FC4t2 no good without LILO

[Graydon]

On Wed, Apr 13, 2005 at 08:12:12AM +1000, Res scripsit:
> On Tue, 12 Apr 2005, Peter Jones wrote:
> >available.  That time has long since gone.  Those people who still
> >need lilo, for whatever reason, would serve themselves better by
> >helping debug and fix things than by posting really whiny flames with
> >an unacceptable lack of detail about any real problems to the test
> >list.
> 
> From: 	NOC
> TO: 	All CSR's
> RE: 	Mail Web DNS News Radius DB Accounting Cacheflow servers
> STS:	High
> 
> All staff may take an indefinate period of non paid leave effective
> immediately, your return depends on how long it takes for us to
> lodge bugs and have them resolved so we can acutally boot at least
> some of the servers in the datacenter, our OS developers have said we 
> should not install a small 100k programme we know will resolve but yet
> keep everything stuffed until we help them fix the bugs with there 
> prefered non universally standard boot loaders
> 
> Kind Regards
> Some Brave Engineer

Oh, piffle.

Leaving aside that Fedora is specifically intended to be a technology
testbed, with the concomitant risk and aggressive schedules of a
testbed, and leaving aside that the problem that started this thread off
_is not solved by lilo_, which won't boot > 2 TB partitions *either*,
and leaving aside that it's tremendously bad, *bad*, *BAD* practice to
boot from the RAID array in the first blessed place on any sort of
production system, and leaving aside that grub has a whole lot of
serious advantages, starting with much more active development and
greater flexibility, you can't leave aside that _no one even vaguely
sane_ -- not anyone who knows sane's postal address -- is going to be
doing untested migrations of the data center servers.

Note I didn't say 'untested migrations of the data center servers to a
test release of a technology development platform distro'; doing it *at
all* just isn't going to happen.

If you think that is an even vaguely realistic scenario, then, well, ok,
you're frustrated.  But you also don't know what you're talking about.
(Or you've been dealing with utter drooling madmen, which I suppose is
possible.)

Linux *changes*.  This can, and will, mean that something you understand
and are comfortable with gets replaced by something that you don't
understand and aren't comfortable with for reasons that have nothing to
do with any problem you have or situation you need to or want to deal
with.

That's a general trend with open source software; tools and the
underlying support mechanisms become more general over time.  That makes
a lot of simple things less simple and a lot of difficult things
possible, and from the point of view of the folks dealing with the
difficult things this is an obviously good trade.

So is getting rid of duplicate ways of dealing with things; cost -- in
time and effort as well as money -- scales with parts count, and from
the point of view of a linux distro, packages are parts.

No amount of rhetoric will get anyone around those two things; they're
fundamental consequences of any prolonged process of complexity
handling, even those which don't involve computers in any way.

A certain comfort with change, a reliable neophilia, is a requirement
for dealing with the penguin.

Demanding that strangers give you what you want, against their own
interests, because this batch of free stuff isn't the free stuff you
want, is *not*.

You can try figuring out how what you want is in their interest, though;
that generally works rather better.