Crash - regression from R20 to R27

[GoogleCodeExporter]

What steps will reproduce the problem?
1. Open the supplied v12_extract.ifc file
2. Crash with IFCPP R27, not with R20 (R24 modified RepresentationConverter.cpp 
could be the cause?)

What is the expected output? What do you see instead?

IFCPPSimpleViewer R20 didn't crash (but didn't handle an opening), R27 crashes.

What version of the product are you using? On what operating system?

Using the github repo based on R27.

Original issue reported on code.google.com by sylvain....@vtreem.com on 22 Aug 2014 at 9:05

Attachments:

• v12_extract.ifc

[GoogleCodeExporter]

The first entity in the file causes a problem, because it is invalid (missing 
arguments):
#4= IFCPOSTALADDRESS(.USERDEFINED.,$,'x',$,('x'));

There's another postal address entity in the file which is valid:
#59= IFCPOSTALADDRESS($,'','','',('','',''),'','','','','');

But does the application really crash? I get an error message, but not a crash.
I just added a line in IfcPPReaderSTEP that fixes the number of arguments in 
case they are missing.

Original comment by fabian.g...@gmail.com on 22 Aug 2014 at 11:07

[GoogleCodeExporter]

Thanks for the check for the number of arguments.
R27 did crash, R18 doesn't anymore.
But with R28, windows 2888 & 3803 don't cut the wall 1878.

Original comment by sylvain....@vtreem.com on 22 Aug 2014 at 1:44

[GoogleCodeExporter]

The missing arguments do not cause a crash, only an exception. So it is not the 
reason for the crash, also I cannot reproduce a crash.
If it is not reproducable, I can only close this issue.
The CSG problem on entity 1878 is probably the same as issue 18. It is related 
to carve internals and I can't spend time on it at the moment, so issue 18 will 
remain open for now.

Original comment by fabian.g...@gmail.com on 22 Aug 2014 at 1:56

[GoogleCodeExporter]

Here is another example showing the openings problem. R20 opens it correctly, 
but R27 only shows the wall with no opening cut. Carve didn't change between 
these 2 revisions, I'd rather suspect R24 which modified 
RepresentationConverter::subtractOpenings() ?

Original comment by sylvain....@vtreem.com on 22 Aug 2014 at 2:08

Attachments:

• 001-06_extract.ifc

[GoogleCodeExporter]

I get the exact same results with older version and current version. Also Tekla 
Bimsight shows the exact same wall with a window...

Original comment by fabian.g...@gmail.com on 22 Aug 2014 at 3:27

[GoogleCodeExporter]

Thanks for testing this. Knowing that R28 was working on WIN32 lead me to think 
of some cross-platform bug. R29 confirmed this and fixes the bug :-) 

Original comment by sylvain....@vtreem.com on 23 Aug 2014 at 9:48

[GoogleCodeExporter]

Original comment by fabian.g...@gmail.com on 23 Aug 2014 at 10:03

• Changed state: Fixed