> Von: [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED] Auftrag von Sergei > Organov > > > Jonathan Larmour <[EMAIL PROTECTED]> writes: > > Sergei Organov wrote: > >> Jonathan Larmour <[EMAIL PROTECTED]> writes: > > [...] > > >>>Therefore I think we have a choice of solutions: we can > either change > >>>the pure virtual functions to be something lik e.g.: > >>> > >>>virtual Cyg_ErrNo write( const cyg_uint8 *buffer, cyg_ucount32 > >>>buffer_length, cyg_ucount32 *bytes_written ) { > >>> CYG_FAIL("write method called directly on Cyg_OutputStream"); > >>>} > >>> > >>>However even that depends on assertion support being on, so is not > >>>ideal. > >> > >> Why not ideal? Either you want assertions and then you compile with > >> assertions on, or you don't. I think it's the best way to > handle this > >> provided you don't want to bring C++ runtime to every > application. [I > >> have my applications in C++, so I do battle with these > issues anyway] > > > > Firstly, since this would be a fundamental problem, if it > > occurs with assertions off, it could be unnoticed. > > The question is how is it different from any other occurrence > of CYG_FAIL in the eCos source base?
That's what I thought too... Alex
