[Rivet] Problem with AGILe and FHerwig

[James Monk]

Hi Adam,

I'd been seeing some strange behaviour from AGILe and elsewhere on my mac.  I think it was because the binary distribution of gfortran that I downloaded came with its own version of libstd++ - I was seeing these kinds of weird malloc errors in other code that had worked fine before the gfortran "upgrade," and AGILe was having weird problems with the streaming operators (see http://projects.hepforge.org/agile/trac/ticket/122 ).  Did you get gfortran from hpc.sourceforge.net ?

In any case, once I'm done with this jet cross section stuff I'll be happy to mess about with your laptop (if you're happy to let me ;)

cheers,

James

On 21 Jun 2010, at 12:51, Adam Davison wrote:

> Hi,
> 
> I've been trying to set up and run FORTRAN Herwig via AGILe on a 64-bit 
> Intel Macbook.
> 
> Every time I run agile-runmc Herwig:6.510 I get a crash with an error like:
> 
> AGILe.Loader: TRACE  Successfully loaded 
> /Users/adamd/local/x86_64/build/genser/herwig/6.510.2/x86_64-Darwin-gcc42-opt/lib/libherwig_dummy.dylib 
> (0x1002d4770)
> python(58746) malloc: *** error for object 0x7fff70bb0500: pointer being 
> freed was not allocated
> 
> I dug into the code and with the aid of some printfs determined that 
> agile is causing this malloc fault in FHerwig::FHerwig() on, of all 
> things, this line:
> 
>     _particleNames[ELECTRON]    = "E-      ";
> 
> Now that seemed rather strange, I tried some stuff, and in the end I get 
> to the point where I have inserted this code above that crash:
> 
>         std::map<int, std::string> b;
> //      b[1] = std::string("hi");
>         b[2] = std::string("hi2");
> 
> Uncommenting the middle line, causes agile-runmc to crash on the second 
> line. All of this makes me think there's some kind of memory corruption, 
> probably heap, given the fact that it's a malloc error ultimately.
> 
> Does anyone have any ideas what could be going wrong here?
> 
> I'm using gcc4.2 from the Apple developer tools, let me know if anything 
> else would help.
> 
> Thanks, any help much appreciated,
> 
> Adam
> _______________________________________________
> Rivet mailing list
> Rivet at projects.hepforge.org
> http://www.hepforge.org/lists/listinfo/rivet