Re: Improving [was: Massaging] load-history

Sean MacLennan writes: Sure, the point is that editing the installed Lisp is going to fail unless you have root, and then it's going to fail to be reflected in the running XEmacs the next time you start it because dumped Lisp doesn't check for an edited version on load-path and reload itself. I haven't had a problem with this in ages, although I don't refer to Lisp sources very often on Windows XP and may never haver tried to find a dumped library in that environment. See above. AFAIK at present the only ways to determine if a Lisp library is dumped are (a) parse the dump list (in a non-dumped library whose name I forget) and (b) examine load-history. We could try to implement a better way, but it needs to be better, not just different. Dontcha just hate it when your bug reports are closed with "works for me"? I don't remember the exact details but there were corner cases where improved code got "lost" because people were changing installed Lisp which got overwritten by the next install. We need to be careful that this continues to work for those that it currently works for. Why you don't need load-history? The find-library functionality falls back to load-path if load-history is null, for one thing. In general, with the main exception of dumped Lisp, searching load-path and searching load-history should give the same results. Only if load-path is changed after initialization such that a different instance of the file ends up earlier on load-path would you expect to find a difference. I do think it's bad if you can't find source for a function which is available in your running XEmacs. Currently XEmacs does (locate-file FILE (or load-history load-path) EXTENSIONS) but I think it could try harder with something like (or (locate-file FILE load-history EXTENSIONS) (locate-file FILE load-path EXTENSIONS)) No time to think about it tonight though (and you should check the locate-file API and make sure it returns nil for not found). BTW, I don't argue with your logic, and certainly don't claim you screwed up somehow. It's possible I'm too conservative about these things, but I follow That Other Emacs and I don't much like the way some of their committers regularly upset other users with "works for me" tweaks. (Fortunately Stefan takes a really dim view of that kind of thing, but even Stefan typically does little to ensure backward compatibility if it would significantly complexify the code.) _______________________________________________ XEmacs-Beta mailing list XEmacs-Beta(a)xemacs.org http://lists.xemacs.org/mailman/listinfo/xemacs-beta