WeatherFax ...

[rgleason]

Ok, I did it again. It seems to be a scaling panning thing combined with incorrect entry of the coordinates using 'TEST-ppam50bw.gif' which is attached.
TEST-ppam50bw.gif (96hr 500mb Feb)

Coord1
132 -90
133 60
Coord2
1547 30
821 0
Mercator

What happens is you zoom in and then zoom out far, until the fax image snaps smaller. Then start panning and the fax will follow, but not at the same speed. If you keep panning in the same direction it follows until it disappears. Then pan back several times, and it follows. Then zoom in and the image gets bigger in scale and then zooom out an the image gets smaller in scale.

The fax coordinates seem to span the equator and not be in the correct location initially. The images below show what happens.

Rick

[seandepagnier]

How did you enter 1547 for a coordinate? For me when I entered this number it automatically changed to 1301 which is the maximum dimension of the image. So the coordinates you specified for me were completely impossible to test.

I tried with the above settings exept with 1301 and had no floating issues. What happens if you use 1301? Are you sure you can enter 1547?


Is that fax real? 85 knots?

[rgleason]

Dear Sean,
The fax is real from NWS/NCEP for Feb 11. We had tremendous storms and 85 knots across large parts of the Atlantic. Environment is upset with our abuse and this is not a good thing for sailors! We must be ever vigilant and more careful than ever. Enough of that, but it is why Weatherfax is an important tool.

I have delved into this panning/scale anomaly further. The file TEST-ppam50bw.gif is 1920x1080 according to GIMP.

The correct coordinate entry is as follows:

Coord 1
132 -90
133 60
Coord 2
1547 0
821 30

The incorrect coordinate entry which causes pan/scale issues:

Coord1
132 -90
133 60
Coord2
1547 30
821 0

I found that when I used a "Chart Group" titled "None" that has just a few ENC charts

[because we are prevented from having a chart group with 'no charts' which uses just simple 'Gshhs Base' - (I think opencpn should make this an automatic chart group that we can select)]

to speed things up, but I could not Zoom out as far (I think this is an issue with Opencpn, perhaps) but that is the chart setup where this anomaly occurs. There are two conditions which much be met for it to occur:
1. Chart Group is set to "None" which is a few ENC charts.
2. Incorrect Coordinates are entered as above.
What Happens: When panning or scaling the WxFx dances across the globe (at a lesser rate than the pan) until the fax disappears when you have panned too far.
Also there is a lot of High CPU use particularly in MSVC Debug mode. The WxFx plugin never had a "break" and was pretty stable all the way through the entire exercise. I could hit edit, change coords to be correct, or not correct and Finish, run tests, and then hit Edit, etc. Never failed. All this was done with Opencpn OFF.

When I turn on Chart Group CM93 two things happen:
1. The pan/scale antics stop and the WxFx file above stays in the right place.
2. I can zoom out further and faster.

Forgive me, but I have screen shots too, in sequence. The titles of each png should be explanatory.

Sean I am going to send you TEST-ppam50bw.gif so it does not get processed by the forum upload. Ok?

I don't know if you can protect users from incorrect data entry! Aren't there too many cases to test? I think this error stemmed from switching from lat/long to long/lat and my previous scribbles were from before that.

Perhaps we need a User beware note in the Plugin writeup, similar to what I wrote earlier? In any case, if there are some broad coordinate data validation tests that can be done, that would be nice. (I expect you are already doing that though.)

The final screen shot shows the transparency issue which seems to persist.

Hope this helps. Sending the file as attachment to your email next.
Rick

[LeaseOnLife]

Originally Posted by rgleason

Dear Sean,
The fax is real from NWS/NCEP for Feb 11. We had tremendous storms and 85 knots across large parts of the Atlantic.

naaaa, this is a 500mbar chart, showing wind speeds at ~5500m height in the atmosphere, 85knots up there is not uncommon.

[rgleason]

Yes that's right 500 m. --I thought of that. Sorry. We did have bad weather in Atlantic however. Good for Coriolis (Vorticity) winds at mid level atmosphere, which helps determine lower surface level.

Sean, the 1547 value was filled in automatically when select coords graphically.

[jackdale]

I was away for a couple of days. What is the latest version?

[rgleason]

See the wxfx.dll attached to Post #302. I believe that is the most recent. Sean has been working on wx-routing, lately I believe, but he may have an update for wxfx in a little while.
http://www.cruisersforum.com/forums/...ml#post1189599

Sean: It seems to me that some of the functions for obtaining weather faxes via the internet are similar to those needed for download of gribs, so perhaps that code can be used in both Gribs and WxFx or similar.

Internet Download: Saving, deleting, moving in the list, a url with text descriptor, in a text file or in the opencpn.ini, and then executing a download to a certain disk location (and remembering that location) are similar. (when close enough for wifi or via cellphone data -internet)

Email: I don't know if Wxfx has an email server like saildocs, and others. (They're are probably other email servers too.) Maybe in future users can add such services. (via Internet or SSB)

Electromagnetic Waves: MF Navtex 510 and 530 kHz, for example. Global Maritime Distress Safety System 518 kHz and 490 kHz for digital text broadcast. Sean has developed these functions in Linux.. Can be done externally with Windows PC software radio. Various programs. (via SSB)

[seandepagnier]

I realize there is some bug with transparency on windows, but I cannot reproduce it, and therefore I am waiting for someone on windows who cares and who can debug the source to fix it. It doesn't seem to be a complete blocker and the plugin still works.

As far as the capture functionality goes.. I did not implement anything with the usb stick, that is for future versions.. it is just a quick hack on linux to read directly from the sound card.

As for the floating bug.. also cannot reproduce and it is a problem only for wrong coordinates anyway.

Anything else before I declare the plugin is ready for general use?

Of course I will continue to develop it.. programs are never done.

[rgleason]

Sean, the transparency issue in the first screen does not seem to fail or break. What techniques would you suggest using to identify it? Do you have any suggestion what section of code it might be? (Maybe something in the first screen?)

I agree with you about floating wxfx...---> User input error. Get your coordinates right.

Yes, I think you should declare it ready. THANKYOU. Great job.

Do you want me to compile anything?

[ChuckSK]

Originally Posted by boat_alexandra

Anything else before I declare the plugin is ready for general use?

There is an Assert error in weatherfaxwizard.cpp:54 when loading a WAV file.

[seandepagnier]

I think I must build wxwidgets from source with the correct cflags then I can catch these myself. Easy to fix.. but thanks for reporting (fixed in git)

I guess I'll wait another day now.

[rgleason]

Sean, do I have correct cflags setup? How would I determine this?

[ChuckSK]

There is another Asssert error in weatherfaxwizard.cpp:160. These errors mask others following.

[seandepagnier]

ok... getting closer. I'm building wx 2.9 with debugging so I should get asserts too soon (I've been using wx release all this time) so I won't tire the helpful testers with these trivial asserts.

Anyway.. I fixed this one as well in git.. should work now

[rgleason]

Sean I think you nailed it!
Much much better. I like the lat/long again. Thank.

(Why do I have to delete the old dlls in build/plugins/debug to get a new one with today's current date to appear? Is it just over-writting into the existing file and not changing the file date? Never noticed that before.)

There were no breaks. I was able to do polar files I had breaks and trouble with. I was able to do polar files that I had not done before and I am reasonably confident that this is really working better. In debug mode it was smoother and more responsive and the zoom in zoom out problems seem better too!

I am pretty sure the attached release DLL will be 98% good.
By the way, I found that the transparency thing on the first screen goes away when you just grab the window and shake it a little, then everything is fine! So shake the window when you first load a file.

Test this file out by taking the doc of the end and placing in your plugins directory, after deleting your old weatherfax_pi.dll