD1 tree that openCPN needs.

Dave, I still see the program slowing down if I load too many ENC charts or scroll through too many. No slowdowns with raster or CM93 charts, but using NOAA ENCs if I scroll around loading more than, say a dozen charts it suddenly slows down noticeably. Let me know if you want more details.

[FuzzyFeat]

Well, I'm no expert here, but Dave has said that opencpn was designed as a navigation tool and not well suited for planing. It would be extremely unlikely that under the conditions for which the application is designed, en route navigation, that more than two or three charts be opened.

Just MHO.

[gjorgensen]

I'm just trying to be a good user and report what I see, especially since Dave has asked if this problem was still occurring. It's of course up to the developers to decide if they want to address the issue.

As for your claim that that two to three charts is the typical in-route maximum, a simple trip from my slip in San Diego to Avalon will cause anywhere from 5-7 charts to open, depending on how many of the large-scale charts you have loaded.

[Islander]

Originally Posted by gjorgensen

Well, never mind, I answered my own question. Once I loaded up CD1 in the same tree CM93-2001 magically came to life. Obviously there must be something in the CD1 tree that openCPN needs.

Any chance you could post some details on what this structure looks like?

I'm currently trying to get mine to work.

Thanks.

[bdbcat]

For gjorgensen...

re Opencpn slowdown on multiple ENC loads:

Opencpn caches charts in memory as they are loaded. On Windows platforms, there is currently no limit to the number of charts which will be cached on opening. Charts are never "unloaded". Eventually, as physical RAM is exhausted, Windows will resort to disk-based virtual memory paging. This is, as you know, much slower than RAM based access.

The reason you notice it more clearly on ENCs is that the internal representation of ENC in opencpn is memory intensive, for better chart rendering performance. Loading ENCs will more quickly fill physical memory.

On Linux, we cache a small fixed number of charts, and discard the "oldest" chart in the cache when a new one is to be loaded, using a "Least Frequently Used" cache replacement policy. Chart loading under Linux is substantially faster that Windows, due to the higher performance file systems and memory management subsystems of Linux.

Perhaps we need to have a look at implementing this LRU policy under Windows. I will investigate.

Thanks for the feedback, and I'd like to hear more comments regarding your 2001 CM93 dataset and its coverage/accuracy/etc.


Dave

[gjorgensen]

Islander,

The CM93-2001 set came on 2 CDs. CD1 has a number of folders, of which the CM93 and CM93ATTR folders appear to be the key. The CM93 folder contains the charts, and also a number of .DIC files which it appears openCPN uses to interpret the charts. CM93ATTR contains a bunch more .DIC files that again openCPN seems to need. Everything else is optional. Mine had a copy of CMAPENC viewer, which is fairly terrible, but which you could use to verify the charts are ok.

CD2 only has a CM93 folder which contains the rest of the charts.

What I did: First, create a folder on the hard drive named C-Map. Then copy the entire contents of CD1 into that folder (go to the top of CD1, select all, and drag the contents to the C-Map folder). Next, copy the contents of the CM93 folder on CD2 into the CM93 folder on the hard drive.

In openCPN I added the path to the CM93 folder on the hard drive (e.g. C:\C-Map\CM93) to the chart path and the charts came right up. I deleted the paths to my normal raster and vector charts when using CM93 since openCPN seems to bog down pretty bad with all of those charts loaded. That's not really surprising - most commercial programs would probably start struggling with that many layers of charts.

That was it. CD2 would not work by itself, I think because the .DIC files were not present. Let me know if you need more details about the folder contents.

Dave,

I spent quite a bit of time playing around with the CM93 chart set and it works great in openCPN. Interestingly, you can spend hours cruising the world zooming way in and out, opening probably hundreds of charts with no change in performance. It's all automatic and pretty much seemless with a couple of exceptions. I have had openCPN crash on me when opening too many ENC charts, but I haven't made it crash with CM93 yet.

The biggest exception I've seen is when getting near 180 degrees latitude, which you said you're already working on fixing. Depending on the chart, getting beyond 160 degrees is flaky, and of course you can't scroll across 180 degrees. Tonga, which has charts that straddle those points, gets all goofy if you try to zoom in.

The charts themselves vary quite a bit. Any place with a lot of shipping has great detail. For example, the Torres Strait has as much detail as you can deal with through the shipping lanes. If there's not a lot of shipping it's a lot more hit and miss. Places like French Polynesia or Saint Helena island have a lot of detail around major harbors based on English and French surveys but little detail away from those harbors, while Tonga has no detail at all (although perhaps the 180 line issues are causing the charts not to load?). The southern California charts are as good as the latest NOAA ENCs from what I could see. For laughs I entered a few waypoints from SSCA bulletins in places like Fiji and Vanuatu. The Fiji locations looked right, while the Vanuatu locations were invariably half a mile inland. That lines up with what I've heard from other people. I'd bet the Marquesas charts are all off by a mile also.

Your comment about how ENCs are loaded is what I thought it would be, but I'm mostly an embedded programmer (cell phones) and not a Windows app programmer so I didn't want to assume. For now it probably makes the most sense to load only the charts I need, and skip the layers of wide area charts (I don't need the eastern half of the Pacific to get to Catalina) which probably eat up gobs of memory. FYI, I've never really seen openCPN slow down with raster charts.

You guys have done a great job with openCPN so far!

[Islander]

... I'll play around with it, hopefully get it working.


Cheers.

[Islander]

Got it working, thanks.

Cheers.

[phiggins]

Dave,

I just installed openCPN on my XO computer running Intrepid Ubuntu. So far all the basic things to work ie loading .kap charts and connecting a GPS.

But I have a simulation program then sends GPS and AIS data into port /dev/ttyS1 and I set that as both the NMEA and AIS input port in openCPN. When I run the simulator the GPS data makes it to openCPN and see the ship but I don't see AIS targets. Can the NMEA and AIS ports be the same?

Attached is the opencpn.conf.doc (I added .doc) so it would upload.

Thanks

[bdbcat]

For phiggins...

Thanks for the note. Glad XO machine works.

Version 1.3.0 of opencpn does not support GPS and AIS on the same port.

This has been an often requested enhancement, and is scheduled for inclusion on opencpn 1.3.2, due out by 30 June, 09.

I'd like to know more about your GPS/AIS simulator. Maybe it could be used for testing the upcoming opencpn release?

Thanks again
Dave

[phiggins]

Ok thanks Dave.

My simulator is written in Python and works both on Windows and Linux. I've used it to test many navigation programs to see if all their features work and also to see how they display things like AIS closest point of approach.

In simulation mode it reads an AIS script and picks one of ships as being "MyShip" and creates GPS data for it. I then feed both the AIS and generated GPS data to the navigation program through a virtual port (pts for Linux). It creates its own AIS list of targets and also calculates the range, bearing , CPA and TCPA as well as the other data in the AIS message like speed and COG, destination, etc. It can be Paused where it will keep sending the last GPS RMC message and the AIS targets so I can check how the navigation program is displaying it.

It also works in a non-simulation mode where it reads real AIS and GPS input and passes it through to the navigation program. I did this to support this: It has an option to create TTM data so that I can feed my older Furuno radar set data (that doesn't support AIS) the 10 closest AIS targets as TTM targets so it gives it pseudo AIS support.

I'd be glad to give you a copy of it.

Paul

[bdbcat]

phiggins....

Your AIS/GPS simulator sounds very cool. I'd like to get a copy and try it out.
The pseudo AIS for radar sounds very useful. Never thought of that....

Could you zip and email attach a copy to me?

Thanks
Dave