[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Where is Debian going?



On Wed, Jul 10, 2002 at 05:51:20PM -0400, Thatcher Ulrich wrote:
| On Jul 11, 2002 at 02:04 +0800, csj wrote:
| > On Wed, 10 Jul 2002 11:03:17 -0500
| > "Jamin W. Collins" <jcollins@asgardsrealm.net> wrote:

| > Maybe instead of stable, testing, and unstable, we can have: server
| > (must be stable), desktop (with newer but not bleeding edge stuff) and
| > developer (because they're the ones who're in the best position to fix
| > it).

As long as the names' connotations convey their intent, it really
doesn't matter what they are.

Just be careful with calling one release "server"  and one "desktop" :
    Do you want people to think that they can only run "server" on a server?
        (I run testing/unstable on some servers, and I say it is still
        better than RH)
    Do you want people to think they can't install debian on their
        desktop because "desktop" doesn't have official CDs?

(that's what [wc]ould happen if you keep the same release model and
just change the names)

| MHO newbie opinion: emphasize version numbers (2.x, 3.x, 4.x) and

That sounds fine for a "you can only get 'stable'" release method
(like every commercial software house runs it), however what version
is sid/unstable?  What about testing?  Only stable releases have a
version number.  If you tried to tack a version on testing and
unstable, you would have to change it every time the dist changed,
which would overflow my capacity for a version.  Using a name is the
simplest way to label a moving target.

How hard is it really to learn that
    "unstable"  means latest-and-greatest-but-who-knows-if-it-breaks-your-system
    "testing"   means tested-a-bit-and-should-work-but-YMMV
    "stable"    means tested-a-lot-and-is-really-stable

Then you create a map that is correct until the next release, which
right now looks like
    sid     <-> unstable
    woody   <-> testing
    potato  <-> stable

It is 6 names, but 3 of them are redundant.  Half of the names are
"timeless"; that is, they will refer to the same thing later as they
do now.  The other half are relative names; they refer to a certain
stage in the release process, and those are moving targets.

Since there are 6 different concepts at work here (5 really), 6
different names are needed.

| start using mnemonics that are less intimidating to the casual user,
| or just de-emphasize the mnemonics altogether.

I think the real "problem" there is that debian's development model is
so different from every other software vendor that people find it
"confusing" (IOW new and different).  In fact, every other OSS project
uses mnemonic names like "stable", "released", and "CVS" to refer to
the different stages of their release cycle.  Most projects aren't
quite as large nor do they have as many stages, though.

| The Toy Story code names are cute but totally confusing; I say ditch 'em.

Everyone has names -- even RH and Mandrake (Seawolf, Enigma, Cooker).
There's nothing wrong with the names, really.

-D

-- 
 
The crucible for silver and the furnace for gold,
but the Lord tests the heart.
        Proverbs 17:3
 
http://dman.ddts.net/~dman/

Attachment: pgp33dhuk31hc.pgp
Description: PGP signature


Reply to: