On Fri, 16 Apr 2010, Guillem Jover wrote: > The difference with this variable and the currently exported ones, is > that this one is always passed to the script through argv[0]: > > self=$(basename $0)
It's not clear to me that this is reliable: it some cases we would get <package>.<control> (when installed in /var/lib/dpkg/info/) and other case <control> (when called from tmp.ci). It seems also wrong for maintainer scripts to rely on this which exposes rather internal information. Suppose a script assumes one of both formats above, it might break when we get <arch>:<package>.<control> just because we decided to store the files that way for multi-arch (I know we're more likely to use different directories, but still...). > n is a constant across all maintainer scripts, in > comparison with the package name which would need to be hardcoded > on each maintainer script, or the architecture which is known only at > built time and would need modifying it then. If you consider that the maintainer must be the only one to have full control on what the script does, yes. But it can make sense for the maintainer to delegate that control by putting some functions which are more like specialized hooks. That's the approach I followed for the conffile handling stuff, it gives more freedom to change the internal implementation and make it do more without requiring intervention from the maintainer who added that hook in its maintainer scripts. Cheers, -- Raphaël Hertzog Like what I do? Sponsor me: http://ouaza.com/wp/2010/01/05/5-years-of-freexian/ My Debian goals: http://ouaza.com/wp/2010/01/09/debian-related-goals-for-2010/ -- To UNSUBSCRIBE, email to [email protected] with a subject of "unsubscribe". Trouble? Contact [email protected]
