[Bug tree-optimization/17509] [4.0 regression] ICE in verify_cgraph_node

["bangerth at dealii dot org" <gcc-bugzilla at gcc dot gnu dot org>]

------- Additional Comments From bangerth at dealii dot org  2004-09-16 12:44 -------
> We use error to report problems, and internal_error to make the compiler die  
> if we also found some problems.  If you call internal_error for the first  
> problem you find, you cannot check the rest of the graph/funcion/whatever for  
> more errors. 
 
Then we need a third function. All lines starting with "error:" and "warning:" have traditionally 
always indicated problems in the source code that require attention by the programmer. 
Internal compiler errors are clearly something different and should be marked as such, 
to avoid confusion. 
 
For a message such as 
   x.i:11702: error: Inlined_to pointer is wrong  
I am almost certain that we will eventually get a PR about confusing diagnostics with 
people wondering why the compiler is trying to second-guess the values of variables 
they haven't even in their source code... 
 
W. 

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17509