Quoting Adam Borowski (kilob...@angband.pl):

> Hell yeah.  I have a completely opposite view: if a piece of software hasn't
> been packaged, it's most likely because it is crap -- otherwise someone
> would bother to do the work.  And, usually upstream developers know how to
> deal with the particular problem their software is made for, but have no
> idea how to integrate it with the rest of the system.

There is much, much to this.  I understand that much of Gerrit Pape's
work in making Debian package dbndns (updated djbdns) is politely
countermanding peculiar DJB software-architecture decisions, for example
(not least, Dan's jihad against FHS).

As another example, I think Laurent Bercot's s6 suite and s6-rc init are
pretty good, but fear that exactly such pain-in-the-ass obstacles as you
list lurk for packagers.  (I could be impolitic and point out that a
recurring theme in the 'gratuitously making things difficult for
distributions' area is 'influenced by Dan and daemontools'.)

> You can build from a tarball on your home toy box.  It simply doesn't scale
> once you try to deploy that onto a bunch of servers -- even within just your
> organization.  A distribution is far more complex than that.

There's room for both, though -- and for middle grounds.  I'm deeply
sympathetic to Steve's view of 'Screw all these problems with distro
packages.  I'm going to just make it locally.'  I may end up doing that
in order to run my next-generation home server on mdev.  But I'll
probably at least make a .deb out of it.

_______________________________________________
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Reply via email to