(sorry about the late reply, holidays, you know :-) Adam Heath <[EMAIL PROTECTED]> writes:
> On 21 Dec 2001, Kjetil Torgrim Homme wrote: > > > I'd prefer a non-executable file to mean "disable". This makes it > > possible to use 0 byte files only (execute bit means yes or no :), > > although packages should not use this trick. > > But this won't work. The .$file has to exist in /etc, so it can be > modified. It also must be a conffile. Which means it will exist > AFTER the package has been removed, but not purged. > > So, this means it must be an executable, that checks for a file that > exists in the package when it is installed. Okay. I thought you'd install .$file in the post remove script. So there will be a .$file for most every $file, then? > > It's all very flexible and nice, but pretty complex and not very > > intuitive for the enduser. It might help a little to call .$file > > something more explicit like .disabled?.$file instead. (The question > > mark may be a bad idea, though.) > > I don't find it at all complex. Of course not, you came up with it :-) My point is, a normal user will not have read a description of how this works. He will find a .$file and have no idea why and how. Making the filename more self-documenting would help, IMHO. Kjetil T.