Re: The redisplay engine is getting really messy ...

Hrvoje Niksic writes: Hrvoje> When we talked in Paris, you said that you considered the redisplay Hrvoje> engine (relatively) well-written and worth upgrading. That's not exactly what I said. My feeling about redisplay is that inside a certain conceptual domain (see below), the stuff is quite well written: for instance, it was trivial for me to make the dim face property work under X, or to add new specifier like the divider width/height, or some shadow thickness somewhere... those things were trivial to implement because they don't interfere with the current high-level model of redisplay (what I called the conceptual domain). At the time we met, I was thinking mainly of those things that I'd done. But the point is that nowadays, we're beginning to experiment the limits of the current model, in two ways IMHO: * first, the increasing number of features and their complexity make several aspects of the current redisplay less efficient and robust. The critical example is what/when to redisplay. It's probable that the current model got confused during the last major redisplay modifications before Andy's subwindow support (IIRC, Kirill and I were quite active at that time). * second, there are new features we'd like to have but I think would require a major rewrite. Pixel based scrolling, real 2d formatting and the like doesn't work well with the current, line-oriented model. I've talked several times with Chuck and I know he doesn't totally agree with me on this, but for instance, I found quite hard to implement the bevel extent when I tried it. So hard (or more precisely kludgy) actually that I turned out thinking that we'd better rethink several lower-level functionalities before. Obviously, those are my impressions based on the current knowledge I have of redisplay, which is for sure incomplete. Hrvoje> Like getting your PhD done. :-) My what ? :-) -- / / _ _ Didier Verna http://www.inf.enst.fr/~verna/ - / / - / / /_/ / E.N.S.T. INF C201.1 mailto:verna＠inf.enst.fr /_/ / /_/ / /__ / 46 rue Barrault Tel. (33) 01 45 81 73 46 75634 Paris cedex 13 Fax. (33) 01 45 81 31 19