Re: [NEWS] status of boot-flopppies
Moin Massimo Dal Zotto,
> Your proposal of self-consistent single cd is interesting but IMHO it is
> applicable only to updates of an already installed distribution or to a
> small distribution which can be fitted on a single cd.
Well !nobody! wants to install a !complete! debian, but anybody wants
to select his favourites.
First of all I think, we need some "standard" way to produce single
CDROM "selection of potato" Debian-Starter-Pack. Those Debian starters
are necessary to for install parties and other events. Our Debian starter
based on a merged "get-selection" of those people in the /usr/group/bremen
who answerd the mailing list query of Peter Ganten for a get-selection
of people running Potato.
Now I think we need the Debian-Booster-Pack. Each Debian-Booster-Pack
should be able to upgrade a basic debian system, should be usable as
a rescue CDROM, and should provide some "up to date application groups",
like lets say the brand new GIMP+Mozilla+XML development. Each Booster
Pack may also be used to install a brand new Debian that is centered
around those "featured application" of course.
Ok to awoid the tragic gathering of hundrets of boosters to collect
a complete Debian-Unstable (which is impossible of course :-) we also
need the Debian-Potato-Decatlon, for those who need all.
But in times of cheap 25MB++ hard disks, many would prefer to copy
those CDs to hard disk once, instead of changing them. And as Debian
is not a fixed point, but a moving target, anybody who is not on the
internet, would like to upgrade parts using the boosters. So the most
usefull and necessary tool, is a script for checking a "selection of
potato" for constency, that we dont end with a mirror that would
break "dselect install" with a "file not found". Which is quite common,
if you pick the wrong time and method for mirror.
> My discussion was mainly about the official Debian distribution, also if
> all my experiments have been made on a single cd distribution like yours.
well Debian Potato is not yet stable - so a single CD starter, for those
who want to install Potato is imho currently the best idea. Those people
need contact to the internet to upgrade. But wait, we made our CD on Nov11,
I installed it and made a "dselect update" to current. The install was
about 100MB download, while the CD contained 420 of software. So our
selection of Potatos changed by around quarter in 2 weeks. I think in a
month or so, nobody wants to download a complete upgrade by phon.
> The next full distribution will be 3 or 4 cd and we can't make assumptions
> on which packages the user wants to install, so having self-consistent cd
> can be difficult to achieve and probably not very useful for general use
> as the user will need anyway to insert at least 2 or 3 cd.
My idea with a multi CD is, that we would currently 3 CDs for one platform.
If I use 4 CDs any of them would be self consistent. If we now have a file
on any CD that tells that this, CD is beloning to a collection, so we could
tell dselect to use a "collected cdrom" as update, and not a "single cdrom"
changing CDs for installing packages would be reduced to a minimum.
> Your idea can be interesting if we want to make periodical updates for
> security fixes or major components upgrades.
Updates and Starters are also my base idea with those self consitend CDs.
> In this case the update cd
> could be bootable and contain also the apt lists of all previous cd's.
> In this way it could be used for upgrade and for first installation of
> the updated distribution.
We need some other ways for multi CD Debians of course, as a multi CD
distribution needs to appear in a single "dselect select", I think the
most important tool is a "mirror checker" - I'll ask Peter to publish
his script on copyleft.de, so people building multi-cd debians may
take a look on it.
Bye Michael
--
mailto:kraehe@copyleft.de UNA:+.? 'CED+2+:::Linux:1.2:13'UNZ+1'
http://www.xml-edifact.org/ CETERUM CENSEO MSDOS ESSE DELENDAM
Reply to: