No definition of virtual function, omits inline and virtualtable

[James F. Carter]

But the point is, that compilation unit was botched.  As you say, the
program as a whole was malformed, and could not have executed (at best
getting an undefined symbol at link time), but the missing virtual function
*could* have been defined in a different compilation unit and linked
in.  That should be legal, but no inline bodies or virtual table would have
been generated.

It was this exact situation that clued me in to the bug.  I never saw the
linker try to find the (incorrectly undefined) virtual function, which is
why it took so long to figure out what was triggering the bug.

James F. Carter        Voice 310 825 2897	FAX 310 206 6673
UCLA-Mathnet;  6115 MSA; 405 Hilgard Ave.; Los Angeles, CA, USA  90095-1555
Internet: jimc@math.ucla.edu (finger for PGP key)
UUCP:...!{ucsd,ames,ncar,gatech,purdue,rutgers,decvax,uunet}!math.ucla.edu!jimc

On 24 Sep 2000, Peter Osterlund wrote:

> Jim Carter <jimc@math.ucla.edu> writes:
> 
> > /* bug03.C -- No definition of virtual function, omits inline and virtual table
> 
> The C++ (draft) standard says
> 
> 	A virtual function declared in a class shall be defined, or
> 	declared pure (_class.abstract_) in that class, or both; but
> 	no diagnostic is required (_basic.def.odr_).
> 
> so your program is ill formed. T::twiddle() is never defined. You
> should be lucky you got an error message at all. The standard also
> allows the compiler to create a program that crashes mysteriously at
> run-time.
> 
> -- 
> Peter Österlund          Email:     peter.osterlund@mailbox.swipnet.se
> Sköndalsvägen 35                    f90-pos@nada.kth.se
> S-128 66 Sköndal         Home page: http://home1.swipnet.se/~w-15919
> Sweden                   Phone:     +46 8 942647
>