On sábado, 31 de março de 2012 19.02.32, Stephen Kelly wrote: > > If I can, I try to make sure that each commit is testable on its own. > > > > > > > > However, I will sacrifice testing for atomic changes. I don't like having > > big changes doing lots of things because it's hard to understand how > > certain changes came by. > > Agreed. I certainly never suggested you should squash commits together.
Which, incidentally, is another source of non-testability. The original QUrl series contained 23 patches. Besides the basics of working on IP addresses, etc., it started by introducing the URL recoder, then adapting it according to some discussions on this ML; then it added the QUrlQuery class, full with passing tests, and then proceeded to change the recoder to be more efficient given the code I had written for QUrlQuery. Then, given the discussion here, we decided that QUrlQuery should keep the order of its elements instead of letting QHash order them. When I uploaded the changes, one of the reviews requested the QUrlQuery addition, documentation and then later change of behaviour should be squashed. Which I did, but it necessitated moving the optimisations and improvements to the recoder before QUrlQuery was added. I've never tested if that reordering introduced failures in existing tests. -- Thiago Macieira - thiago.macieira (AT) intel.com Software Architect - Intel Open Source Technology Center Intel Sweden AB - Registration Number: 556189-6027 Knarrarnäsgatan 15, 164 40 Kista, Stockholm, Sweden
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ Development mailing list Development@qt-project.org http://lists.qt-project.org/mailman/listinfo/development