Hrvoje> Jan Vroonhof <vroonhof(a)math.ethz.ch&gt; writes: Hrvoje> It's not. My point is that I want to have a *choice* Hrvoje> whether I want `.xemacs/' or not, and for that choice to Hrvoje> be respected. You seem to be confusing "having a choice," and "having the choice default to your preference." To be more constructive, I agree that having the default be "create ~/.xemacs" looks hard to implement cleanly, although maybe I'm just not creative enough. That is not your burden, it's ours, of course (ours == the "create .xemacs/{init,options}.el and be done with it" camp). But we'd appreciate your help if you choose to give it! Oscar> Grouping things together under a single Oscar> .xemacs/ directory is much cleaner and I am yet to see any Oscar> good and objective reason for not doing so since until now Oscar> I've only seen vain religious arguments. Religious arguments are never vain. "Cleanliness is next to Godliness," not the other way around ;-) In any case, Hrvoje's primary argument is that we should try to avoid doing anything that breaks existing .emacs files, as that is the supported interface. Further, we have a burden to provide tools to help the users that choose to do the migration. I don't know how to do it, and nobody has analyzed it, at least on xemacs-beta. This is a religion that you also seem to be a member of. You know, I haven't seen _any_ comment on this that runs "I support 150 XEmacs users, and how am I going to migrate all of them?" It would be nice if all the admins like that are thinking "I support 150 XEmacs users, and this is gonna be a piece of cake!" I don't think so. (In fact, I believe Hrvoje himself is a counterexample.) Oscar> not to have XEmacs add more... I believe that .xemacs/ Oscar> should clearly be the default and multiple files in ~/ the Oscar> option, not the opposite. Well, we need to show that can be implemented cleanly. I can see enough potential for messiness that I agree that we should postpone it to 21.1, and that the _very first patch_ creating 21.1-beta1 should be the new, quite possibly very broken, user-init-file scheme. That might allows us to work out most of the bugs by 21.1 release. :-/ As long as we're postponing this, I think we should completely disable searching for .xemacs/init.el for 21.0. XEmacs-beta readers can emulate the functionality easily enough (I have done this already, both the disabling and the emulation, although the testing of GNU Emacs, and further automatic customization of the files, has been zero). Anybody else who is doing it is likely to be in for nasty surprises when we decide on the One True User-Init-Process. I recognize I'm probably in a minority of one on this, as even Hrvoje has signed on to retaining .xemacs/init.el as an undocumented feature.