Travis build failure

[sgrogan]

Ah, thanks for the information.

What should the "rule" be? I see three cases;
FreeCADcmd, no gui,I think the test should fail as it did.
FreeCAD -c, console mode, but the gui stuff has been built, I think the test should pass here, I think this is how Travis is set up, less the "packaging" issue.
FreeCAD, full gui, maybe with something like this https://forum.freecadweb.org/viewtopic.php?f=4&t=42264 we could test the gui as well.

[vocx]

...
FreeCAD -c, console mode, but the gui stuff has been built, I think the test should pass here, ...

I think the test should fail here. We are telling FreeCAD we want console mode, so it should turn off the GUI somehow, if that is at all possible. Otherwise this -c option makes no sense, and we should only have two options, FreeCAD or FreeCADCmd.

[wmayer]

FreeCADcmd, no gui,I think the test should fail as it did.
FreeCAD -c, console mode, but the gui stuff has been built, I think the test should pass here, I think this is how Travis is set up, less the "packaging" issue.

FreeCADCmd and FreeCAD -c from a user point of view is exactly the same. So, if for the one case FreeCADGui shouldn't be allowed then for the other case it shouldn't be allowed either.
The technical difference between both versions is that FreeCADCmd doesn't link against any GUI libs so that it can be used on pure server systems where no GUI stuff is installed.

FreeCAD, full gui, maybe with something like this https://forum.freecadweb.org/viewtopic.php?f=4&t=42264 we could test the gui as well.

When doing this then it should be done additionally to the console tests.

[sgrogan]

When doing this then it should be done additionally to the console tests.

Agreed.

What do think about disabling SEH for the Travis builds. I can't find a way to return an error if SEH catches a segfault, so a test will time out and the rest of the tests will be skipped. So for Travis it may be better to let the app crash. If we ever deploy for win from Travis this would change.

[wmayer]

When doing this then it should be done additionally to the console tests.

Agreed.

What do think about disabling SEH for the Travis builds. I can't find a way to return an error if SEH catches a segfault, so a test will time out and the rest of the tests will be skipped. So for Travis it may be better to let the app crash. If we ever deploy for win from Travis this would change.

Yes, for the travis tests SEH can be switched off.

[triplus]

Reading the discussion i am not sure if any other issues need to be resolved ATM, looking at the logs, the GCC caching issue wasn't resolved yet. Hence lets try to do that:

https://forum.freecadweb.org/viewtopic.php?f=10&t=42615

[wmayer]

Any idea what happened to the Windows build that it again fails?

C:\Users\travis\build\FreeCAD\FreeCAD> % Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 606 0 606 0 0 606 0 --:--:-- --:--:-- --:--:-- 1549
100 5291k 100 5291k 0 0 5291k 0 0:00:01 --:--:-- 0:00:01 5291k
No output has been received in the last 10m0s, this potentially indicates a stalled build or something wrong with the build itself.
Check the details on how to adjust your build configuration on: https://docs.travis-ci.com/user/common- ... s-received
The build has been terminated

Apparently an operation takes more than 10min without output and thus it gets cancelled by the system but I wonder which operation exactly?

EDIT:
See also: https://forum.freecadweb.org/viewtopic. ... 10#p367206

[triplus]

As said in the other thread, currently Travis Windows jobs will hang and timeout, whenever cmd.exe /C is used. Not much we can do about it, something for Travis people to sort out.

[triplus]

Windows builds are back.