Subject: Re: Package Paths Proposal v2
To: Curt Sampson <email@example.com>
From: Todd Vierling <firstname.lastname@example.org>
Date: 12/16/1998 16:49:44
On Wed, 16 Dec 1998, Curt Sampson wrote:
: > Frankly, I don't give a damn about new users to UN*X that won't even read
: > the README file that tells them how to install the stuff.
: Well, what can I say? This speaks for itself. I obviously want to
: appeal to a broader class of users than you do. (I'd especially
: love to start converting members of my local Linux users group;
: this would help in that regard.)
This didn't answer the question I asked that you didn't quote.
What audience are you trying to convert that _doesn't_ read the instructions
on how to install the stuff?
: > If they do, then it should be created by a program, not by a human, and the
: > starter files should go in libdata. (Binary scorefiles in particular are
: > typically host-byte-order dependent.)
: Indeed, they should. But do all programs do this nicely, or not?
: I'd rather leave the option open so that we can have somewhat
: ill-behaved programs in our package tree. What compelling advantage
: is there to not acommadating these programs?
What programs? Ones that don't need starter files are moot; ones that do
simply need to have pkg_add create the /var files for them. I must've
missed something here.
: > Uniquely named SINGLE config files.
: I thnk you missed my example here. Say we have a program that uses
: two config files in /etc (conf1 and conf2), and provides three
: examples of each beyond the default ones (conf1.example1,
: conf1.example2, etc.). How do you set this up under share?
A directory, of course. That's not a _single_ file. :)
-- Todd Vierling (Personal email@example.com; Bus. firstname.lastname@example.org)