On Wednesday 26 September 2012 13:54:10 Stephen Kelly wrote: > Hi there, > > In > > https://codereview.qt-project.org/#patch,unified,35581,1,/COMMIT_MSG > > we have a discussion about whether to reimplement a virtual method (the > obvious way to fix this bug, as was done in the Qt 5 patch), or to avoid > doing so. > > In this case, code which gets recompiled will get the bug fix, and code that > does not get recompiled will not get the bug fix. > > Do we care about cases like that? Existing compiled code will continue to > work as before. From the point of view of the dynamic linker, it is BC as > there is no change to the virtual table size etc (AFAIK). > > Do we have a hard policy for this? 'It's fine to add virtual method > overrides as long as the existing compiled code continues to work without > new bugs' makes sense, but do we have it written down anywhere authoritive?
I personaly think it is safe to say that one need to recompile to get a fix. What is important with binary compatibility is: "an application compiled against an old Qt will continue to work". In this case, the application already had the bug, so there is no regression. -- Olivier Woboq - Qt services and support - http://woboq.com _______________________________________________ Development mailing list [email protected] http://lists.qt-project.org/mailman/listinfo/development
