[EMAIL PROTECTED] (Aaron M. Ucko) writes: > Daniel Brockman <[EMAIL PROTECTED]> writes: > >> To get your Emacs back, remove dictionary-el, install emacs21, and >> then install dictionary-el again (if you want it). > > Good catch; I'm not sure why I never ran into the problem myself.
This bug only bites when you install a new flavor (really, who would reinstall?), and Emacs 21 has been out there for a looong time. :-) >> To fix this, I suggest avoiding loading things from the package >> startup file. Here, I've simply replaced (load "dictionary-init") >> with the contents of dictionary-init.el, which just contained a >> bunch of autoloads anyway. This seems to be what most packages do. >> (Except a few that also suffer from this problem; I'll get to those >> if this patch is accepted.) > > That's how I originally did it, but then I switched to using load to > reduce redundancy Yeah, I saw that while grepping the changelog. > (though I admit that that can also be accomplished by generating the > file automatically), improve conffile stability, and avoid the > possibility of dangling autoloads if the package is removed but > not purged. Good point. I'm not sure I understand why these files are conffiles in the first place. So you can have control over what gets loaded? > However, there are other solutions, such as calling load with a > non-nil second argument and then printing a non-error message if it > fails; I plan to fix the package that way when I get a moment. Oh, interesting, I didn't know it took all those arguments. :-) >> By the way, is there any reason to keep the deb/ directory around, >> when there is a debian/ directory which is better in all respects? >> If we removed deb/, we could nuke dictionary-init.el completely. > > I use upstream tarballs as is whenever possible, which is the case > here. (However, the upstream author did initially say that he would > remove his deb directory when I released official packages; I guess > he just forgot about that, and I didn't consider the issue worth > bugging him about.) Ah, of course. > Anyway, thanks for the report, which I'll try to address before > too long. Thanks for the quick reply! -- Daniel Brockman <[EMAIL PROTECTED]> -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
