Boost logo

Boost :

From: jsiek_at_[hidden]
Date: 2000-10-05 16:58:32

Hi John,

I feel your pain. I've also been running into internal compiler errors
using VC++. Often times these errors are quite mysterious and hard to
track down... I sometimes resort to binary searches... #ifdef'ing out
portions of the source code.

I'll get back to you tomorrow on this one.



P.S. I think VC++ should be used as a prime example in the Justice
  Dept. case against Microsoft. How else (other than monopolistic
  pressure) could such a bad product also be the most popular?

John Britton writes:
> > If one takes libs/graph/examples/connected_components.cpp, and
> > encloses the main() function in a namespace, the VC++ compiler fails...
> This problem is driving me nuts, and I've discovered it to be more insidious
> than simply having, or not having, a namespace. If one simply edits a
> pristine libs/graph/examples/connected_components.cpp and changes
> int num = connected_components(G, &c[0], get(vertex_color, G),
> dfs_visitor<>() );
> to
> int num = connected_components(G, &c[0] );
> the same VC++ error "C1001: INTERNAL COMPILER ERROR" is reported.
> When examining the code, I was wondering why the first version of
> connected_components() was used, instead of the second one above, which if I
> understand things correctly, should be equivalent. Was it to avoid this VC++
> bug?
> Not having fun today, JohnB

Boost list run by bdawes at, gregod at, cpdaniel at, john at