Beta Test / Technical

[idpnd]

Originally Posted by PjotrC

Still, I would appreciate more data for testing: Aid To Navigation, Base Station etc. Maybe someone can record and upload an interesting AIS stream?

Cheers
Piotr

There's a couple of AIS simulators floating around..

[sredna]

Originally Posted by cagney

I have done some testing. Works great when only using CM93. But as you say, raster charts are a bit problematic.
I had full coverage of small scale charts loaded for the route areas, but generally only the background vector chart (black/blue) was shown when zooming to route. It was impossible to change charts, the chart bar was blocked, as was manual zooming with the mouse.

Good. The issue with raster charts should possibly be solved outside the route manager code, but that is a design issue, so we should discuss this. I am also not very familliar with the code base yet, so I do not know what is the best way to select a chart.

A possible workaround while figuring that out, is disabling the zoom to for raster charts.

Another small thing. I imported existing routes from the hard-disk, these had no names so it was difficult to differentiate between them. Why not print the file name of the imported gpx-file if a route name is missing?

Yes, that could be an idea. It should be done in the gpx import code, but if we can agree with Dave and Mark that this is a good idea, I can fix that.

All in all, ...congratulations to this nice feature, you have already come a long way with this feature. I'm looking forward to see more of your ideas, turn into code

Thomas

[sredna]

Originally Posted by idpnd

I did ./configure, but I've never used auto(re)conf before? WIll try in a bit

Autoreconf will regenerate Makefile.in which is again used by the configure script to generate the Makefile. Makes me think again that we should look more at using cmake...

[Psyches]

Originally Posted by sredna

Autoreconf will regenerate Makefile.in which is again used by the configure script to generate the Makefile. Makes me think again that we should look more at using cmake...

I believe Dave uses cmake; I haven't, yet. I hand-built a vsproj to (a) learn the code structure a bit more, and (b) because it's the "standard" way for Win. But if cmake works ok, I'd switch to that too.

Originally Posted by sredna

... but that is a design issue, so we should discuss this. ...
A possible workaround while figuring that out, is disabling the zoom to for raster charts.

Agreed, and the temp workaround sounds good too.

Yes, that could be an idea. It should be done in the gpx import code, but if we can agree with Dave and Mark that this is a good idea, I can fix that.

Thomas, Anders, adding auto-naming code at gpx import sounds like a great idea to me. I think any name at all is better than none, the file name sounds just fine.

[Psyches]

Originally Posted by Netsurfer

I never made this before and it seems a little bit 'complicated' , but I'm sure to get it to work with a little help.
So I wait and see if someone can help me.

Hi Gunther, it won't seem that complicated after you build it once...it's actually remarkably simple. Remember you'll also need a build of wxWidgets, as per the build instructions in cvs.

If you want to continue with the "standard" VSE2008 approach, I suggest you start with this thread if you don't have a project. Also, you should put the test source files (.cpp, .h) into existing opencpn source and include folders...it's easier.

Then, add the new .cpp in VSE2008. That is very simple: start VSE, open the project, right-click on the "Source Files" folder, and select "Add" => "Existing Item...". Add the .cpp there, click ok, and you're done.

BTW, this always needs to be done for every new .cpp...so these steps are done fairly often. Note that .h files in existing folders don't need any changes. But for every new folder containing .h files, you'll need to add the folder into a compile options page (also simple but not needed here).

Mark

[idpnd]

Whilst we're doing noob handholding exercises, would someone inform me why my autoreconf says:

$ autoreconf
Can't exec "libtoolize": No such file or directory at /usr/bin/autoreconf line 190.
Use of uninitialized value $libtoolize in pattern match (m//) at /usr/bin/autoreconf line 190.

Did you have do go through these hoops on 9.10 cagney?

[sredna]

Originally Posted by Psyches

I believe Dave uses cmake; I haven't, yet. I hand-built a vsproj to (a) learn the code structure a bit more, and (b) because it's the "standard" way for Win. But if cmake works ok, I'd switch to that too.

Lets talk about this when Dave returns from his cruise then. Having one build system for all platforms would be very nice!

Originally Posted by Psyches

Agreed, and the temp workaround sounds good too.

I'll add the workaround in next version

Originally Posted by Psyches

Thomas, Anders, adding auto-naming code at gpx import sounds like a great idea to me. I think any name at all is better than none, the file name sounds just fine.

I'll come up with a patch.

I consider moving the mark/route storage to gpx as my next project, because it will be much easier only having one storage, and I think seperation from the config file is very sane. What do you think?

Anders

[sredna]

Originally Posted by idpnd

Whilst we're doing noob handholding exercises, would someone inform me why my autoreconf says:



Did you have do go through these hoops on 9.10 cagney?

I believe ubuntu has a package named "buildessentials" or such, did you install that? Or a autoconf package, that should install dependancies.

Anders

[idpnd]

I've got that, and the autoconf package..

[cagney]

Did you have do go through these hoops on 9.10 cagney?

No! Fortunately I had all dependencies installed, and autoreconf was not necessary in my case.

/usr/bin/libtoolize is provided by the ubuntu package called
"libtool"

Thomas

[Psyches]

Originally Posted by sredna

Lets talk about this when Dave returns from his cruise then. Having one build system for all platforms would be very nice!



I'll add the workaround in next version



I'll come up with a patch.

I consider moving the mark/route storage to gpx as my next project, because it will be much easier only having one storage, and I think seperation from the config file is very sane. What do you think?

All sounds good. I also agree separation from the core config file is a good idea.

For that matter, I think the config code itself should move. I like putting import/export of things into the corresponding class code. That way the code to read/write, and do all other operations on an object is together rather than spread all around...and when an object changes, its new import / export data aren't as easy to miss. It'd make sense to make this an interface (pure virtual) implemented by Config, Routes, Route, Waypoint, Track, etc. so higher-level config code can recurse to lower-level, and r/w can happen: Config=>Routes=>Route; and if we choose to split up routes (say, independently index by lat-lon bounds and separate into SQL or individual files) we can delay-load routes easily when we need them.

BTW, I also agree with your "suspicious code" comment on the route point index - it'll be interesting to ask Dave about that, but I'm guessing it can be greatly simplified. Good to look for that kind of thing...

Mark

[HamishB]

Originally Posted by sredna

Hi Mark,
I have found roughly a working formula, details posted in the sf forum (https://sourceforge.net/projects/ope...2/index/page/1). I work on getting a testing version ready for testing, I just have a few more things to fix before I am ready, maybe later tonight (CET).

I use cm93 vector maps for testing, they are precice enough I believe.

Anders

the missing term you are looking for is cos(lat).
see comments in the bug report.

also, I'll mention that you do need dpi, but it is not very easy to guess exactly. You can try to grab a value from X11 or Windows, but even if you can it will generally only be sort of close. In the end you will probably just have to hardcode some compromise value if you want to display a 1:xx,xxx value on the screen, and it will be wrong for 95% of viewers. But close enough so that 1:50,000 and 1:250,000 are fairly close to what you think they are. Just don't try to calculate any exact distances from them.

fundamentally "true" scale factor only works on an original printed page, where 1inch == 1inch for everyone. Any time you can have different monitor physical sizes and different monitor resolutions, any number you had calculated for scale goes out the window. Scanning charts at 300dpi or whatever just exacerbates the problem more. You need to keep track of the entire chain.

One way around this is to allow the user to recalibrate their dpi every time they change monitor or resolution, like GIMP does.


Hamish

(been fighting with this in GpsDrive for years, & I think we've finally nailed it down there, or at least got it to a point where it is as good as it is going to get. see gpsdrive/trunk/src/map_download.c among others)

[sredna]

Hi Hamish, thanks for the tip

I do not think I need to consider dpi, I just want to be able to maximize a specified lat/lon box to the screen.

Probably, I should use the width at the side of the box away from equator, but unless the route is very long on the north/south axis that probably does not matter much - I also need a bit of space around the box, that will eat small variances.

In any case, I will see what I can do with cos(lat)

Anders

[cagney]

Idpnd, and others, who like to try srednas route manager.
This works.
Download the latest code from the CVS.
Download srednas patch.
Copy patch files
routemanagerdialog.cpp to opencpn/src
routemanagerdialog.h to opencpn/include
routemanagerdialog.patch to opencpn

Then run

..../opencpn $ patch -p0 < routemanagerdialog.patch

Then run the normal ./configure, make, sudo make install.
All this leads to a clean run with patch and a clean compilation, provided that all dependencies are installed.
Note that the present CVS code compile as is, no extra acrobatic is now necessary.
Also remember to run "make clean" before compiling a second time (if you need/want to do that).

Thomas

[Netsurfer]

Would be nice if someone could also compile a windows version and make it publicly available for download (or mail it to me and I'll put it online).