Manoj Srivastava <[EMAIL PROTECTED] (va, manoj)> writes: > I find myself agreeing, except that I feel that way as soon as > people get away from tried and tested POSIX commands and > dpkg-dev. There are far more people who are competent with cp, > install, mv, make, and other common POSIX commands, and may not be up > to date with a distribution specific mini helper language. I > appreciate the build system for certain red hat and suse packages not > being arcane and distribution specific when I try and incorporate > changes made in packages on those distributions, and I tend to return > the favour.
> Ultimately, this comes down to preferences, and what one is > used to. I would hope that most people are used to cp, mv, install, > make, nad gzip, which is what the vast majority of my rules files are > comprised of. I'm not sure the following is expression of a different preference or support for your statement above. Probably some combination of both. The difficulty that I would have with modifying a hand-rolled package is the same thing Joey pointed out, namely that I'd be afraid I'd miss something. One of the great advantages of debhelper from my perspective is that it takes care of a lot of bits of Policy that I then don't have to think about, things like knowing when to compress documentation files. My worry would be that in the process of changing something else, I'd forget some dictate of Policy. This is, in some sense, an argument in favor of your approach; I can see some real advantages to more people being directly familiar, from experience of exactly how to roll a Debian package from the ground up using nothing but POSIX commands and dpkg-dev. (Policy explains how to do it, of course, but there's nothing like practical, hands-on experience.) On the other hand, I feel a lot more comfortable with debhelper for routine packaging, since I don't trust my memory and I do trust the debhelper scripts. As you said, though, this is all preference, and most of my concern only applies when maintaining the package rather than NMUing it. The minimal modifications that one makes in an NMU are hopefully unlikely to result in other unrelated Policy violations due to forgetting to do something that a debhelper script might have remembered. -- Russ Allbery ([EMAIL PROTECTED]) <http://www.eyrie.org/~eagle/> -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
