"Stephen J. Turnbull" <turnbull(a)sk.tsukuba.ac.jp&gt; writes: I'd say simply because XEmacs is *big*, we are numerous to hack on it, and it's an open development process. This means in particular that: - XEmacs is so big that nobody on the list can pretend [s]he knows everything about XEmacs. There are more than 150.000 lines of C and more than that in lisp. - In that, we're all ignorants, but we're all experimented as well. All technicians of some sort. People rarely throw general ideas or remarks. People usually send patches to implement or modify the concerned feature, and this is because the most important feedback we get on XEmacs usability is from ourseleves, the people who develop it. Hence the "defaults" are intrinsicly biased. - Moreover, as Stephen says, for us it sucks to try the defaults because we know more, so we want more. I don't think so. The fact is that all the *behaviors* are preferred by somebody. Whether such or such behavior should be the default is the real question. And the real problem begins when people are so used to the behavior they like that they begin to confuse it with the default thus breaking simpler things. Yes, but we need to answer another question: "do we need a definition of 'default'?", and I think we do too. Exactly. Consistency /is/ the point, and it has a-priori nothing to do with complexity or obviousness. To me this leads not to one definition of 'suck', but two: something can suck as a matter _taste_, or as a matter of _consistency_. As a matter of taste, there's nothing much we can do. Some people like dark backgrounds and some others prefer light. This is a matter of taste and this is not arguable. There can be flame-wars between the dark-background-camp-by-default and the light-background-camp-by-default, but this will be sterile anyway. For this kind of debate, we should probably vote or something. As a matter of consistency, there's much that we can do. Like what's happening for the `push-mark' debate, it's always harder to fix a-posteriori an inconsistent behavior when it's the result of a graduate increase of complexity in a feature. But unfortunately, this is how it will happen most of the time because since we're all rather experimented users, it's even more difficult to keep a genuine and high-level view on what we're doing.[1] Now about the other question: "which definition for 'default'?" - The first thing is that at a first glance, any default should be okay, as long as it is consistent. And consistency this is part of our job, as developers. - The second thing is to ask "who the defaults are for?". The answer is easy: for people that don't know (yet) how to modify them. I'll probably make people shoot in my direction there: when Steve says that "xemacs -vanilla stinks", I don't consider it a valid argument. At least, not apart from consistency matters. Of course it stinks; we're so much used to an in-depth usage of it on this list! How could it be any other way? Now take a random guy out there, give him a vanilla xemacs, ask him to edit something and listen to what he says. What *he* will find stinking will probably be stinking. Before yelling at me, let me just precise this: xemacs for novices doesn't have to be dramatically simple and powerless, on the contrary to what Kyle thought I was saying. This is not a matter of /complexity/, but a matter of /obviousness/. I mean we can have something very complex by default. However, if the behavior the user is facing is not obvious, in other words "why the fuck did xemacs do this?!" then, it's the wrong default. Defaults doesn't have to be stupid, but for sure, they are not for us. Now if we'd all like a complex behavior by default, but this makes other things non-obvious, then, we should stick to a stupid default, and add a few documented lines in sample.emacs. I think that's the way I see the answer to this question of Stephen: | If there are only two, then we can have a massive flamewar over which | is going to be hard-coded as defaults and which is going to be | enshrined as sample.emacs. Jane User who decides the background color, I dunno, and it doesn't really matter. Jane User who decides that something is inconsistent or obscure, or even _not powerful enough_ in a vanilla xemacs, I'd say as many non computer scientists as possible. You're joking, but actually I think you're getting closer to the source of the problem. Add an xemacs-custom-default-value write-only mailing list and you're done. :-) I have very few things either, and I think that's what makes me very sensitive to the defaults' consistency. Footnotes: [1] This is the reason why I keep talking about beginners. Because I'm convinced that we need them as consistency-checkers. -- / / _ _ 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