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

Re: Proposal: Debian release numbers



On Tue, Jan 07, 2003 at 10:08:07PM +1100, Martijn van Oosterhout wrote:
> On Tue, Jan 07, 2003 at 10:54:53AM +0100, Wesley W. Terpstra wrote:
> > On Tue, Jan 07, 2003 at 09:20:54AM +0000, Scott James Remnant wrote:
> > > I therefore propose that we alter the scheme for release numbers as
> > > follows.
> > > 
> > > Debian stable releases will be identified by their code name and a single
> > > release number, starting at the next whole number (4) and incrementing it
> > > for each new stable release.  The next stable release would therefore be
> > > "Debian 4 (sarge)", the release after "Debian 5 (etch)", and so on.
> > 
> > I think this is a great idea!
> > 
> > When people talk about debian, they always refer to 'hamm', 'potato',
> > 'woody', 'bo', etc. This is a good indicator that our version numbering
> > system is not memorable. We should definitely be including the codename
> > in all version numbers so that people understand which debian they have.
> 
> I'm not sure I agree here. You're just encouraging version number inflation.
> Is this a plan so we can catch up to other distributions?

No. :-) [ and given our release history, would we even catch up then? ;-) ]

> Seriously, the reason people remember the name is because you never use the
> number. apt lines use the name, the directories on the CDs/archives/websites
> use the name. Very, very few places use the version numbers, so it's no
> surprise you don't remember it.

That is true. This agrees with what I was meaning to convey:
	the version number should include the codename

> So perhaps the real solution is to drop the number altogether and talk about
> woody.0 :)

I am almost fine with that idea. 

My only complaint would be that it is hard for a person who is not familiar
with debian to deduce that woody is older than sarge. Debian 3 (woody) vs
Debian 4 (sarge) is clear.

> > Not to mention, I think every one of our code-named releases constitute a
> > major upgrade. Even as a debian developer I would be hard pressed to tell
> > you what the version number of potato(!) was without looking it up. A single
> > number always accompanied by the codename people remember is sure to be more
> > understandable to users as well as reinforce the codename<->number mapping.
> 
> All our releases are major because there's such a big delay between each. If
> we released faster they wouldn't be so major.

Again, a good point. However, I still have doubts about the purported new
'fast release' schedule. :-) I have yet to see one actually happen. :-)

More seriously, I think the original post is correct here. The decision
about what is a major release and what is a minor release are a source of
contention. Eliminating this is good.

I also find it hard to keep track of the version number. I can tell you that
potato is two releases prior to sarge though. Hence arithmetic works better
on the proposed system. We all know the release order by name, this should
be transferrable to the version number.

Then we have a nice simple timeline in our heads.

The whole major/minor thing requires us to make value judgements. If X4
justifies a Debian 3.0 who is to say that KDE3 doesn't justify Debian 4.0?

If we say no (in this hypothetical example) we are saying something about
debian's view of the relative importance of packages. I think avoiding this
is best.

Simply: a new release! --- a new number.

> > > Revisions of the stable release, which currently use the "X.YrZ" form of
> > > release number, would instead append and increment a minor number to the
> > > release.  The first revision of sarge would therefore be
> > > "Debian 4.1 (sarge)", the second "Debian 4.2 (sarge)", and so on.
> > 
> > This makes more sense too. People often ask me what the rZ is about. I
> > explain it is like a patch-level and they understand. However, this is not
> > accurate. I believe that because people ask this question and that because
> > the easiest way to explain it is wrong, we should conclude that this system
> > is also not working.
> 
> Isn't it a patch level? It's merely very minor updates to an existing
> distribution.

It is very close. Some people probably disagree about this.
Hence, I retract that point content that my other arguements are strong. :-)

People asking about what "rZ" means is still a bad sign.

> > > I think that sarge is the most appropriate time to adopt this proposal.
> > > "Debian 4" would naturally follow on from "Debian 3.0" without causing
> > > any confusion, while at the same time implying the change in release
> > > number scheme by dropping the ".0" suffix.
> > 
> > I whole-heartedly agree with this!
> 
> I don't. If we don't remember or use the version numbers now, why do we care
> what they look like?

Because numbers tell us what is newer.
Because some users refer to debian 2.0 and I have no idea what they are
talking about.
Because it is printed on our CDs so it shouldn't be meaningless
Because the numbers are used in /etc/apt/preferences

---
Wes

Attachment: pgpFX_ChUBhNh.pgp
Description: PGP signature


Reply to: