OMEdit crashes when trying to open a Modelica file

[Christian Kral <dr.christian.kral@…>]

The nightly built version 23549 crashes when I try open a Modelica file. This happens both by clicking on the open icon and choosing Open from the File menu.

If I open a file listed under recent files, OMEdit opens the file without problems.

The error refers to Ubuntu 12.04 32 bit.

[Martin Sjölund]

For me it worked fine using Ubuntu 14.04 64-bit and 12.04 32-bit (Unity and Gnome Classic)...

[Martin Sjölund]

I will try to make omc/OMEdit be built by gcc instead of clang 3.0 to see if that resolves the issues.

I remember the crash happened first on my computer on Friday Nov. 21.

Happens to coincide with OMEdit and omc sharing the same address space. But the stacktrace suggests it is not the issue.

[Christian Kral <dr.christian.kral@…>]

I tried it with

• Gnome Classic
• Unity
• XFCE

and they all lead to a crash of OMEdit.

[Martin Sjölund]

Adeel: When can I get a stacktrace of all active threads in the crash logs?

[Christian Kral <dr.christian.kral@…>]

No more crash in r23584. Opening a Modelica file now works without problems. Thanks a lot for fixing this issue.

[Martin Sjölund]

OK. Then I believe it was an issue with clang 3.0 requiring some flag to disable an optimization. But since I could not reproduce the issue, I can't disable this optimization. Just hope people don't use clang 3.0 to compile omc...

[Christian Kral <dr.christian.kral@…>]

A new issue arouse in r23584. and I am not sure how it is related to this ticket. In r23584 some simulation examples are simulating with no problems. Other simulation examples fail -- but worked with previous versions of OpenModelica. An example of a failing example is
Modelica.Magnetic.FundamentalWave.Examples.BasicMachines.AIMC_DOL
Failing means: the modelica file gets compiled but simulation hangs after a few seconds, so that the progress bar is frozen and the simulation is not completed.

Martin, could you please check if you can reproduce this issue.

[Martin Sjölund]

It's sort of related. gcc 4.6 is used now. And it seems the dassl routine doesn't like the GNU compiler... It's spinning somewhere in the lapack routines of dassl.
On 14.04 it works fine (uses clang). I assume if you change solver method, it works fine in 12.04?

[Christian Kral <dr.christian.kral@…>]

For Modelica.Magnetic.FundamentalWave.Examples.BasicMachines.AIMC_DOL I tried a couple different integration methods (Ubuntu 12.04 32 bit):

• Euler
• Rungekutta
• Radau3
• Radau5
• Lobatto2
• Lobatto4

None of these worked. Three different failure scenarios occurred:

1. The simulation terminated after a couple integration steps
2. The simulation terminated after the switch closed
3. The simulation terminated normally, but the results were wrong (showed DC current instead of AC line current)

[Christian Kral <dr.christian.kral@…>]

I switched from Ubuntu 12.04 to 14.04 in order to avoid this issue. Modelica.Magnetic.FundamentalWave.Examples.BasicMachines.AIMC_DOL now works fine.

I wonder whether this ticket can be closed...