Re: Reasonable maximum package size ?
On Mon, Jun 11, 2007 at 01:24:34PM -0600, Warren Turkal wrote:
> On Monday 11 June 2007 13:09:40 Roberto C. Sánchez wrote:
> > That may be true when it comes to breakdowns. However, I challenge you
> > to show me a "cheap" desktop disk that is also SCSI or SAS *and*
> > hotpluggable.
> While not SCSI or SAS, there are SATA controllers that support hotplugging
The point wasn't that you can't set up a professional RAID array using
cheap desktop hard disks; you can, if you really want to, though I
wouldn't recommend it. And yes, you're completely free to ignore that
particular advise, so long as you don't expect me to become a customer
The point was that a single 500G hard disk just doesn't cut it for a
publically-accessible Debian mirror; that you need much more than that,
both in terms of quality (hotpluggable controllers don't really help if
your disks aren't ready for hotplugging--which is usually the case for
cheap desktop-class hard disks--possibly other interfaces than the
popular SATA connection, and preferably also disks that have had way
more testing than desktop-class hardware) so that "there is this set of
cheap crappy hard disks that are large enough so that you can store an
entire Debian archive on it" is a ridiculous argument in favour of
"throwing every insanely large piece of junk into a Debian package and
onto the Debian archive".
Note that that also doesn't talk about *what* exactly defines the line
between "an insanely large piece of junk" and "a piece of interesting
data large enough to be useful, but not so large as to become a
problem". The borderline there would seem to be rather gray.
Personally, I'd advice the OP to try to find the smallest data set that
is still at least moderately useful for the package, and to package that
(so that people not familiar with the software or the type of data it
requires can still try it out), unless that would require a package
larger than a few tens of megabytes. Additionally, it would be useful to
point in the package description and/or its documentation to either a
way to automatically generate debs out of a larger data set so that
users can generate their own data debs and distribute them on their own
network, or to a separate debian archive that they can add to their
sources.list if they want.
I don't think setting a hard limit (as in, with numbers) on maximum
recommended package size is a very good thing to do; I'm tempted to
think that this kind of limit is not really static over time, and thus
that wording such as "roughly two times the average package size at the
time the package is created" or so would be more appropriate.
<Lo-lan-do> Home is where you have to wash the dishes.
-- #debian-devel, Freenode, 2004-09-22