On Aug 2, 2012, at 10:58 AM, ext Paul Olav Tvete wrote: > On Thursday 02 August 2012 10:31:59 ext Olivier Goffart wrote: >> If that's only a documentation problem, we can workaround that with some >> macro magic. > > Very good point. For some reason I thought we would have to hack on qdoc > itself, > but creative use of \fn and \internal should be enough.
#ifdef QDOC can also be used to hide functions from the documentation. > >> Should that even go in the constructor? Maybe QImage::setColorProfile. > > The problem with that is that it is not clear from the name whether it > converts > the image data to the new profile, or just overrides the profile. Also, > overriding the profile is really just needed for specifying the profile when > constructing the image. > >> On the other side, if we add it now, and realize this should not be pointer, >> or should not be even in QImage. Then it is too late to change. > > Those are good points. Of course we could still add new constructors, and use > qdoc magic to hide the pointer argument in that case :) > > The only other point I can see is saving 6 extra exported symbols in Qt 5.1, > and > I realize that that's not a very strong argument. > > TL;DR: I will not disagree if the consensus is to revert the change. I say revert. We don't bind ourselves to potential future API. > > - Paul > _______________________________________________ > Development mailing list > Development@qt-project.org > http://lists.qt-project.org/mailman/listinfo/development _______________________________________________ Development mailing list Development@qt-project.org http://lists.qt-project.org/mailman/listinfo/development
