c++/4966: [3.2/3.3/3.4 regression] g++ generates incorrect debugging information

[Christian Ehrhardt]

The following reply was made to PR c++/4966; it has been noted by GNATS.

>From ehrhardt@mathematik.uni-ulm.de  Mon Jan 13 00:56:03 2003
From: "Christian Ehrhardt" <ehrhardt@mathematik.uni-ulm.de>
To: gcc-gnats@gcc.gnu.org, gcc-prs@gcc.gnu.org,
  meixner@rbg.informatik.tu-darmstadt.de, rodrigc@gcc.gnu.org,
  gcc-bugs@gcc.gnu.org, nobody@gcc.gnu.org
Cc:  
Subject: Re: c++/4966: [3.2/3.3/3.4 regression] g++ generates incorrect debugging information
Date: Sat, 11 Jan 2003 02:38:22 +0100

 It seems that g++ emits two identical copies of the constructor
 for B from the same source lines. When gdb sets the breakpoint
 at the specified line it chooses one of those copies but gcc happens
 to call the other copy. Thus the breakpoint is never reached.
 This means that this is probably a gdb bug (gdb should set two break
 points) but someone should verify that the debugging information
 is really ok.
 
     regards  Christian
 
 -- 
 THAT'S ALL FOLKS!