Celestial Navigation Plugin Redux

[rgleason]

Stelian, Saw this

printf("Corrupt record in file %s on line %d.\n", mdfile, fileline);
I guess this would be the file error message.

Thanks

[seandepagnier]

Originally Posted by stelian

It wasn't easy, and I don't have a full explanation.

However, after starting to add traces, I quickly found out that adding one trace (or even just a function call), at specific places, made the program work.

It turned out that those specific places where between the ListItem creation and their display in the List.

This smelled like stack corruption.

When I modified the ListItem allocation (allocate it in the heap and no longer on the stack), it didn't crash anymore. I was on the right track.

I looked for some time in the wxWidgets source code, and I saw that when the listitems are converted into native Windows list items, the attributes (TEXT, DATA, etc) are treated in different ways. And maybe (I didn't push the analysis too far), at this point they reference some value on the stack which in no longer there.

Next, I was grepping for wxLIST_MASK in the other OpenCPN plugins and the comment in WeatherRouting_pi put me on the right track...

Note: When I compiled with wxWidgets 3.1, the bug didn't seem to be there. Or maybe it was, but more difficult to trigger...

This was never the fault of the plugin but in wxwidgets.

Perhaps wxwidgets should be fixed? Is it fixed in 3.1?

[P_Dub]

Investigating further, I arrived with a rather large difference of 9.6miles with 2 stars sights as compare with thee position calculated by hand. This could be not acceptable by many people.
Ship's position:- 12-26.6 S, 125-30.6 W
O plugin :- 12-27.2 S, 125-40.4 W
The stars used:- Fomalhaut and Altair
26 June 1958
What do you thing of that.

I shall investigated more tomorrow in case of errors.
bye

[rgleason]

I have written NAV to see if he can check the plugin too.

[CarCode]

Originally Posted by boat_alexandra

This was never the fault of the plugin but in wxwidgets.

Perhaps wxwidgets should be fixed? Is it fixed in 3.1?

It is well known that wxWidgets is extremely buggy. Therefore the lead developer has made several attemps to circumvent them. After many years he finally switched to a newer wxWidgets version when OCPN does not compile for macOS. But sadly not to the latest version.
That means using other wxWidgets functions not have been already used and tested in main OCPN sources is a high risk.

Gerhard

[stelian]

Originally Posted by rgleason

Stelian, Saw this

printf("Corrupt record in file %s on line %d.\n", mdfile, fileline);
I guess this would be the file error message.

Thanks

Done, and pushed.

[P_Dub]

Well, I apologised, I had forgotten the height of eye and the result are perfect.
The position calculated was by the book.
Lat 12˚26.8`S
Long 125˚30.8`W
So, the plugin work correctly in Windows and should in others.
Good job and thank you from a navigator retired.

[rgleason]

Thanks P_Dub. You can't retire yet!

[P_Dub]

I suppose I still have some to offer.
Bye.

[P_Dub]

Originally Posted by rgleason

Thanks P_Dub. You can't retire yet!

I suppose I still have some to offer.
Bye.

[P_Dub]

rgleason
After 4 more sights calculation (double sights) I can confirm the accuracy of the plugin, either in the Northern or Southern Hemisphere. Their maybe a difference of 0.1 mile but nothing to be concern about. So nobody should be unhappy of it. Although it would be nice to have a more elaborated interface, as one may want, but it is not difficult to use for experience men. all you need is a little practice with it.
Best Regards

[rgleason]

Thanks p_dub

[rgleason]

Remembered some test sightings in the manual which might have been useful.
https://opencpn.org/wiki/dokuwiki/do...l_navigation&s[]=celestial#examples_of_test_data

[P_Dub]

One thing I would suggest for the celestial navigation plugin, have a space on the interface for entering the DR position prior to entries. Otherwise you may have a result in the Northern hemisphere instead of the Southern and vice-versa. This is undesirable and definitely inappropriate.
Thank you.

[rgleason]

P_Dub,
Stelian has made a new version 4 days ago and I am not getting it to work very well.

I've compiled and uploaded it to Opencpn Beta File thingie rguser rgpass
Find it under 4.8.0 Test Directory.

I was trying entering the Examples at the bottom of the Cel_Nav manual

Celestial Examples_of_test_data

I kept getting "Warning: Measurement certainly set too small, sight may not appear."

Also it bombed several times on me, rather than prompting "Values are wrong" or something. It should not bomb.

Perhaps I am doing something wrong?