Jonathan Larmour <[EMAIL PROTECTED]> writes: > Sergei Organov wrote:
[...] > The only potential problem could be extra footprint caused by pulling > in the various versions of the write() and get_error() methods, along > with the desstructor. But in this *particular* case, this isn't so bad > as those functions do not themselves impose extra requirements of any > note, and people are always going to want to write() a stream. So it's > not a free-for-all for virtual functions by any means - the same logic > would not apply to all methods, but this particular case is ok. Yes, I agree that an application that wants to read() but not to write() ought to be almost impossible to meet in practice. > > If you could repost the patch with the changes we did agree on, that > would be great, thanks. Although I won't be able to commit it before > your copyright assignment arrives. That's fine with me. I then defer the effort of changing the patch till then. -- Sergei.
