[omniORB] Bug in inheritance across namespaces.
Steven W Brenneis
brenneis@SURRY.NET
Wed, 28 Apr 1999 21:37:48 -0500
Sai-Lai Lo wrote:
>
> Frank,
>
> The bug with MSVC++ has been around for so long that I think it ought to have
> been fixed by Microsoft. Perhaps you can help by reporting this bug to
> Microsoft instead of asking omniidl2 to generate workaround code.
>
> We've put in considerable effort to workaround bugs in MSVC++ to the extend
> that some part of the source code starts looking rather ugly and
> incomprehensible. I sometimes can't remember why some apparently rather
> convoluted code was written.
>
> Back to the inheritance namespace bug, we have put the fix Dietmar May has
> suggested for the special case he mentioned. It is an ugly workaround I do
> not want to be used for a general case.
>
> What I could do better is to document this bug and suggest workaround on
> our web page. If you can write me a concise summary of what is the IDL and
> what workaround should be put into the stub manually, I would put it on
> our web page to make the info more accessible.
>
> Sai-Lai
>
> >>>>> Frank Lynch writes:
>
> > I'm having a hellish time with inheritance across namespaces using
> > Microsoft Visual C++ 5 and 6.
> > Clients are crashing in the generated code (fooSK.cc) with a stack
> > overflow.
>
> > I think this is the same problem that Dietmar May outlined a "patch" for
> > in his mail to this list on June 29 1998. with the subject line [omniORB]
> > MSVC 6.0 bug with inheritance across namespaces (modules).
>
> > I have not been sucessfull in my attempt to apply this patch. Is there any
> > reason why this patch was not included in omniORB 2.7.1
>
> > Can someone please help me - with a workaround or some help in applying
> > the patch.
>
> --
> Sai-Lai Lo S.Lo@uk.research.att.com
> AT&T Laboratories Cambridge WWW: http://www.uk.research.att.com
> 24a Trumpington Street Tel: +44 223 343000
> Cambridge CB2 1QA Fax: +44 223 313542
> ENGLAND
Sai-Lai,
I have to agree with you in principal, but let me relate my experience
with Microsoft in reporting the namespace bug.
I reported this bug about 18 months ago. My company has a service
contract with MS which gives us 4 hour response time. After 45 painful
minutes of trying to describe the problem to their phone support
gatekeeper (who knew absolutely nothing about C++ by his own admission),
I was passed to the next level where I was given the opportunity to
explain the problem all over again to yet another non-technical flunky.
I definitely got the idea they were trying to frustrate me enough to
make me give up, but I'm not built that way. After another 45 minute
phone conversation, I was told I would be contacted by a software
engineer within 4 hours. Three days later, I was contacted by a lady
who identified herself as being part of the compiler team. The young
lady was very polite and helpful, but after 15 minutes of discussion she
told me, "I will have to ask about this. I don't know very much about
namespaces." She assured me I would be contacted within 4 hours by
another engineer. I began developing workarounds (which, BTW, I am
really tired of doing). About a week later I was contacted by a fellow
who did not even identify himself. He told me that it was a compiler
problem and that I needed to report it through Microsoft's website and
then he said good-bye and hung up. I abandonded all hope and did as he
suggested and finished my workarounds. As a side note, a colleague of
mine was told by a member of the compiler team at MS during another
phone call that they never look at the bugs reported on the website.
I am sincerely hoping the egcs group will provide us with a standard
library soon to go along with their excellent compiler. At that point I
will probably *lose* MSVC and recommend likewise to all my friends and
clients. Until then, we will all have to keep hacking workarounds and
complaining.
P.S. In case I haven't mentioned it lately, keep up the good work you
and your group do. Maybe the Evil Empire will learn by example!