Re: Directory structure
At 11:41 PM 3/14/00, Adam Di Carlo wrote:
Ross Boylan <RossBoylan@stanfordalumni.org> writes:
> At 02:56 PM 3/14/00, J.A. Bezemer wrote:
> >Some time ago I posted my proposed directory structure (at
> My take on this was that proved to be too big a change to implement for
> this release cycle. I'd love to see it go in, if it could.
> My take was as a participant in making the proposal, but purely a
> to the actual doing it.
Actually, I disagree. The scheme has many problems:
- the model doesn't demonstrate how subarches and flavors would work;
The top-level division is flavors: compact, standard, .....
True (at least for me) coming from i386 perspective, but the addition of
subarchitectures doesn't seem to change that much.
- big-disk sucks as the dir name for base2_2.tgz
Well, a name is easy to change. We kicked around several ideas. I thought
we were going to have just a single entry for all "large media" (hard disk,
- why are disk-2.88 and network peers of each other; the layout has not
logic to it
The logic of the proposal (at least my interpretation) was to approach it
from the standpoint of someone who had figured out what media and flavor
they wanted, and wanted to get all the relevant files and no more. 2.88
and big-disk (or whatever) are alternate media. This lacks logic only from
the current perspective, in which directories with media names are used
only to hold floppy images.
Top level directory: flavors (with "standard" one of the flavors)
2nd level directory: media (with 'big-disk' as one of the media)
Go to a directory and get it.
Again, I'm pretty happy with the scheme we have now and am disinclined to
Figuring out which files to get is a guessing game given current layout and
docs. For example, you go disks-x.xx/safe, and then have to figure out if
you need the safe/ files at top level. You also need to collect additional
files (rawrite). Someone switching media has it even tougher.
The current layout, in which the top level directory mixes organization by
media and by flavor lacks readily apparent logic.
.....Adam Di Carlo....adam@onShore.com.....<URL:http://www.onShore.com/>
P.S. Yes, I realized the ppp-pam was not a base system issue toward the end
of my post on that.