Re: errors in packaging packages
In article <[🔎] 199612191536.KAA28314@mercator.math.uwaterloo.ca> you write:
> I find it somewhat ironic that 'hello', the example of how to
> package a package has no man page.
(Should at least be a symlink to undocumented(7)!)
Well, it's GNU hello, and GNU don't believe in man pages these
days... (Myself I think there's a need for both info and man pages;
info lends itself well to tutorials and the like, whereas man pages
tend to be better for quick reference.)
> Andreas' program to check packages for errors is a good one.
Apart from the extremely unhelpful Subject: lines in the bug reports
generated, yes.
> I realize he is simply currently going through the current packages,
> but the right way to do this would be to have packages checked
> before they get into the system.
One would hope that people would read the standards documents so as to
avoid most of these mistakes in any case!
>This could be done a few ways:
>- distribute the program so people can check their package themselves.
>- have debmake use the script (debmake may do some checking already.
> I don't know). Doesn't help if debmake isn't used though.
Indeed, and I'd prefer the use of debmake (as was the case with
dchanges) to remain an optional aid for package maintainers.
>- have any non-compliant packages get rejected. Must be a way to
> override rejections.
I'd disagree with this one; it'd be daft to reject a package solely on
the grounds that its manual page isn't compressed, for example.
> - accept all packages, but have them checked automatically and send
> the maintainer mail with any problems.
Perhaps; an auto-`auto bug report'...
(S)
--
TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to
debian-devel-REQUEST@lists.debian.org . Trouble? e-mail to Bruce@Pixar.com
Reply to: