David> Aha. But it doesn't seem safe, generally, to make such an David> assumption about something described as "byte-compiled'. David> It's easy enough to break something in source. Of course it's not safe. But if all you've got is the binary, and you need to get work done, you patch it. Heck, I did that to Wordstar on DOS. David> Well, strictly speaking, all I know is that it's the first David> source found on the load-path. Right. And on your machine, you've set things up so that that's the one you want to edit, if I understand correctly. But I suspect that if you ask M-x list-load-path-shadows, it will tell you you have shadows. In other words, if you know what you're doing, fine. However, XEmacs will not support you (at least not out of the box), and therefore you have to provide the discipline (or infrastructure) to keep things consistent. David> Sigh! I'm always exhausting my energy honing my tools, David> rather than using the tools to build something. Well, maybe you should consider changing tools, then. However, you could also think of this as a medium term project. Just do one variant build a day until you either get a build or see something that you haven't reported to the developers before, that might be a clue to XEmacs's fussiness. I did that with Sparc Linux + XEmacs, back in the dark days before glibc became standard. Took three hours to do a full build, and the machine was pretty worthless while it was doing that. So I'd just start the build and go to bed. :-) After about 3 weeks, I had a working XEmacs, after another week, they'd fixed two generic bugs in XEmacs memory management based on better stack traces (accident of the Sparc architecture) and I had a bug logged against the Sparc Linux kernel. Took about 15 minutes a day once I got the system going. David> SWITCHING TOPIC SLIGHTLY: I have the impression that, at David> the moment, the Xemacs team is severely underpersoned. Is David> this the case? Yes and no. In the run up to the release of 21.4, there were about 10 more or less active reviewers, which was probably the biggest "core staff" XEmacs ever had, including in the Lucid days. 6 of those guys got jobs, married, or pissed off by somebody and are currently completely inactive. On net, I'd say we have a 50% personnel reduction since those days, with some new people coming in but others reducing activity. The big recent drought, though, was because for 18 months Ben Wing, who has made over 50% of the code and documentation contributions since 1995 or so, was out of the picture. He recently reactivated. For day to day stuff, Norbert has the packages well-managed, Adrian has the web site under control, and I run the mailing lists and try to keep track of the bug list (informally) and improve the infrastructure bit by bit. We could really use a new postmaster, we could really use a real issue tracker, and we probably want to move away from CVS for development purposes because we need better tracking of people's workspaces, although CVS (or subversion) is a good way to distribute to beta testers. -- Institute of Policy and Planning Sciences http://turnbull.sk.tsukuba.ac.jp University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAP