Virtual Base Bug

[Mike Stump]

> To: Mike Stump <mrs@windriver.com>
> From: Alexandre Oliva <oliva@lsd.ic.unicamp.br>
> Date: 08 Nov 1999 16:24:19 -0200

> On Nov  8, 1999, Mike Stump <mrs@windriver.com> wrote:

> >> Date: Mon, 08 Nov 1999 03:26:56 -0800
> >> From: Mike Bird <mgb@yosemite.net>

> >> P.S. A sincere if belated thank you to H.J. Lu for trying to head
> >> this one off at the pass.

> > You misunderstand who introduced the bug then.

> This is not fair.  I've just re-read the thread H.J. Lu pointed to,
> and I got the impression that Lu tried to go back to
> -fno-vtable-thunks before it was too late, but was overruled by
> others.

?  What's not fair?  The facts are, Lu submitted the `fix', and it was
put in.  The facts are p6610a.C in the testsuite alerted anyone that
wanted to know, that introducing this patch would break users
expectations. p6610a.C goes way back, predating the start of the egcs
project even, as well as the introduction of the bug in gcc.  The
facts are that I told Lu within 24 hours after he reported the `bug'
that `Turning on thunks causes this regression. Some users will expect
this testcase to work. The failure mode of this with thunks is
particularly bad as the this pointer will be wrong at times, and that
shift can cause all sorts of very hard to figure out problems.'  These
are the facts.  24 hours is not `too late' in my estimation.  We have
lived with the bug for over two years, I don't buy the notion that 24
hours was way too much time under the bridge for reversion.  The
testsuite worked as expected, the manually alerting people of the
problems with their changes worked as expected.

To be not fair, you have to show that Lu took due diligence (ran the
testsuite) before submitting his work, you have to show that we didn't
tell him it would break users code, you have to show that he didn't
know about the breakage, you have to show he didn't break it.  I
contend my remark was fair.


Wed Aug 27 01:24:25 1997  H.J. Lu  (hjl@gnu.ai.mit.edu)

    * config/linux.h (DEFAULT_VTABLE_THUNKS): New. Defined as 1 if
    USE_GNULIBC_1 is not defined.

    * config/rs6000/linux.h (DEFAULT_VTABLE_THUNKS): New.  Defined as 1.

    * config/sparc/linux.h (DEFAULT_VTABLE_THUNKS): New. Defined
    as 1 if USE_GNULIBC_1 is not defined.

http://www.cygnus.com/ml/egcs/1997-Aug/0584.html is the start of the
thread.

date: 1997/08/27 07:26:17;  author: law;  state: Exp;  lines: +7 -1

        * config/linux.h (DEFAULT_VTABLE_THUNKS): New. Defined as 1 if
        USE_GNULIBC_1 is not defined.
Linux patches from HJ.

bash[516] cvs log p6610a.C 
revision 1.1
date: 1997/08/19 07:34:51;  author: law;  state: Exp;
branches:  1.1.1;
Initial revision