On Tue, Jul 03, 2001 at 02:50:08PM -0800, Ethan Benson wrote:
> On Tue, Jul 03, 2001 at 12:40:05PM -0500, Steve Greenland wrote:
> >
> > As a specific question: what is the big deal over the uid? I don't want
> > to force it on existing systems, but I don't see how changing it for new
> > installs is that big a compromise. Ditto runlevels: of course we aren't
> > going to mess with existing setups, but I personally would rather the
> > defaults were what was in the LSB: the benefit of having 4 identical
> > runlevels has so far escaped me.
>
> the benifit is leaving what all 4 of those runlevels do solely up to
> YOU not some so called standards body.
The reason why the run levels are specified was to handle cases where
An LSB application may wish to have some kind of daemon which is only
running when X11/xdm is running.
Now, we could have added yet another level of indirection (in computer
science no problem can't be solved by adding another layer of
indirection), but after a while, the layers upon layers of indirection
become nasty to deal with. In cases where we needed to add an extra
layer of indrection, we did so; for example, structured comments
define boot facilities such as $named or $remote_fs, instead of
"S46foo.d".
In the case of run levels, most distributions were already using this
as a standard, or thought it would be very easy to switch to this
scheme. During the discusions, *no* *one*, not even the one or two
Debian developers where involved at the time, thought it would be
difficult to change the meanings of the run levels so that everyone
could be harmonized on one scheme.
So that's why we didn't add another indirection layer. Everyone else
ended up changing the meanings of the run levels to some degree or
another, and they were OK with that. I don't think any distribution
escaped unscathed on this one; we had general consensus that while we
tried to standardize existing practice where it was there (run levels
0, 1, 6) that trying to make something which was logical was of
primary importance, and of secondary importance was to follow existing
practice as much as possible.
If we had a lot of people complaining about this particular point,
perhaps we could have added Yet Another Indirection Layer. Of course,
then we would have people like Nick Petreley of LinuxWorld writing a
poison pen editorial about how the LSB folks had no spine, and was
afraid to dictate anything to the distributions, and instead added
needless complexity. The bottom line is that it's pretty much
impossible to make everyone happy in this business; no matter what you
do, someone is going to throw stones at you.
- Ted
