OpenCPN Version 2.1.0 Beta

[oem]

I have had numerous crashes since last, but I still cannot find any consistent clou's:

1) The no-start problem, - sometimes I have to delete CHRTLIST.DAT to make the program start. In fact I have had to do that quite frequently. Does that indicate anything? Or is that part of the cause?

2) The panning-and-zooming around problem. My original problem. Sometimes followed by the no-start problem. Sometimes the program seems to run ok, but then just just suddenly opens again with the no-start screen.

I have (again) removed the program completely and re-installed, but still with same result, even with no-start.

I have also (again) copied just two charts (SK_BRJNG.KAP and S_157_25000.KAP) to a separate folder (no danish special characters) and rebuilding the chart database, but doing so I could get both no-start and the panning-and-zooming crash (two charts).

Bottom line: If this problem is only my problem, then leave the problem, maybe the solution will come up some day.

[Gilletarom]

hello,

First :

Installing the 620, I change language from english to french. After I clicked twice "Add to select" on the button. Thus, it were twice the same charts.
I had a mistake: See the first attach line.
Buts answering OK concealed continuing without problem.

Second :
Why the message is it in english. I looked in the file language, this item is not present. And then we cannot translate it.

Third :
The same question for the second attach file.

Why is it also in english

Best regards,

Gilletarom.

[BluegillBoy]

Originally Posted by BluegillBoy

Does this also affect the behavior of S-57 charts published by NOAA? For example, using the Lake Michigan ENCs, there's a black stripshown when quilting across Lake MI just south of Kewaunee WI. Clicking on this area changes some areas to display charts, changing scale also turns some on. There's a small section in the middle of the lake that never displays a chart and clicking/changing scale on this area several times ultimately throws it out of quilting mode unexpectedly.

image001 is first moving up to the area in question and shows black areas where no chart quilts.

image 002 is after clicking on the change scale buttons.

image 004 and 005 shows other instances where charts don't quilt in.

I've got the NOAA ENCs for lake Michigan loaded and the USACE ENCs for the Illinois Waterway. Next I'll try with only the NOAA charts loaded to see if there's some weird but improbable interfereance.

[BluegillBoy]

I tried with only the NOAA ENCs loaded... same issues.

blu

[bdbcat]

OEM....

Thanks for your patience. I found the bug. Happens only on Windows, and only if the PP parameter in the kap header is "UNKNOWN". This leads to an uninitialized variable, which will produce different results on different runs. Completely unpredictable behavior.

Fix was simple, will have no regression problems. We will now assume that if the PR=UNKNOWN and PP=UNKNOWN, and the format is BSB3, then the projection is Mercator, and the PP (Projection Parameter, for Mercators this is the projection latitude) is 0.0. All will be well with this assumption, since we are going to use the embedded georef solution for this chart, and it presumably is sufficiently accurate.

I thank you for your persistence in finally working this problem to the top of the list.

Dave

[bdbcat]

blu....

Thanks for the images.

1 and 2: Hard to believe, but I think there is really no coverage on the NOAA ENCs for this little sliver in the middle of the lake. Probably comes from the conversion of polyconic paper charts to non-projected ENC dataset.

Hope there are no rocks out there!

3 and 4.
A subtle point. Here is (partly) how OpenCPN quilts ENCs.
a. Include all ENC cells that are co-located with the center point of the screen.
b. Include any other ENCs which intersect the screen in any way, as long as the area covered by each candidate cell is greater than 10 % of the screen area.

Why do we do this? Because at small view scales, there may be hundreds of large scale ENC cells potentially on the screen. There is not enough memory nor processor cycles in the world to open and display all of these cells. Pan performance would be impossibly slow. I tried it. Looked good, but my linux box was saturating the swap space, and no cursor interaction was possible.

I guess we could tweak down the 10% value to more "aggressively" quilt, as Manimaul would say. Or a config option? Anyway, this is an area of focus for the next Release after 2.1.0

This is a persistent issue with many NOAA chart regions. The cells do not overlap much, and there seem to be few mid-range scale cells. On IENC cells it is most apparent. They do not overlap at all, and they are all large scale.
We see the problem in Europe as well with IENCs.

What to do now? I recommend that you include cm93 in your chart set. cm93, if enabled, is always used to fill in the last remaining blank areas in a quilt. It may come into the quilt at quite a small scale, but the major navigational features are apparent. Doesn't help the IENCs too much, though, except to help you identify the continent.

Thanks again for bringing this up
Dave

[BluegillBoy]

well good effin vodka grief. and you're absolutely right, there is no ENC coverage for those gaps in the Lake MI charts from NOAA. Given that they're about a mile wide, at 500 foot depths, I guess I'm not in too much danger.

: )

and the 10% rule makes perfect sense, since when I'm actually cruising at my normal 8 knots, the charts have plenty of time to load at scales that would be effective and informative. It's just when I'm at home and skipping hundreds of miles in a click or trying to view the entire Lake MI that the lack of coverage is even visible.\\

Good choices on the logic and trade-offs.

blu

[bdbcat]

Hi debian fans!

We are getting close now.

I have made available a Beta debian release package for download and test install. This package was generated from the cmake script, and should install normally to /usr/local


http://opencpn.org/downloads/beta/opencpn-2.1.622.deb

I would like for debian (Ubuntu) users to try a new-user-style install of this package to verify the release methodology.

Thanks
Dave

[jpiebrig]

Dave,

I have downloaded

http://opencpn.org/downloads/beta/opencpn-2.1.622.deb

and I have installed it without problem. Need "to play" a little with it now...

Ubuntu 10.04

Jean-Pierre

[oem]

Originally Posted by bdbcat

OEM....

Thanks for your patience. I found the bug. Happens only on Windows, and only if the PP parameter in the kap header is "UNKNOWN". This leads to an uninitialized variable, which will produce different results on different runs. Completely unpredictable behavior.

Fix was simple, will have no regression problems. We will now assume that if the PR=UNKNOWN and PP=UNKNOWN, and the format is BSB3, then the projection is Mercator, and the PP (Projection Parameter, for Mercators this is the projection latitude) is 0.0. All will be well with this assumption, since we are going to use the embedded georef solution for this chart, and it presumably is sufficiently accurate.

I thank you for your persistence in finally working this problem to the top of the list.

Dave

Well, - I like that an error was found . And excuse that I was not able to identify it better. When a program behaves unpredictably it's difficult to stay 'calm'. Now, - after knowing what I might have been able to do to correct it myself, i feel a bit embarrassed. That's life. Thanks indeed!

[petermoree]

checked this and have the same problem
depth of 2,5 meter is found in the VALSOU = 2.5 (m)
when you do an Object Query in the chart where only the X shows

X shows when *paper chart* is ON
Depth shows when *symplified* is ON
You may have to restart after the settings are changed

Tested this is with Vista and build 620

Peter

[petermoree]

Vista Build 620 CM93 2010 set

Test: drop mark EXACTLY on Sala I Gomez 26 26.7 S 105 27.7 W
Test also works on other locations with a small detailed area (large scale) chart that is located in an otherwise empty area ( e.g. reefs in remote locations).


After dropping the mark ZOOM IN with your mouse ON the mark.
You will find the detailed map of the island.
Now move your mouse pointer just outside the detail chart area and zoom out, then zoom in again. NO ISLAND !!!

When you put your mouse on the MARK, you will find the island again with zooming in/out.

The results are slightly different when QUILTING is OFF.

Then you will see at least the chart outline at a certain scale.


I hope the above test will help you find the reason and solution.

I still think that the option to show all chart outlines at all scales will help at least indicating that there is something there. An alternative could be a mark of some kind in all locations for which only the largest scale (most detailed) charts are available.





Please also note that in the first picture the 1000 meter depth contours (small circles to the left of the island) are indicated without their depth value.


Peter

[Colorman]

Hi

Loaded 622 on Ubuntu 10.04 Using CM93 (2009) and NZ raster charts. When I select highest res (left most chart) OCPN hangs. Quilting on or off. Have forced database rebuild no luck. Keep up the good work.

[bdbcat]

Petermoree....

I suggest that you switch OFF Toolbox->etc->"Wheel zoom to cursor"
so that zooming does not shift the viewport center. It can be very confusing...

Unfortunately, cm93 is not a truly seamless worldwide database. Display selections that work well for one area may be not right for another. In this case, display of all chart outlines worldwide makes an impossibly dense and distracting image for most of the littoral navigational areas of interest. Perhaps an selectable option would be best. We will study for the next version.

Meanwhile, you may want to use SingleChartMode when passage planning in the Pacific.

Thanks for your comments.


Dave

[bdbcat]

Colorman....

Welcome aboard.!

Do you have a lat/lon or chart reference for the ubuntu hang case you reported?
Screenshot?

Thanks
Dave