On Wednesday 27 February 2002 7:57 pm, you wrote: > Angus Leeming <[EMAIL PROTECTED]> writes: > > | On Wednesday 27 February 2002 7:12 pm, Lars Gullik Bjønnes wrote: > >> Angus Leeming <[EMAIL PROTECTED]> writes: > >> > >> | Forget it. It isn't going to happen! > >> > >> IMHO a serious regression from figinset. > >> > >> | The graphics cache is external to any > >> | buffer and isn't going to know anything about one. What happens if the > | view > >> | changes dimensions? You'll be asking for the image to change with it. No! > >> > >> yes. > > > | Alright smart arse. So you have two views of the same buffer, but one view is > | wider than the other. Do you want different pixmaps? > > > | No! > > But you still want to scale relatively to the textwidth. You either > you have a clever solution to avoid different pixmaps _or_ you just > use different pixmaps.
Well, I slept on this and have calmed down. It is possible to do as you desire, but it will involve a re-write of the re-written graphics cache. In fact, the graphics cache will return to a form similar to Baruch's original code. Each insetgraphics will then store a GImage instance and will do the croping, rotating and scaling itself. If the scaling request is %t or whatever, then the scaling and generation of the pixmap will be performed in the insetgraphics' draw member function, else all operations will be performed on a changed status signal (as now) and cached That way you'll get what you desire. Happy with the plan? Angus
