Weather Routing

[nohal]

Originally Posted by boat_alexandra

I mean that it records that it had warned the user, so restarting opencpn, will not result in another warning.

I think this is best, but if the users share a machine, or are forgetful it might not be enough.

This is more than enough, I suppose.

How shall the plugin "detect" the gshhs levels?

1) Should it perform a known land crossing segment that will pass only from the highest level (this is subject to fail with updated gshhs
2) Check disk for the files (and where are they located, normally plugins don't know these paths)
3) add function to plugin api for this?

3 - 1 is unreliable, 2 is doable as the path can be found in the ini file, but not future proof. If you are fast, API 1.15 is still not totally frozen, I think.

Pavel

[rgleason]

What's wrong with a search for "Settings/Dnldr" then finding the GSHHG basemaps path, "D:\\Data-Juno\\Chart\\GSHHG" for example. Then actually going there to
D:\Data-Juno\Chart\GSHHG
and checking if these exist:

poly-l-1.dat
poly-i-1.dat
poly-h-1.dat
poly-f-1.dat
poly-c-1.dat
GSHHG_catalog.xml
etc ?

[Settings/ChartDnldr]
ChartSources=All GSHHG basemaps|https://raw.githubusercontent.com/ch...\Chart\\GSHHG|

[nohal]

Nothing Rick, nothing. Exactly until the next time I, or someone else, for some reason, refactor the GSHHG stuff again...

Pavel

[rgleason]

Dear Pavel, you are really speaking a different language which I do not understand. What is refactor? I really do not know what that means.

I appear to have upset you. I am sorry about that. What should I do now? I got it working, I thought that was the goal. Where do you want GSHHG to be put? I will move it there directly.

I am an Architect, you can call me a hack if you wish, but you're that one who showed me how to compile along with Jesper years ago. I am afraid you created your own monster!

[rgleason]

"Refactor" does not exist in the dictionaries I've checked. "factor" does exist.
https://en.oxforddictionaries.com/definition/factor

[did-g]

Originally Posted by rgleason

"Refactor" does not exist in the dictionaries I've checked. "factor" does exist.
https://en.oxforddictionaries.com/definition/factor

That's jargon
https://en.wikipedia.org/wiki/Code_refactoring

[nohal]

The API is prepared in https://github.com/OpenCPN/OpenCPN/pull/974

[rgleason]

Originally Posted by nohal

The API is prepared in https://github.com/OpenCPN/OpenCPN/pull/974

Wow,

Now I am beginning to understand

"Nothing Rick, nothing. Exactly until the next time I, or someone else, for some reason, refactor the GSHHG stuff again...
Pavel"

I guess refactoring is something you guys are doing all the time, making the code clearer and better, but so it still does the right things. The link was very helpful did-g. Thank you.

Aren't programmers linguists then? Are there tools that can be used to identify code that "smells" as the wiki link suggested?

[rgleason]

Gilletaron, I was re-running your tests and found that they all complete when 10 min interval is selected except this one, which results in endless computing because the destination is out of range of the computing area.

In this case I think we need to increase max diverted angle from 120 to 130 and perhaps try max search angle (under advanced) increased from 120 to 130 and rerun again with a 5 min interval. The idea is to get the calculation area limits to encompass the destination point.

I suppose the at the plugin could initially check if the Destination point and the Start point are within the calculation area limits, but then it would still be necessary to do a lot of calcs anyway, so it is better and probably quicker to just run the routine and see what happens.

Since I had to manually stop the calculation because it will never reach its destination, there is no message... I wonder if there should be some time limit or checking for the the destination being within "range" of the calculation area? With a stop calc and a message "Failed: Destination not in range" ?

Which leads me to ask if there is a way of "fine tuning" the

• time interval
• max diverted angle
• max search angle

so that a destination point is actually within the calculated area, and is successfully reached?

I will show the result in the next post, once it completes. This is taking a long time.

[rgleason]

Look at that. It did complete. Took awhile however. You do an excellent job at picking out of the way completion marks. If you continue to challenge WxRte that way you will need a Univac!

I have no doubt that all "Failures" are simply a failure of the person in charge "PIC" to input adequate instructions to the Abacus!

Next you ask why doesn't the Abacus "know" what instructions are necessary? Can't it adjust itself so it never fails? - I expect these magical characteristics will take more time, but I think the plugin does do very well when given good instructions.

[rgleason]

Pavel added to the Plugin API 1.15 so that GSHHG detection is possible! Thankyou Pavel.
Now Weather_routing just needs to use it I think.

[Gilletarom]

Hello RGleason,

I did a lot of other tests. In particular another test with two searches with timestamps of 1h and 30 mn from Cherbourg and Cowes..

What is curious is that, with the same data, no other change in the configuration, WR the first passes east of the Isle of Wight and the second west of the island. But with 30mn, WR was not able to finish and stopped very close to the finish.

Hum ... Already, we must put less boundary because if there is too much, it increases much the amount of calculation.
Then you have to put a frame around the area, to prevent WR from going for a walk too far.
JonGough, Did-g, probably also Stelian, and probably also Sean are currently working "together" to improve and improve everything.
Let's be patient.

[rgleason]

Gilletarom, I had an idea.... What about allowing the user to stop the calcs (or recompute) from a user selected isochrone, with some changes/adjustments to the parameters?

In other words the user can take a "failure" and turn it into "complete" by giving better instructions to the abacus. This would allow fast calcs for the big crossings, and then fine tuning for the coastal shoreline work and it would be quite flexible. You would work with a configuration to get it to completion.

[rgleason]

Cherbourgh west to Newlin
1 hr interval
110 max diverted course
120 max search angle
Completes quickly with the destination within the computing area.

[Gilletarom]

Hello RGleason,

Originally Posted by rgleason

...... What about allowing the user to stop the calcs (or recompute) from a user selected isochrone, with some changes/adjustments to the parameters?
In other words the user can take a "failure" and turn it into "complete" by giving better instructions to the abacus. This would allow fast calcs for the big crossings, and then fine tuning for the coastal shoreline work and it would be quite flexible. You would work with a configuration to get it to completion.

This is a very good idea that I like very much. And I often thought about it for a few days. BUT, it is still necessary that our comrades developers are ready to do the necessary to change the code.

That said, the way WR manufactures isochrones seems to me questionable. And especially how WR each isochronous by segments that snap to the previous isochron.

From what I understood, Sean and Did-g seem to me to have begun a reflection on this subject.