On Tuesday 23 April 2002 10:49 am, Juergen Vigna wrote: > On 23-Apr-2002 Angus Leeming wrote: > > Yes. Remember that I had to rewrite the code pretty-well from scratch and > > that meant learning how it worked as I went along. I just came up with a > > solution that was logically correct. > > > > Storing the "modified" images in the graphics inset and use of > > boost::shared_ptr will make all more transparent. 1.3 stuff. > > Well I don't know, do you care about insets in the undo buffer? Every > undo instance can have an inset which refers to the image.
Let's talk about the present rather than some 1.3 pipe dream. At present we store a list of insets with each image in a GraphicsCahceItem, so it doesn't matter if the inset is in the buffer or in the undo buffer; it's in the list. Only when the list is empty is the GraphicsCacheItem removed. Or that's how it should work! > And one more thing why do you don't reload the image if we don't have > a GC for it? Because something has gone wrong. Part of the code thinks we have an image, part of it doesn't. Best to resolve this conflict correctly, not fudge the issue. Angus
