OpenCPN Version 5 Beta Release 4.99.1426

[Rasbats]

Building a local copy of OpenCPN from yesterdays GitHub master for testing on Linux. This warning when opening the program. Happens each time:

11:11:27 GMT: can't seek on file descriptor 10 (error 22: Invalid argument)
11:11:27 GMT: can't find length of file on file descriptor 10 (error 0: Success)
11:11:27 GMT: can't seek on file descriptor 10 (error 22: Invalid argument)
11:11:27 GMT: can't find length of file on file descriptor 10 (error 0: Success)

Installed to /usr/local/. Ubuntu 16.04.03, 32bit. Does not stop OpenCPN. Seems to work normally.

Probably my system but reporting it here in case anyone else has seen this.
Mike

[did-g]

Originally Posted by Rasbats

Building a local copy of OpenCPN from yesterdays GitHub master for testing on Linux. This warning when opening the program. Happens each time:

11:11:27 GMT: can't seek on file descriptor 10 (error 22: Invalid argument)
11:11:27 GMT: can't find length of file on file descriptor 10 (error 0: Success)
11:11:27 GMT: can't seek on file descriptor 10 (error 22: Invalid argument)
11:11:27 GMT: can't find length of file on file descriptor 10 (error 0: Success)

Installed to /usr/local/. Ubuntu 16.04.03, 32bit. Does not stop OpenCPN. Seems to work normally.

Probably my system but reporting it here in case anyone else has seen this.
Mike

Are you using wxWidgets 3.1 ?


On linux file under /proc/ aren't seekable and 3.1 is complaining about it while testing for file 'seekability'.


Too bad O is using /proc/meminfo a lot and it makes 3.1 unusable on Linux.


Must be fixed in wx.

[Rasbats]

Should have added this from the log file:

2:24:28 GMT: Metrics:  m_display_size_mm: 338     wxDisplaySize:  1280:1024   
12:24:28 GMT: Error: can't seek on file descriptor 20 (error 22: Invalid argument)
12:24:28 GMT: Error: can't find length of file on file descriptor 20 (error 0: Success)
12:24:29 GMT: Metrics:  m_display_size_mm: 338     wxDisplaySize:  1280:1024   
12:24:29 GMT: Error: can't seek on file descriptor 20 (error 22: Invalid argument)
12:24:29 GMT: Error: can't find length of file on file descriptor 20 (error 0: Success)
12:24:29 GMT: Error: can't seek on file descriptor 20 (error 22: Invalid argument)
12:24:29 GMT: Error: can't find length of file on file descriptor 20 (error 0: Success)
12:24:29 GMT: Error: can't seek on file descriptor 20 (error 22: Invalid argument)
12:24:29 GMT: Error: can't find length of file on file descriptor 20 (error 0: Success)
12:24:29 GMT: Error: can't seek on file descriptor 20 (error 22: Invalid argument)
12:24:29 GMT: Error: can't find length of file on file descriptor 20 (error 0: Success)
12:24:29 GMT: Error: can't seek on file descriptor 20 (error 22: Invalid argument)
12:24:29 GMT: Error: can't find length of file on file descriptor 20 (error 0: Success)
12:24:29 GMT: Error: can't seek on file descriptor 20 (error 22: Invalid argument)
12:24:29 GMT: Error: can't find length of file on file descriptor 20 (error 0: Success)
12:24:29 GMT: Error: can't seek on file descriptor 20 (error 22: Invalid argument)
12:24:29 GMT: Error: can't find length of file on file descriptor 20 (error 0: Success)

And in other places as well. wxWidgets 3.1.2

Mike

[fgd3]

I have experienced several Windows crashes (blue screen) while running OpenCPN 4.99.1426+c1bad7c on my Toshiba Satellite running Windows 7. I first noticed the problem while I was exploring some charts of Cuba. The most recent crash occurred with NOAA charts only. Maybe there is a problem with vector charts, maybe it is a function of how long the program has run. Maybe it's a completely unrelated Windows problem, but so far it only happens when running OpenCPN, so I suspect there's a connection. I'm attaching a log file with every session of the 1426 build as well as the two Windows dump files from the latest crash.


I am using OpenGL if that makes any difference.


Fabbian

[Baikal]

Here there is PhotoLayer, vfkaps plugins for overlay small raster to vector!
Why spoil MBTiles which was designed to store a large amount of information and cover large areas!
This format should not be used as an overlay, but as an independent source, and satellite maps are more accurate and visible than vector maps. Combine satellite maps and vector in two canvases and get the maximum benefit.
I do not understand!
They killed the possibility of using only satellite maps of large coverage from different sets of sources

[skipperearly]

W10 O4.99-1428
degration: when waypoint is named e.g. @~~Vigo the next tide values is not shown anymore in the route details - this was a very useful, however somehow not promoted feature in previous versions. Please enable linking to tides again.

[bdbcat]

Baikal...


1. "I made the whole world of satellite maps up to scale 9, the volume is only 650mb, it was beautiful and clear. Now this map is not displayed."
If you could provide a link, I'll check this file. It should display...

2. "Maps now do not apply a display scale and are always shown. This is a very heavy load on the PC."
Tiles display if zoomed in far enough, relative to their advertised scale. Zoom out, and the tile will disappear.



3. "Previously, I could place a low resolution map under a high resolution map. Now only the base map is under it and the tile cutting grid has begun to shine through."
I cannot see how a photo-tiles-only display could in any way be suitable for navigation. No NAVAIDS, no depth soundings, no text.... Normal use would suggest having a useful cartographic chart as a backing image.


4. "Need a guide on the proper preparation of MBTiles. Similar to the ENC cells and their scale."

A good idea, and I agree. We also need to reasonably handle the existing base of MBTiles sets widely available on the net today.


Thanks for the comments
Dave

[fgd3]

Here's more information about the crashes I've been experiencing on my Windows 7 Toshiba Satellite with OpenCPN 4.99.1426+c1bad7c.


In the 08:40:09 session I started OpenCPN with my home raster chart and let it sit for almost ten minutes. Then I exited the program. No crash.


In the 08:50:35 session I started OpenCPN, switched to the vector chart of the same area. I didn't move the boat, scroll, pan, or anything else. After about a minute the system crashed. The Windows crash files I've attached are from that crash.


In the 09:04:18 session I started OpenCPN, turned off OpenGL, and exited the program.


In the 09:04:52 session I started OpenCPN, switched to the vector chart with the same area as my home chart, and let it sit for 32 minutes. No crash.


It seems pretty clear the problem has to do with using vector charts when OpenGL is selected. You may not want to pursue the issue since Windows 7 is pretty old and probably includes a buggy version of OpenGL. Switching off OpenGL seems to be a satisfactory workaround for me.



Fabbian

[Ptizef]

Hi All,
About compiling on Windows 10, am a little bit confused.
In the manual it is said that we should use wxWidgets-3.1.2 and v141 but what is coming from sources and buildwin-4.99 is version 3.1.1 and v140 for the DLLs. The last beta is also based on 3.1.1.
Do I have to revert back to 3.1.1 to be at the same level as the current development.
Thanks for your help
JP Joubert

[bdbcat]

jp...


Windows builds should be using wx3.1.2.


What says your logfile?

08:40:09: wxWidgets version: wxWidgets 3.1.2 Microsoft Windows NT 64 bit wxMSW

Dave

[stelian]

Originally Posted by nohal

Looking at your crashes, the first one does not even seem to be in wx, but rather lower, maybe in Wayland's X11 emulation layer.
The other one seems to be in wmm plugin, probably in https://github.com/OpenCPN/OpenCPN/b...mm_pi.cpp#L593 (Never seen this crash before) and I would first try to disable it as well to se how far you can get.

When disabling the wmm plugin, there is no crash anymore, but OpenCPN hangs instead with a partially rendered window (see attachement):
#0 0x00007ffff6292af6 in wxStatusBar::GetFieldRect(int, wxRect&) const ()
at /usr/lib/x86_64-linux-gnu/libwx_gtk3u_core-3.0.so.0
#1 0x00007ffff629257e in wxStatusBar:rawField(wxDC&, int, int) ()
at /usr/lib/x86_64-linux-gnu/libwx_gtk3u_core-3.0.so.0
#2 0x00007fffb9d94536 in wxStatusBar::OnPaint(wxPaintEvent&) ()
at /usr/lib/x86_64-linux-gnu/libwx_gtk2u_core-3.0.so.0
#3 0x00007ffff5d3b4be in wxEvtHandler::ProcessEventIfMatchesId(wxEventTable EntryBase const&, wxEvtHandler*, wxEvent&) () at /usr/lib/x86_64-linux-gnu/libwx_baseu-3.0.so.0
#4 0x00007ffff5d3b5c3 in wxEventHashTable::HandleEvent(wxEvent&, wxEvtHandler*) ()
at /usr/lib/x86_64-linux-gnu/libwx_baseu-3.0.so.0
#5 0x00007ffff5d3b98b in wxEvtHandler::TryHereOnly(wxEvent&) ()
at /usr/lib/x86_64-linux-gnu/libwx_baseu-3.0.so.0
#6 0x00007ffff5d3ba13 in wxEvtHandler::ProcessEventLocally(wxEvent&) ()
at /usr/lib/x86_64-linux-gnu/libwx_baseu-3.0.so.0
#7 0x00007ffff5d3ba75 in wxEvtHandler::ProcessEvent(wxEvent&) ()
at /usr/lib/x86_64-linux-gnu/libwx_baseu-3.0.so.0
#8 0x00007ffff5d3b7d7 in wxEvtHandler::SafelyProcessEvent(wxEvent&) ()
at /usr/lib/x86_64-linux-gnu/libwx_baseu-3.0.so.0
#9 0x00007ffff626ae8f in wxWindow::GTKSendPaintEvents(_cairo*) ()
at /usr/lib/x86_64-linux-gnu/libwx_gtk3u_core-3.0.so.0

...
#37 0x00005555557e633e in ChartCanvas::SetCursorStatus(double, double)
(this=0x5555565d8000, cursor_lat=48.813038540262269, cursor_lon=-3.4411027639623879)
at /home/stelian/TMP/OpenCPN_stelian.git/src/chcanv.cpp:4034
#38 0x00005555557e6dc8 in ChartCanvas::OnCursorTrackTimerEvent(wxTimerEvent& )
(this=0x5555565d8000, event=...) at /home/stelian/TMP/OpenCPN_stelian.git/src/chcanv.cpp:4016
...

Originally Posted by nohal

I have done a little experiment on my machine, removed the the WX libs linked against gtk2, installed the ones linked against gtk3, built opencpn (nothing special, cmake -DOCPN_FORCE_GTK3 ..;make) and the product works as expected for me, so if I didn't see your backtraces, I would say it simply has no problem with GTK3 and WX as delivered on the latest LTS Ubuntu...
Pavel

Ok, if it works on Bionic I'll assume the WX/GTK3/Wayland libs are buggy on Cosmic. So I'll remain for now with wx 3.0.2 / gtk2 which works ok.


So, if we go back to the original report, I have problems detecting my monitor physical width because of the commit dd12fde0. The detection is resolved if using GTK3 (either using the new or the old API). So I guess the real question is are there more users of hidpi screens or more users of multi-monitor setups ? (I don't even think of those who have mixed hidpi multi-monitor setups...)


Stelian.

[nohal]

Originally Posted by stelian

When disabling the wmm plugin, there is no crash anymore, but OpenCPN hangs instead with a partially rendered window (see attachement):
#0 0x00007ffff6292af6 in wxStatusBar::GetFieldRect(int, wxRect&) const ()
at /usr/lib/x86_64-linux-gnu/libwx_gtk3u_core-3.0.so.0
#1 0x00007ffff629257e in wxStatusBar:rawField(wxDC&, int, int) ()
at /usr/lib/x86_64-linux-gnu/libwx_gtk3u_core-3.0.so.0
#2 0x00007fffb9d94536 in wxStatusBar::OnPaint(wxPaintEvent&) ()
at /usr/lib/x86_64-linux-gnu/libwx_gtk2u_core-3.0.so.0
#3 0x00007ffff5d3b4be in wxEvtHandler::ProcessEventIfMatchesId(wxEventTable EntryBase const&, wxEvtHandler*, wxEvent&) () at /usr/lib/x86_64-linux-gnu/libwx_baseu-3.0.so.0
#4 0x00007ffff5d3b5c3 in wxEventHashTable::HandleEvent(wxEvent&, wxEvtHandler*) ()
at /usr/lib/x86_64-linux-gnu/libwx_baseu-3.0.so.0
#5 0x00007ffff5d3b98b in wxEvtHandler::TryHereOnly(wxEvent&) ()
at /usr/lib/x86_64-linux-gnu/libwx_baseu-3.0.so.0
#6 0x00007ffff5d3ba13 in wxEvtHandler::ProcessEventLocally(wxEvent&) ()
at /usr/lib/x86_64-linux-gnu/libwx_baseu-3.0.so.0
#7 0x00007ffff5d3ba75 in wxEvtHandler::ProcessEvent(wxEvent&) ()
at /usr/lib/x86_64-linux-gnu/libwx_baseu-3.0.so.0
#8 0x00007ffff5d3b7d7 in wxEvtHandler::SafelyProcessEvent(wxEvent&) ()
at /usr/lib/x86_64-linux-gnu/libwx_baseu-3.0.so.0
#9 0x00007ffff626ae8f in wxWindow::GTKSendPaintEvents(_cairo*) ()
at /usr/lib/x86_64-linux-gnu/libwx_gtk3u_core-3.0.so.0

...
#37 0x00005555557e633e in ChartCanvas::SetCursorStatus(double, double)
(this=0x5555565d8000, cursor_lat=48.813038540262269, cursor_lon=-3.4411027639623879)
at /home/stelian/TMP/OpenCPN_stelian.git/src/chcanv.cpp:4034
#38 0x00005555557e6dc8 in ChartCanvas::OnCursorTrackTimerEvent(wxTimerEvent& )
(this=0x5555565d8000, event=...) at /home/stelian/TMP/OpenCPN_stelian.git/src/chcanv.cpp:4016
...





Ok, if it works on Bionic I'll assume the WX/GTK3/Wayland libs are buggy on Cosmic. So I'll remain for now with wx 3.0.2 / gtk2 which works ok.


So, if we go back to the original report, I have problems detecting my monitor physical width because of the commit dd12fde0. The detection is resolved if using GTK3 (either using the new or the old API). So I guess the real question is are there more users of hidpi screens or more users of multi-monitor setups ? (I don't even think of those who have mixed hidpi multi-monitor setups...)


Stelian.

Stelian...
I wonder where the wx 3.0.2 libs come from - https://packages.ubuntu.com/search?s...searchon=names says it should also be 3.0.4 like in bionic...

It would be interesting to see what the pixel resolution reported for multi-monitor systems actually is. If it was the total, I would say we can easily assume that ratio closer to 32:9 than to 16:9 is a multi-monitor setup and pick the correct, or better say less broken, method...

Pavel

[Baikal]

Originally Posted by bdbcat

Baikal...
1. "I made the whole world of satellite maps up to scale 9, the volume is only 650mb, it was beautiful and clear. Now this map is not displayed."
If you could provide a link, I'll check this file. It should display...
2. "Maps now do not apply a display scale and are always shown. This is a very heavy load on the PC."
Tiles display if zoomed in far enough, relative to their advertised scale. Zoom out, and the tile will disappear.
3. "Previously, I could place a low resolution map under a high resolution map. Now only the base map is under it and the tile cutting grid has begun to shine through."
I cannot see how a photo-tiles-only display could in any way be suitable for navigation. No NAVAIDS, no depth soundings, no text.... Normal use would suggest having a useful cartographic chart as a backing image.
4. "Need a guide on the proper preparation of MBTiles. Similar to the ENC cells and their scale."
A good idea, and I agree. We also need to reasonably handle the existing base of MBTiles sets widely available on the net today.

Thanks for the comments
Dave

1. If you could provide a link, I'll check this file. It should display ...

Under the link all my satellite maps involved in the testing.
https://www.dropbox.com/s/5imxrprs43b8grn/SAT.zip?dl=1

2. Tiles display if zoomed in far enough, relative to their advertised scale. Zoom out, and the tile will disappear.
3. I cannot see how a photo-tiles-only display could in any way be suitable for navigation. No NAVAIDS, no depth soundings, no text.... Normal use would suggest having a useful cartographic chart as a backing image.

I recorded a video of how everything worked well on version 1426 and how it became in 1428.
https://www.dropbox.com/s/jsy7nll21l...Tiles.mp4?dl=0
I note that I use satellite maps as a separate group of maps. Split into two canvases where to the left are satellite maps on the right ENC. On high-quality satellite images you can clearly see stones, underwater destroyed obstacles to shallow waters and coastal structures. We need experience for the interpretation of information on satellite images. In any case, the two types of charts complement each other.
And if you can implement a hybrid overlay of information from ENC to MBTiles with enabling and disabling this function in the chart Options, it will be just great!

You can leave as is now.
But if you use MBTiles as one main source in a selected charts groups, then the wish to make it functional as in version 1426

[stelian]

Pavel,

Originally Posted by nohal

Stelian...
I wonder where the wx 3.0.2 libs come from - https://packages.ubuntu.com/search?s...searchon=names says it should also be 3.0.4 like in bionic...

Sorry, it was a typo. I have the standard Ubuntu 3.0.4 libwx libs.


Stelian.

[cagney]

Left-clicking on a quilted MBTile chart, in the piano bar, hides the chart. Using a separate MBTile chart group.

Surely a mistake??

Thomas