Beta Test / Technical

[sae140]

Can't get 1.3.4. to run on Win98SE.

Error message reads:
OPENCPN caused an invalid page fault in
module OPENCPN.EXE at 016f:0042aef0.
Registers:
EAX=013ef9e8 CS=016f EIP=0042aef0 EFLGS=00010202
EBX=00000000 SS=0177 ESP=013ef9ac EBP=ffffffff
ECX=024c97b0 DS=0177 ESI=0000002c FS=0e6f
EDX=00000000 ES=0177 EDI=20203a65 GS=0000
Bytes at CS:EIP:
8b 46 d8 50 e8 0e 64 fd ff 3b 44 24 1c 75 0a 8b
Stack dump:
022a6fd0 00000000 ffffffff 0000004f 00428cea 013efa00 013ef9e8 00000001 53b6d083 023b8c10 00000001 013efcc0 00000001 024cb98c 00000000 023c6a3c


Hope the above info is of use.

[bdbcat]

Sae140....

There is a bug in 1.3.4.
This may be worked around as follows:
1. If you start from a clean install, opencpn will ask you to select the directory(folder) containing your charts by immediately opening the Toolbox->Charts tab. If you skip this step, and cancel the dialog, it will crash. Ooops...
2. So, simple workaround is to select and add a chart directory on the initial prompt into the toolbox->charts tab.

3. Alternatively, you may install 1.3.2, and then install 1.3.4 without un-installing 1.3.2. This will use your old config and chart database files.

I think this should work for you. I use W98 on one of my main everyday navigation machines, probably for the same reason you do. Boots fast, no call-Redmond authorization required, shuts down fast, etc.....

Good luck
Dave

[sae140]

Originally Posted by bdbcat

Sae140....

There is a bug in 1.3.4.

simple workaround is to select and add a chart directory on the initial prompt into the toolbox->charts tab.

I think this should work for you. I use W98 on one of my main everyday navigation machines, probably for the same reason you do. Boots fast, no call-Redmond authorization required, shuts down fast, etc.....

Many thanks Dave - that was exactly the problem - I thought I'd fire-up the program first, then select the chart folio afterwards .... Hah - that's me trying to be clever !

Win98 ? Yes, exactly - it does the job - fires up and closes down quickly, DOS-based with direct access to ports, and I find the '98Lite' config is rock solid.
Also enables me to run older nav programs (plus hopefully OpenCPN in the future) on 'disposable' Thinkpads with 300Mhz processors. It's either Win98 at full speed, or XP on the crawl - and I know which I prefer !

Many thanks, Colin

[phiggins]

dave,

Running openCPN 1.3.4 under XP I occassionally get exceptions. I have sent you the Dr. Watson dump.

[manimaul]

Originally Posted by sae140

Win98 ? Yes, exactly - it does the job - fires up and closes down quickly, DOS-based with direct access to ports, and I find the '98Lite' config is rock solid.
Many thanks, Colin

For a computer that is never on the internet, w98 isn't a completely horrible idea. Another thought, since we're running on the linux platform, creating a special bootable linux+opencpn build for usb sticks and installs on old hardware could be cool.

It should be feasable for it to run on as little as 300mhz and 128mb ram.

[dacust]

Minor issue:

Sometimes when I switch from another application back to OpenCPN, the scroll wheel isn't recognized. I switch back to a different application, scroll the wheel, switch back to OpenCPN and then it works.

Doesn't matter what the other application is. Doesn't matter if the other application even does anything on scroll. Just matters that I get out of OpenCPN and flip it once.

Only happens occasionally.

If this isn't something you can track down easily, I'll do some testing and see if I find a way to reliably replicate it.

WinXP OCPN 1.3.4, but I noticed it in earlier versions, just never thought to report it before.

-dan

[blubaju]

Originally Posted by dacust

Minor issue:

Sometimes when I switch from another application back to OpenCPN, the scroll wheel isn't recognized. I switch back to a different application, scroll the wheel, switch back to OpenCPN and then it works.

Doesn't matter what the other application is. Doesn't matter if the other application even does anything on scroll. Just matters that I get out of OpenCPN and flip it once.

Only happens occasionally.

If this isn't something you can track down easily, I'll do some testing and see if I find a way to reliably replicate it.

WinXP OCPN 1.3.4, but I noticed it in earlier versions, just never thought to report it before.

-dan

SAME xpERIENCE, didn't want to report as I thought I am the only one.

The occasional "silent" crashes when GPSGate is on still happen, as, with a factor of 1:10, crashes with a C++ error message (bad function call). No time now for further testing, in a few weeks again.

[Gone Troppo]

I have a problem with OpenCPN either not showing some cm93 detail or not finding a more detailed chart. 1.3.4 on XP, location is Darwin Harbour at 12 28.268S, 130 50.867E
Best detail chart shown is F 1/20000 2005, also tried some older charts eg 1996. On Cmap I seem to go to a Harbour chart(G) where it shows berth numbers and the anchors securing the ro ro dock. Not sure if it is my setup or a bug.
Otherwise runs fine.
Regards, Stephen

[ColdFusion]

Confirmed (v1.3.4 on Vista).

In OpenCPN Darwin harbour will only go to 1:20,000 (2009). In CMap ECS using the same chart directory the scale will progressively go to 1:1,500.

Also, with chart detail set to maximum no ro-ro berth anchors (etc, etc) show up in either OpenCPN or CMap ECS at 1:20,000 (this is not a complaint, just an observation). They start to appear in CMap ECS at 1:15,000 but you simply can't get to that scale in OpenCPN. I suspect this is a general OpenCPN chart scaling issue (not just related to Darwin harbour) because I had thought my old version of CM93 charts went to a much smaller scale but I changed to CM93 May 2009 ed at the same time as starting to test the beta versions of OpenCPN. Didn't think anymore of it until I read Gone Troppo's post

[Edit] Exactly the same problem at Dale 051.42.446 N 005.10.034 W (OpenCPN won't go beyond 1:20,000 and is missing lots of chart detail. CMap ECS will go to 1:1,500 + shows much more detail). Ditto Abermenai Pool 053.07.701 N 004.19.573 W.

[sae140]

I'm seeing some strange artifacts on CM93 charts with OpenCPN - kind of symbols on lines.
They're difficult to describe, so I'm attaching some graphics of the region 53.00N / 000.24E from OpenCPN, and images of the same area from MaxSea, DKartNav and CMWFW as comparisons.

[sinbad7]

SAE140..

You will see this caution area also in cm93 in zoom 18500

Feature Index: 121
Attributes
INFORM (PORT INFORMATION SERVICES
For procedures and working
details of the Boston and
King's Lynn Port Information
Services see Admiralty List
of Radio Signals.)

[Wester]

Strange artifacts - CM93/2

I see them to around Sweden and Norway

Leif

[bdbcat]

Gone Troppo / ColdFusion....

There are several things going on here:

1. Failure to show anchors/chains and berth numbers is a bug. These are not S57/S52 standard objects, but are cm93 private objects. I will track them down, and get them rendered...

2. Failure to zoom further is not a bug, it is by-design behavior. Long time forum members will recognize my hot-button.

At this geo point, there is no larger scale chart in cm93 than the 1:20000 F scale chart. As you can see in the lower right corner box of the status bar, the chart may be zoomed to a factor of 4X overzoom. 4X overzoom is all that is allowed by opencpn. For a 1:20000 F scale cell, this equates to an actual onscreen scale of 1:4800, as indicated by the TrueScale value. Zooming further provides no more information, and exaggerates, (or overstates by implication) the inherent accuracy of the chart.

Theoretically, if we were to be compliant with S52 standards for big ship ECDIS displays, we should zoom no further than 2X.

I'll try to put it another way. Looking at the CMAP ECS coverage of the same area at 1:1500 on an F scale chart is nearly the same as looking at the 1:20000 paper chart of the same area with a 13 power magnifying glass. The only difference is that the features on a vector ENC are always rendered cleanly, without the pixellation that would result from looking thru a glass, or zooming a raster chart. Thus the misrepresentation.

I'd recommend you read Calder on this subject. He certainly writes better than me.

Frankly, the users coming from CMapECS and MaxSea have worn me down on this issue. My soapbox is getting thin. I'll say it one more time, and then throw in the towel.

Excessive overzoom on vector charts is a dangerous use of ECS, and will lead to misinterpretation of chart data.

The next version of opencpn will have an "expert" mode option, allowing near infinite overzoom. I will not use it.


Good Luck, and safe sailing
Dave

[bdbcat]

SAE140....

What you are seeing is "Symbolized Boundaries" of a caution area.
Part of S52 specs. You can change the symbolization rules in Toolbox->VectorCharts->Boundaries. "Plain" may be less distracting.

Dunno why the other ECS samples choose not to show the boundaries at all....

Thanks
Dave

[ColdFusion]

Originally Posted by bdbcat

Gone Troppo / ColdFusion....

There are several things going on here:

1. Failure to show anchors/chains and berth numbers is a bug. These are not S57/S52 standard objects, but are cm93 private objects. I will track them down, and get them rendered...

2. Failure to zoom further is not a bug, it is by-design behavior. Long time forum members will recognize my hot-button.

Dave

Thanks for the explanation and apologies for raising something that has been discussed before my time here. However, could I highlight a few points...

With regards to 1) please note that it isn't just the anchors/chains that are not displaying - there are various and numerous other items that aren't displaying too, including charted depths at Dale, for example. As many items (such as anchors/chains) only show up in CMap ECS at 1:15,000 and below, I'm guessing this might be why they're not showing up in OpenCPN.

On point 2) I do find it odd that a commercial package such as CMap ECS (over)zooms to 1:1,500 (and displays more chart detail) given your explanation. Ditto MaxSea. This is not a criticism - just an observation.

Additionally, in the Feature Requests forum there have been requests for increased zooming together with an anchor watch function.

Sorry to say this but the present lack of zooming to the same level as CMap ECS together with the missing chart detail will see me using CMap in preference to OpenCPN, but I'll look forward to the next version with the 'expert' overzooming function. Given the exaggeration of chart inaccuracies that you describe, could the text, "CAUTION! OVERZOOM" be added after the chart scale to alert the user of possible inaccuracies?

Keith