Am Montag, den 14.05.2018, 13:18 -0400 schrieb Richard Kimberly Heck: > I've been thinking about this, too, and had come to a somewhat > similar > conclusion, though not with nearly as much detail. I.e., it seemed to > me > that there was something wrong with how the parents were being > assigned > when we clone. Unfortunately, there is a reason that we clone from > the > master: There may be macros defined there that are needed in the > child. > (See 2d6173d8103.) It may be that we should have said that wasn't a > bug: > If you view only the child, then you shouldn't expect macros defined > outside it to work. Maybe the include_only support allows people to > do > what they were trying to do here by viewing only the child?
That's definitely wrong. A child compiled standalone should not inherit anything from a parent. This opens up for all sorts of bugs. > It seems to me that, in some ways, this is tied up with different > understandings people have of the master-child relationship, and what > it > means to view "only the child". Maybe. But "Viewing the child" is actually not what is the case here. It's not a child in that case (we have includeonly support for that case). Jürgen > > Riki >
signature.asc
Description: This is a digitally signed message part