On Sun, Jul 01, 2012 at 11:17, Tobias Stoeckmann wrote: > Your patch breaks this compatibility for every branch except the initial > one. GNU CVS is not consistent about its behavior in many ways. > >> Ha! This is the same bug I was looking into last week. I wasn't sure in >> what circumstances adding the magic (like wtf?) is necessary. I'd say >> this is probably ok. > > I disagree: > > $ pwd > /home/tobias/src/lib/libssl/src/crypto > $ cvs log mem.c > RCS file: /cvs/src/lib/libssl/src/crypto/mem.c,v > Working file: mem.c
> > > Magic, magic everywhere! Except in this special initial branch one. ;) > > If there is no need to be compatible to GNU CVS as strictly as the project > initially tried to, I'm fine with that. But as far as I know, this patch > might break scripts and other tools that expect this weird behavior. I had trouble finding a decent file where there was a difference, thanks. I guess we need an uglier patch to only skip magic if the version is 1.1.1?
