On Thu, 6 Jun 2013, Matteo Semplice wrote:

> On 05/06/2013 16:56, Roy Stogner wrote:
>>
>> If you're seeing a segfault, the first thing you should do is switch
>> from opt to debug mode.  Then typically you'll fail an informative
>> assertion before you ever get to the segfault point, and even in the
>> worst case at least you can start things up in a debugger and have
>> more context to step through.
>>
> Hi, I am afraid I don't seem to hit any assertion failure... The
> backtrace at the bottom of my email was obtained with the program
> compiled in debug mode.

This bit stands out!

> /home/matteo/libmesh-0.7.2/libmesh/include/mesh/vtk_io.h, line 153,

libMesh 0.7.2 is a year and a half (and about 3 releases) old, and
although I don't recall any bugfixes since then affecting FEMSystem
quadrature, I wouldn't swear to that.  And if there are any *unfixed*
bugs affecting FEMSystem quadrature, it'll be easier for others to
help you if you're using 0.9.1 or the git trunk.

I'll pull down a 0.7.2 copy and see if I can see anything obvious
there, though.

Strange.  On line 575 it's not actually a quadrature access that's
segfaulting; it looks like it's an invalid pointer to one of the FE
objects.  You didn't clear or swap or anything to those, did you?

> Anyway, here it is again. But perhaps I should change my question to
> "what is the proper way to change the quadrature rule in a
> FEMSystem?"

Three steps:

Delete the current element_qrule (and/or side_qrule)

Set it (or them) to the address of new-allocated QBase(s) of your
choice.

Loop over element_fe (and/or side_fe) and attach_quadrature_rule() to
each.

I don't know if this has been tested, though.
---
Roy

------------------------------------------------------------------------------
How ServiceNow helps IT people transform IT departments:
1. A cloud service to automate IT design, transition and operations
2. Dashboards that offer high-level views of enterprise services
3. A single system of record for all IT processes
http://p.sf.net/sfu/servicenow-d2d-j
_______________________________________________
Libmesh-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/libmesh-users

Reply via email to