Beta Test / Technical

[bdbcat]

Sean/Pavel...

I partially reverted support for c++11, taking Sean's suggestion.

So, please adjust/cancel your pull requests to conform to this new method.
Feel free to use std c++11, if it suits, and add files to the list as necessary.

Sorry for the run-around.

Dave

[Hakan]

W10 VS2013 A Debug issue
I've problem debugging BR24Radar_pi. It seems to hang when updating a menu and I use a click button.
My first trace is it could be connected to this alert I get when I click on O-Tools button when BR24 is not included in the debug.
Any bells?
Thanks Håkan

[transmitterdan]

Originally Posted by bdbcat

I'll implement for ENC. Should be quick and clean.

Comments?
Dave

Dave,

Looked at your commit that hashes the path to create unique S57 file names. I am sure this will work.

I just like to be able to efficiently and unambiguously determine if "cached" data is obsolete. The present code is unambiguous I believe. It's probably not as efficient as it could but I'll stick with the old proverb that says "premature optimization is the root of all evil".

Dan

[Hakan]

In Win10 VS2013 I'm not able to debug OCPN (incl Dashboard & WMM)
It starts to open OCPN but on a white canvas and not yet a tools menu I get a break in appbase.cpp: (Call stack below)
I'm not capable of solving this, anyone else please?

>>>>> Break in appbase.cpp row 1077
// default assert handler
static void
wxDefaultAssertHandler(const wxString& file,
                       int line,
                       const wxString& func,
                       const wxString& cond,
                       const wxString& msg)
{
    // If this option is set, we should abort immediately when assert happens.
    if ( wxSystemOptions::GetOptionInt("exit-on-assert") )
        wxAbort();

    // FIXME MT-unsafe
    static int s_bInAssert = 0;

    wxRecursionGuard guard(s_bInAssert);
    if ( guard.IsInside() )
    {
        // can't use assert here to avoid infinite loops, so just trap
Break--->   wxTrap();   

        return;
    }
<<<<<<<<<<


  Important or not but while looping F11 around it is running this 
  wxendbusycursor and eventually return to the wxTrap() break.
>>>>>>> opencpnplatform.cpp row 1233
       #if wxCHECK_VERSION(2, 9, 0 )
//    if( ::wxIsBusy() )
    {
        ::wxEndBusyCursor();
    }
    #endif
#endif    
}
<<<<

>>>>
    / Restore cursor to normal
void wxEndBusyCursor()
{
    wxCHECK_RET( gs_wxBusyCursorCount > 0,
                 wxT("no matching wxBeginBusyCursor() for wxEndBusyCursor()") );

    if ( --gs_wxBusyCursorCount == 0 )
    {
#ifndef __WXMICROWIN__
        ::SetCursor(gs_wxBusyCursorOld);
#endif
        gs_wxBusyCursorOld = 0;
    }
}
<<<<<<<<<<<

Call stack:
[00] OCPNPlatform::HideBusySpinner c:\builds\ocpn\opencpn\src\ocpnplatform.cpp:1233
[01] cm93chart::SetVPParms c:\builds\ocpn\opencpn\src\cm93.cpp:2249
[02] cm93compchart::PrepareChartScale c:\builds\ocpn\opencpn\src\cm93.cpp:5069
[03] cm93compchart::SetVPParms c:\builds\ocpn\opencpn\src\cm93.cpp:4979
[04] cm93compchart::RenderRegionViewOnGL c:\builds\ocpn\opencpn\src\cm93.cpp:5389
[05] glChartCanvas::RenderQuiltViewGL c:\builds\ocpn\opencpn\src\glchartcanvas.cpp:2762
[06] glChartCanvas::RenderCharts c:\builds\ocpn\opencpn\src\glchartcanvas.cpp:3074
[07] glChartCanvas::Render c:\builds\ocpn\opencpn\src\glchartcanvas.cpp:3637
[08] glChartCanvas::OnPaint c:\builds\ocpn\opencpn\src\glchartcanvas.cpp:1084
[09] wxAppConsoleBase::HandleEvent c:\wxwidgets-3.0.2\src\common\appbase.cpp:612
[10] wxAppConsoleBase::CallEventHandler c:\wxwidgets-3.0.2\src\common\appbase.cpp:624
[11] wxEvtHandler::ProcessEventIfMatchesId c:\wxwidgets-3.0.2\src\common\event.cpp:1394
[12] wxEventHashTable::HandleEvent c:\wxwidgets-3.0.2\src\common\event.cpp:998
[13] wxEvtHandler::TryHereOnly c:\wxwidgets-3.0.2\src\common\event.cpp:1589
[14] wxEvtHandler::TryBeforeAndHere c:\wxwidgets-3.0.2\include\wx\event.h:3671
[15] wxEvtHandler::ProcessEventLocally c:\wxwidgets-3.0.2\src\common\event.cpp:1522
[16] wxEvtHandler::ProcessEvent c:\wxwidgets-3.0.2\src\common\event.cpp:1495
[17] wxEvtHandler::SafelyProcessEvent c:\wxwidgets-3.0.2\src\common\event.cpp:1613
[18] wxWindowBase::HandleWindowEvent c:\wxwidgets-3.0.2\src\common\wincmn.cpp:1526
[19] wxWindow::HandlePaint c:\wxwidgets-3.0.2\src\msw\window.cpp:4857
[20] wxWindow::MSWHandleMessage c:\wxwidgets-3.0.2\src\msw\window.cpp:2856
opencpn.exe has triggered a breakpoint.

[bdbcat]

Hakan...

This is wxWidgets being picky, but we do have some logic issue on busy spinner.

To get by temporarily, make this change to OCPNplatform.cpp:

    #if wxCHECK_VERSION(2, 9, 0 )
    if( ::wxIsBusy() )
    {
        ::wxEndBusyCursor();
    }
    #endif

Remove "//" from line 1231.

I'm off line for a few days hiding from Matthew....

Good Luck
Dave

[Hakan]

Dave
Thanks, that solved it for now.
Happy hiding and please take the strength out of Matthew before he cross the north Atlantic and pass us.
Håkan

[transmitterdan]

In latest git (6 October 2016 8PM EDT) I see a strange opengl problem. When zooming charts from US ENC no issues. But when zooming charts from Inland ACE lots of problems with failure to erase and replace the region outside the chart boundaries.

[rgleason]

Yes Tdan, I get this too. The river rewrites itself multiple times at different scales, etc

[Hakan]

I hope Matthew has passed you all without serious damages. I've seen some movies also over here and realize my earlier not so wise attempt to joke isn't the most clever thing I've said. I sorry if I was misunderstood and it wasn't at all my intention to be flippant.
Håkan

[seandepagnier]

Originally Posted by transmitterdan

In latest git (6 October 2016 8PM EDT) I see a strange opengl problem. When zooming charts from US ENC no issues. But when zooming charts from Inland ACE lots of problems with failure to erase and replace the region outside the chart boundaries.

Maybe these charts store the regions in a different winding direction?

Where did you get these charts? Is this a new problem? How does it work both with and without opengl?

[.Paul.]

Originally Posted by transmitterdan

In latest git (6 October 2016 8PM EDT) I see a strange opengl problem. When zooming charts from US ENC no issues. But when zooming charts from Inland ACE lots of problems with failure to erase and replace the region outside the chart boundaries.

The problem also exists in the official build of OpenCPN 4.4.0 downloaded from the link at opencpn.org. It is not caused by any new code. It is caused by OpenCPN's handling of the 3UABUOYS.000 ENC from USACE. This chart is mostly transparent. When the chart canvas is repainted, and 3UABUOYS is included in the quilt, and 3UABUOYS covers the whole canvas, the code in OpenCPN assumes that the old image on the canvas will be hidden by the image of 3UABUOYS. So, the old image on the canvas is not cleared before repainting as required for the pan or zoom. This works for most charts, but fails if a transparent chart is being painted.

Paul

[transmitterdan]

Paul,

Thanks for the diagnosis. I can check whether there is an easy fix. Maybe someone is already on it.

[.Paul.]

ENC chart 3UABUOYS covers a very large area. The chart border is shown in this screenshot:

In addition to covering most of the IENC charts from USACE, it covers some of the coastal waters of the southeastern US.

Paul

[cagney]

Dave
Coming back to an old bug, that's not really fixed.
Maybe it's a good time now, or I can enter the bug in the tracker.
I have a chart group consisting of nos/geo and BSB2 charts.
When panning around I always get a crash. This particular crash is using the master git, but IIRC the problem has been there all through the 4.* releases.



Thomas

[transmitterdan]

Originally Posted by cagney

Dave
Coming back to an old bug, that's not really fixed.
Maybe it's a good time now, or I can enter the bug in the tracker.
I have a chart group consisting of nos/geo and BSB2 charts.
When panning around I always get a crash. This particular crash is using the master git, but IIRC the problem has been there all through the 4.* releases.

Attachment 133356

Thomas

Thomas,

Can you find a way to share the chart files with me and instructions how to repeat the crash? I'll be happy to fire up a debug version and see if I can trap the crash. I normally debug on Windows but if this is a generic issue (e.g.null pointer) then it will find it right away.

Dan