OpenCPN Version 2.1.0 Beta

[bdbcat]

Ptizef (Jean Pierre)...

I see the problem where sometimes the quilt bar does not match the display. Solution found.

There is a new chcanv.cpp in CVS now. Please update , recompile, and should be improved.

Thanks
Dave

[Ptizef]

Hi Dave

I continue testing ...
I get frequently a crash when clicking alternatively on quilt control bar : see shot attached

regards
Jean pierre

[Ptizef]

Hi Dave

I compiled and tested with the new chcanv.cpp
It's better and quilt bar is updated according with the display , but there is still a problem of speed . The screen moving is very slow , not smooth , frequently hanging .. updating are frequently offset ed by severals seconds .
Furthermore , I didn't understand well what is the logic of quilting in the shot attached :
It's shown an overzoom whereas there are bigger scale charts available but not shown

I hope this will help
regards
Jean Pierre

[MarkJ]

Originally Posted by MarkJ

OK starting on a 25 mile run in 1 1/2 hours.

I'll tell you if it keeps us in the water or runs us over terra firma

We're alive! The program must work!

Mind youy I was thinking of blaiming you for our lousy tie up!

[bdbcat]

JP...

Silly questions:
1. On your post 347: Is this crash happening using the newest chcanv.cpp (today)?
2. On 348: what happens if you click one of the larger scale (dark blue) bars? Nothing?

Thanks
Dave

[Ptizef]

Dave
post 347 ; It happens BEFORE changing chcanv.cpp . I can't reproduce it after. The only crash I got since is (systematically) when I try to rebuilt the chart database 'shot attached)

post 348 : If I click on the larger scale bars , the charts are quilted and displayed normally . In fact in most case I have to click on the largest scale bar to get all bsb charts quilted at every zoom scale .

Jean Pierre

[champ8242]

Hi Dave,

Hope this one helps. (just updated from CVS)

[cagney]

Dave
A couple of crashes, while testing the "Chicago Syndrome".
This in itself seems solved. A side effect of the present quilting algo is that, if using ENC and zooming out all ENCs in view not yet initialized, starts their SENC conversion.....
Could come as a surprise to users in certain situations

To do these tests, I loaded all US ENCs. First crash when zoomed out from Chicago in CM93, and tried to zoom in on Carib rasters that did not show up.
Second crash came when trying rebuilding the existing chart database = CM93 + all US ENCS + ~ 30 Carib raster charts.
I'm using CVS from today.

Thomas

[bdbcat]

Thomas...

The faults you report are very mysterious. It is a problem deep inside the gtk library code, caused I think by bad data sent to to the subroutine.

I cannot reproduce.

So, I have built a special version of cm93.cpp to catch this thing. It catches the fault, and then creates a log entry regarding the data that would have caused a crash, and then exits cleanly.

If there is no fault in the code, then opencpn operates as normal.

I have uploaded the new cm93.cpp to CVS.

So, if you could update and recompile, play with it for a while, and see it it exits at an unexpected time. If it does, please inspect and post the last bit of the logfile. Maybe we can see what the problem is.

Thanks
Dave

[bdbcat]

Oops...

Fixed in CVS now.

Dave

[bdbcat]

Just to be clear:

Posts 352 and 353 are the same problem, introduced today from Ptizef bug, and fixed by latest chartdb.cpp upload.

cm93.cpp gtk region problem relates (so far) only to Cagney, cf Post 336 shot2. Don't you feel special, Thomas?

Anyway, update from CVS right now should have all the latest.

Dave

[bdbcat]

All...

I really do appreciate our current condition:
1. Linux and Windows debuggers online in Europe right now.
2. Sourceforge (wherever) providing the CVS backend.
3. Forum providing an effective chat function.
4. Strong Wifi for me at anchor in a quiet cove on the Chesapeake enjoying a Sunday afternoon debug session.

Pretty cool, don't you think?

Dave

[manimaul]

Originally Posted by bdbcat

All...

Pretty cool, don't you think?

Dave

There is hope for peace in the world yet.

Thanks, and nice work on the progress on the new quilting mechanism everyone.

[BluegillBoy]

i'm jealous, dude. but glaqd for you.

i did get in about 5 hours cruising today using build 501.many crashes after forcing rebuild, even before i could quit and restart. also when changing charts even without forcing a rebuild. i know you're on that one.

here's something weird... when i click on one of the bars at the bottom representing the various charts available, i LOVE the pop-up window that gives full chart info! but almost always, the infor in the window for the first chart that i click never gets replaced when clicking on other bars. no matter which bar i click on, the chart info stays the same.

but anyway, this is way cool.

thanks

[manimaul]

Originally Posted by bdbcat

b. Choosing a chart on the bar (by clicking) builds a quilt with following properties:
The chart chosen, plus all charts of smaller scale of the same type, projection, and skew are shown in the quilt.
c. Right click menu on the bar allows excluding/including a particular chart from the quilt. Not currently saved in config...
d. Charts appearing in the quilt show brighter color bars, as with single chart mode, except there are several shown at once.
e. Note that a small scale chart in the bar may be available for the quilt, but be eclipsed by larger scale charts. In this case the brighter bar color may be confusing, since the chart is really not visible. We may need a special icon or color pattern to indicate this case.
f. As a consequence of this new algorithm, it is not possible to combine chart types (Raster/Vector/Cm93) on one quilt, with one useful exception.
g. Exception to (f). If the quilt does not cover all of the screen as composed, and cm93 charts are available, then cm93 will be used to fill out the uncovered region of the screen.

possible further refinements:

e...This may be a case for the viz and inviz icons. (I will make these more distinguishable.) It could work as such...

1. Only charts that are actually visible in the window get highlight + visible icon.

2. Eclipsed charts get no highlight (Eclipsed = Any charts not actually visible in the quilt window except manually hidden charts)

3. Manually hidden charts get hidden icon + (highlight or no highlight based on if they would be visible if not hidden)

Other possible refinements:

1. Hovering pointer over chart in chart window highlights corresponding chart in bar.
2. Ability to right-click on actual chart in quilt window and choose "remove this chart from quilt"