Weather Routing

[rgleason]

Larfalitl
I believe the Linux version does not run with Climatology, you need to use gribs. I had changed the code for the last functional windows compile because I believed that Sean wanted it to work... Sometimes Sean is not too verbal about these things, and he did not expect me to monkey with his code, but I did, because I have been using Climatology with Wx_Rte from the beginning.... I am planning on letting him call the shots now with respect to working with Climatology, so the next wx_rte version will work as he determines!

There are some bug reports that in Linux you need to open grib file first or Wx_rte will crash Opencpn eventually. Would like to know more about that if you find something out. Tracker is here go ahead and register, so you can participate. Your input will be helpful. Thanks.

[Larfalitl]

What?!? hahaha, I doubt the PI needs a text html file to function.

Cool, I expected the wiki was behind the interface. The interface may need a little work as it is a bit obtuse to use.

Funnily, I bumped in to Sean in a number of places in Vanuatu & downloaded the full year of current data so he could create the tool to process all the other years. It was a long day over 3G to get that file down...
Yup, I've updated to the new current files & suggested the overlay colours may need to be tweaked. I'll find the email & post on flyspray.

Any idea where? We headed south to New Cal & he was talking about going north somewhere.

What is wx? As in wx_rte & fax. Is it just an abbreviation of weather?

Cool, looks like I'm on the right track with the WR PI. Is the forum the best place to discuss features? Once the idea has crystallised move it to flyspray. Doesn't seem to be much discussion there.

Cheers

[Larfalitl]

WR PI works with both grib & climatology PIs under Linux, both together or either one separately. Well, yesterday it did. Today it only works with climatology data, I'm getting "grib failed" when I try grib data. *sigh*

Yeah, Sean definitely wants the PI to work with climatology. We discussed the PI at length. The PI will start the route with grib data (if available) then move to climatology data when the grib runs out (assuming you have both open).

I'm not quite getting the whole grib vs climatology debate. climatology is a statistical probability, grib is guess made by computers based on what weather has happened recently.
Don't get me wrong, I love gribs & use them extensively. But their usefulness degrades exponentially as their timeline increases, at some point the climatology data has a higher probability of occurring that the "guess" the grib computers will make.
It's all shades of grey to me. In the near term the gribs are more probably correct, in the longer term the climatology is, at some point there is a cross over.
Neither is accurate, both are better than nothing.
Yes, today I sail in what I have, the question is, do I sail above or below my rhumbline because I believe the conditions will be better there in the future?

As to the bug reports, I created them . I'm not a coder, so know little about this dark & mysterious art. But if you can point me to crashlogs or any other tools I'm more than happy to help!

Cheers

[rgleason]

...doubt the PI needs a text html file to function. - Yes you do need the data/weatherroutinginformation.html file!

...interface may need a little work as it is a bit obtuse to use. -Suggestions in Tracker please.

... bumped into Sean...in Vanuatu & downloaded the full year of current data so he could create the tool to process all the other years. It was a long day over 3G to get that file down... -Awesome, circumstance. So thats why I had to download another 12gb to the base current data for years 1992-2012! -Thanks. Sean is quite a personality I think, glad you met him.

...updated to the new current files & suggested the overlay colours may need to be tweaked. I'll find the email & post on flyspray. --Thanks.

...What is wx? As in wx_rte & fax. Is it just an abbreviation of weather? -Yes I get tired of typing it out. I could type WR instead.

...Cool, looks like I'm on the right track with the WR PI. Is the forum the best place to discuss features? --yes, then post in Tracker "pi_WeatherRouting" project! Thanks for your Tracker posts.

Once the idea has crystallized move it to flyspray. -Yes.

[rgleason]

...WR PI works with both grib & climatology PIs under Linux, both together or either one separately. -- I am surprised. Which version of the file? Did Sean give you a more current version?

Well, yesterday it did. Today it only works with climatology data, I'm getting "grib failed" when I try grib data. *sigh* - Perhaps try another grib, some types don't work.

...Sean definitely wants the PI to work with climatology. We discussed the PI at length. The PI will start the route with grib data (if available) then move to climatology data when the grib runs out (assuming you have both open). --Yes that was my understanding too, but that's not the way he coded the Versioning portion of the plugins. I believe his version numbers exclude Climatology from working.

I'm not quite getting the whole grib vs climatology debate. climatology is a statistical probability, grib is guess made by computers based on what weather has happened recently. --- Yes, true. Grib is predictive based upon known weather patterns and it's value and accuracy does degrade. Try using qtVlm and join a virtual race and learn about the rate at which gribs change, it is a good exercise.

Don't get me wrong, I love gribs & use them extensively. But their usefulness degrades exponentially as their timeline increases, at some point the climatology data has a higher probability of occurring that the "guess" the grib computers will make. -- Try running long voyage routes in the tradewinds and against the tradewinds, and importantly in areas where the winds are highly variable, you will see that the Climatology routing is too fast where it is highly variable. We need several user selectable algorithms for climatology long term routing (Like visual planner), I have entered this request into Tracker.

It's all shades of grey to me. In the near term the gribs are more probably correct, in the longer term the climatology is, at some point there is a cross over.
Neither is accuraate, both are better than nothing. -- Agreed.

Yes, today I sail in what I have, the question is, do I sail above or below my rhumbline because I believe the conditions will be better there in the future?

As to the bug reports, I created them . -- good job.


I'm not a coder, so know little about this dark & mysterious art. But if you can point me to crashlogs or any other tools I'm more than happy to help! --Your help will be great. You know where the ini and log files for opencpn are? - Pavel and Dave are making better crash tools too. --I use Linux in VirtualBox and I like linux version, but I use Windows more. Its good you are testing in Linux.

[Larfalitl]

Originally Posted by rgleason

...WR PI works with both grib & climatology PIs under Linux, both together or either one separately. -- I am surprised. Which version of the file? Did Sean give you a more current version?
-- OpenCPN Version 3.3.814 Build 2013-08-14 & libclimatology_amd64_pi19_v02_pi.so. No, just the available downloads.

Well, yesterday it did. Today it only works with climatology data, I'm getting "grib failed" when I try grib data. *sigh* - Perhaps try another grib, some types don't work.
-- yeah, that turned out to be that I was using the wrong instance of grib_PI. I created a copy so I could display multiple gribs but only the original works with the WR PI...
This is to be expected of course.

Don't get me wrong, I love gribs & use them extensively. But their usefulness degrades exponentially as their timeline increases, at some point the climatology data has a higher probability of occurring that the "guess" the grib computers will make. -- Try running long voyage routes in the tradewinds and against the tradewinds, and importantly in areas where the winds are highly variable, you will see that the Climatology routing is too fast where it is highly variable. We need several user selectable algorithms for climatology long term routing (Like visual planner), I have entered this request into Tracker.
-- Hmm, will have to investigate this...

I'm not a coder, so know little about this dark & mysterious art. But if you can point me to crashlogs or any other tools I'm more than happy to help! --Your help will be great. You know where the ini and log files for opencpn are? - Pavel and Dave are making better crash tools too. --I use Linux in VirtualBox and I like linux version, but I use Windows more. Its good you are testing in Linux.
-- Okay, i'll attach the log files to the flyspray ticket.

Cheers

[rgleason]

ChuckSK and Rasbats,

I am trying to fix Weather_Routing cutil.h references and
learn more about Git, Branches and Repositories at the same time, so that my local changes can be moved up to my github repository and then I can ask Sean to bring it into his code.

I have read ChuckSK's patch and with bdcat's post understand why we should not make external references which may just break again when the code changes.

I have also read Rasbat's repair of Route_pi for Salty Paws and now have his 2010 cutil_route.h file to work with. At the moment I cannot "find" any reference to "cutil.h" in the files in the src directory.

Also how do I make the changes so they are recorded by git, and can be uploaded to my git repository?

1. Do I have to use MSVC++ to make the file changes or can I use Notepad++?
2. What git branch should I be using? I have "origin/master" which is a remote and "rick" which I believe is a local branch because does not show with "git branch -remotes"
3. If I should be using the "rick" branch I will git pull, then use "git checkout rick". Make my changes and get the plugin compiling locally. Then I will figure out how to get it onto my repository. I have a weather_routing_pi repository https://github.com/rgleason/weather_routing_pi that evidently still compiles.
4. How would I update that with the repaired code in my local repository? --or is that backwards somehow? --Perhaps I should git clone my weather_routing_pi repository and then fix it?
5. Sean's weather__routing_pi repository has progressed past mine see thishttps://github.com/rgleason/weather_routing_pi/network

Perhaps I should update mine to be equal to his. Do I do a Merge?

Then perhaps I remove the Weather_Routing_pi plugin and clone it again, using my updated repository https://github.com/rgleason/weather_routing_pi ? but what Branch should I be using, my local branch "rick"?

I will worry about uploading later, if I have this right so far. Can anyone advise?

[rgleason]

Found a reference to cutil.h
File: Georef.c Line 41 #include "cutil.h"

It looks to me like cutil.h is just missing because there is no path to the main directory!

--However I checked my repository with a fork of Sean's weather_routing_pi and it does not have a cutil.h file! - and it evidently compiles.

[rgleason]

Ok, to heck with the proper way to do this. I want to fix it.

I just took cutil_route.h from Rasbats and pasted it into my local opencpn-git/plugins/weather_routing_pi/src

Then I found the #include reference to cutil.h at line 41 and changed it to cutil_route.h using Notepad++ (any problem with this?)

Then ran cmake .. and cmake --build . and got the attached output with lots of bad references. Any thoughts.

I looked for a more current cutil.h under github Opencpn version 3.2.0 and could not find it under src directory. Where would it be?

[rgleason]

Ok found the Github Search bar at the top. Searched the current repository (v3.2.0)
found cutil.h under /include/ directory! Right here
https://github.com/rgleason/OpenCPN/...il.h?source=cc

Maybe this is the version I should use, last edited by Nohal, file 101 lines (79 sloc) 3.497 kb
Minor Code prettification and copyright header fixes Jan 27, 2013. Minor code prettification and copyright header fixes

1. 702e630013 Browse code
   nohal authored 8 months ago
   

[rgleason]

Now I realize that I do not know how includes work with C++ (I am somewhat familiar with Pascal) and I do not know if I made the change to line 41 correctly

Then I found the #include reference to cutil.h at line 41 and changed it to cutil_route.h using Notepad++ (any problem with this?)

in Georef.c file Line 41 now reads:

#include "cutil_route.h"

The cutil_route.h file is in weather_routing_pi/src

[rgleason]

Ok, it compiled in MSVC++ and seems to work in debug version. Will clean up later.

[rgleason]

The plugin compiled ok and I have tested it. I runs but when I try to calculate a route, it bombs, taking Opencpn down too. It also crashes Opencpn when you go in to Edit, Then try to Edit the Boat. It just crashes. I can load a Boat file ok.

The date that I know weather_routing worked was 7/20/2013 even with the cutil.h as an external file (not in the plugin src). I'd like to find the cutil.h that was being used then to see if I can stop these crashes.

[Rasbats]

cutil.h is referenced in georef.c in the latest version of the weather_routing plugin.

VSC++ 2010, Win7

git cloned weather_routing from https://github.com/rgleason/weather_routing_pi.git

Using the latest O beta (3.3.910) I copied cutil.h from the main include folder and pasted it into plugins\weather_routing_pi. Renamed as cutil_weather_routing.h.

Reference in georef.c changed to cutil_weather_routing.h

When compiling the errors came from the function roundint. This function does not seem to be used in weather_routing. I MAY BE WRONG. Removed this function from cutil_weather_routing.h.

Compiled with only some warnings and this version runs in 3.3.814 and 3.3.910.

Mike

[Larfalitl]

Hi,

So I've been having a few thoughts about the weather_routing plugin.

Was thinking that "comfort sliders" may be a good idea.
Something like, one slider for each of beating, reaching, beam reach, broad reach & running.
Then the sliders are customised for your comfort. I.E. you only want to beat in small waves, so it is set low. In big sea you want to run away with a quartering sea, so broad reach is set to max. In medium sea your happy to reach, beam reach & run, so these are set in the middle.

Where I'm a bit uncomfortable is this assumes the seas are following the wind. Which is not necessarily the case.

So maybe these are wave angle sliders. Where they would be bow, forward quarter, beam, aft quarter & following. In a big sea you want the wave on an aft quarter, so it is set to max. Or perhaps you prefer big seas following. Maybe you don't like roll, so set the beam slider low. etc.
So the plugin would consider the wave direction as well as the wind direction.

Anyone have any thoughts?

Cheers