Debian successor to Lenny has been named [LWN.net]

what is the attraction of confusing code names?

Posted Sep 1, 2008 19:16 UTC (Mon) by jhoger (guest, #33302) [Link]

Seriously?

http://www.debian.org/releases/

-- John.

Seriously?

Posted Sep 1, 2008 21:51 UTC (Mon) by Felix_the_Mac (guest, #32242) [Link]

Yes, seriously.

Of course you can look it up, but that doesn't stop it being confusing when you come across it in a random article.

Perhaps you would like to rename the gear positions in your car:

1st - Sheila
2nd - Sally
3rd - Sheena
4th - Sandy

:-)

Seriously?

Posted Sep 2, 2008 4:00 UTC (Tue) by jhoger (guest, #33302) [Link]

I think that would work fine since my list is short enough, but I would run out of names quickly since I got married fairly quickly after I started dating, if one can call that. Anyway.

My point is, names are just names. There is no sense to them. They, like all names are mnemonic devices that stand in for something more complex. I find names easier to remember than numbers. Sure numbers compare better, but you still have to have context to make sense even of numbers. Version 3.1 in a vacuum doesn't mean any more than Sarge. So the number, like the name only means something in context.

Alternatively, think of it in terms of information hiding. Sometimes release names are equivalent to a variable name or handle. This way you can refer to a release even if you don't know what the final version number will be or if you are releasing many point releases afterward.

Seriously?

Posted Sep 2, 2008 10:52 UTC (Tue) by Cato (subscriber, #7643) [Link]

The part that non-Debianistas find hard is knowing how the various codenames are sequenced - is Etch before or after Lenny, for example? This is important if you stumble on a forum posting for Debian when trying to configure another distro, and would like to know if the information within is of a similar vintage to your version.

Ubuntu uses codenames that are alphabetically sequenced, so as long as you know the latest release it's easy to tell the Breezy is a very old version. (I know Warty Warthog was older but that's before they got onto the alphabetical sequence I think).

what is the attraction of confusing code names?

Posted Sep 1, 2008 22:03 UTC (Mon) by amacater (subscriber, #790) [Link]

Ask somebody about Debian 1.0 sometime :) [There never was a 1.0 - but
someone from ?? Infomagic ?? packaged up a 0.97 or so that wasn't ready for
release and sold that as 1.0 - hence Debian skipped a version to 1.1] Hence
the codenames _until_ release which people carried forward.
The 4.0r4 (where r is short for revision) comes from the fact that CD
vendors wanted to put out whole numbered versions: Debian wanted to make it
clear that there had been changes/security fixes but that "4.0" wasn't
obsolete. As ever, you can update any 4.0rX to 4.0rX+ just by using
aptitude update / apt-get update ; aptitude dist-upgrade / apt-get
dist-upgrade.

All embedded in FAQs and Debian mailing lists

what is the attraction of confusing code names?

Posted Sep 1, 2008 23:25 UTC (Mon) by stevenj (guest, #421) [Link]

This explanation doesn't hold up to scrutiny. No amount of codenaming will prevent an unscrupulous distributor from packaging up unstable and calling it "Debian 5.0" or whatever they want; your only recourse is public shaming and (possibly) trademark law, regardless of how you name/number releases.

Arguably, using confusing codenaming only makes it easier for unscrupulous distributors to put out their own official-sounding releases. If I release "Debian zurg", it doesn't conflict with any official Debian release names, but it sounds as official as any of the others.

It is perfectly possible to assign version numbers for upcoming releases. e.g. gcc gets along perfectly well doing this (modulo one minor glitch with Redhat releasing an unofficial "gcc 2.96", but see above: codenames wouldn't stop this).

(I don't doubt that there are reasons why people choose codenames for their project versions. They just don't seem like rational reasons.)

what is the attraction of confusing code names?

Posted Sep 2, 2008 1:39 UTC (Tue) by bignose (subscriber, #40) [Link]

> (I don't doubt that there are reasons why people choose codenames for their project versions. They just don't seem like rational reasons.)

Here's one: the people working on it need to call it *something*, but at the time work begins they don't know what version number it will have.

Version numbers, which are chosen much closer to the release date, generally indicate how radical the changes are in each release, which isn't known ahead of time.

So, picking a name allows work to proceed before knowing what changes will be in the release.

what is the attraction of confusing code names?

Posted Sep 2, 2008 2:39 UTC (Tue) by stevenj (guest, #421) [Link]

Most projects, when they are not ready to set a version number for the next release, simply refer to the "development branch" of the project or some similarly clear descriptive name, rather than naming it "Shirley".

Unfortunately, code names live on for long after the official version number is assigned.

what is the attraction of confusing code names?

Posted Sep 2, 2008 16:16 UTC (Tue) by njs (subscriber, #40338) [Link]

> Most projects, when they are not ready to set a version number for the next release, simply refer to the "development branch" of the project or some similarly clear descriptive name, rather than naming it "Shirley".

Debian does that too ('unstable', 'testing', 'stable'). But the names provide an additional -- technical! -- capability: you can refer to and track a single release even as it moves from being 'testing' to 'stable' to 'oldstable'. So you can, for instance, upgrade to the next stable release during the freeze, and then stay on that stable release once it's released, without any pain. Or you can put your stable boxes on etch, and they'll stay there even after lenny is released.

That said, and speaking as an exclusive Debian user for over a decade, I personally can never remember which codename corresponds to current testing/stable/oldstable and have to look it up every time.

what is the attraction of confusing code names?

Posted Sep 3, 2008 0:23 UTC (Wed) by bignose (subscriber, #40) [Link]

> Most projects, when they are not ready to set a version number for the next release, simply refer to the "development branch" of the project or some similarly clear descriptive name

A name such as "development branch" is clear, but it's descriptive only during a given time period. Once that description no longer applies (e.g. the branch becomes the "stable branch"), one *still* needs to call it something. Re-naming is much more difficult once it has a body of discussion using the existing name, so "development branch" is a poor choice of name.

Further, when something *else* becomes the "development branch", mass confusion is sure to ensue over *which* "development branch" is being referred to in past and current discussion. Some other way of distinguishing them is needed; since this is entirely predictable, it would be foolish not to distinguish it ahead of time, before the name takes hold in discussions.

Fortunately, choosing an arbitrary name that does *not* attempt to describe such time-sensitive state avoids all these problems.

> Unfortunately, code names live on for long after the official version number is assigned.

That may be unfortunate, but it's a fact. The "code name" approach acknowledges and works with that fact.

This applies not only to software branches, but to names of many other things as well <URL:http://www.ietf.org/rfc/rfc1178.txt>.

what is the attraction of confusing code names?

Posted Sep 2, 2008 2:06 UTC (Tue) by BrucePerens (subscriber, #2510) [Link]

The explanation is accurate. The code names are Toy Story characters, I was working at Pixar at the time. Debian can use trademark law to prevent unscrupulous folks from labeling something "Debian 5.0" and selling the result.

what is the attraction of confusing code names?

Posted Sep 2, 2008 2:17 UTC (Tue) by stevenj (guest, #421) [Link]

Bruce, I wasn't saying the explanation was inaccurate. Just that it doesn't seem to make sense. Debian can use trademark law whether it uses inscrutable codenames or version numbers, so codenames apparently serve no purpose in that regard.

Obviously, the tradition of Toy Story names has turned into a cultural thing with Debian that's taken on a life of its own. But I think it is counterproductive in a larger setting, because it is impossible to tell what the sequence of versions is unless you look it up or memorize it. Moreover, it sets a bad example — nowadays, Debian is not the only free-software project that uses inscrutable code names for their versions.

Code names are a code, a cipher — the purpose of a cipher is to obscure meaning. Why should free software projects be obscuring the meaning of their versions?

what is the attraction of confusing code names?

Posted Sep 2, 2008 2:45 UTC (Tue) by BrucePerens (subscriber, #2510) [Link]

I agree that it's difficult to figure out which version of Debian is after which, once you got past "Woody" and "Buzz" (the ones I named).

The important point is that nothing should get a number before it's released, because the temptation to burn a CD of the mirror and call it "Debian X.Y" is too great.

what is the attraction of confusing code names?

Posted Sep 2, 2008 9:34 UTC (Tue) by eru (subscriber, #2753) [Link]

labeling something "Debian 5.0"

Coincidentally I bought yesterday the September 2008 issue of Linux Magazine, which boasts "debian 5.0" as its cover DVD. The article inside does explain that it is the "July 7, 2008 Debian 5.0 Lenny "testing" version". A 2-sided DVD with 32-binary on one side and the 64-bit version on the other.

As to the names being confusing, I wonder if it has occurred to people criticising the scheme that it could be part of the "cultural identity" of the Debian community, which after all is a volunteer organization? Bit like flags, anthems etc. in other cases that people outside the community in question may find silly, but it is not their business to complain. (Just guessing. I have very little knowledge of Debian. Actually I bought the above-mentioned magazine to finally test drive it a bit).

By the way, what will Debian folks do when they run out of named "Toy Story" characters? Move to other Pixar films?

what is the attraction of confusing code names?

Posted Sep 2, 2008 15:08 UTC (Tue) by rfunk (subscriber, #4054) [Link]

1. Name characters themselves. I'm pretty sure "Squeeze" was just "the aliens" in the movie.
2. Wait for another Toy Story

what is the attraction of confusing code names?

Posted Sep 2, 2008 10:16 UTC (Tue) by dark (✭ supporter ✭, #8483) [Link]

The important part of the explanation is that the distributor wasn't unscrupulous. The distributor simply mirrored the directory called "debian-1.0" and sold it as Debian 1.0. The directory was called debian-1.0 because that's what the version number would be once it was released, and renaming the root of a directory tree just before release time is just asking for trouble. (It would take manual action on every mirror to avoid re-downloading the whole tree.) The distributor was sloppy by not checking whether the stuff in there was ready for release, but it wasn't deliberate.

Thus, to prevent this kind of accident in the future, the project decided to use code names for the directories. These code names stuck and became shorthand for the releases. I personally find them much easier to remember than version numbers. Names have personality.

P.S. I don't remember if "debian-1.0" was exact. It may have been "dist-1.0" or just "1.0", but the version number was in there.

what is the attraction of confusing code names?

Posted Sep 3, 2008 2:26 UTC (Wed) by rahvin (subscriber, #16953) [Link]

Case in point. I don't follow Debian Versions very closely. In fact not at all to be honest. As a subscriber to Linux Pro Magazine I recieved with the latest issue a Debian 5.0 CD. Thinking the release had happened I went to play with the CD and noticed afterward that it was a snapshot of the unstable distribution borking completely my new firewall install.

Thanks a lot guys!

Re: confusing code names

Posted Sep 2, 2008 1:34 UTC (Tue) by ldo (subscriber, #40946) [Link]

I agree. I find it really hard to remember the Debian ones. For some reason, the Ubuntu ones are much more memorable--perhaps because they're alliterative, and also they (now) go in alphabetic sequence.

I think the Debian folk should lose their ongoing in-joke. Nobody else gets it.

what is the attraction of confusing code names?

Posted Sep 2, 2008 14:28 UTC (Tue) by NAR (subscriber, #1313) [Link]

Do you really need to know that potato was released before or after bo? It's not like Debian releases a new version every other week... There's a stable version (etch), a (relatively) soon to-be-stable version (lenny), an unstable (sid) - the rest is history. Of these three, the unstable has a fixed name (like rawhide for a certain other distribution), so most of us only need to remember two names.

what is the attraction of confusing code names?

Posted Sep 4, 2008 3:02 UTC (Thu) by k8to (subscriber, #15413) [Link]

As a debian user of some 6 or so years, I agree completely. The names are useless, obfuscating, and juvenile. It's fine if Debian people want to use cute names, but if the cute name is ever used, it should follow the version number, not replace it. For example: "Debian 3.5 (potato)"

Nearly all Debian discussion is always about these silly tags, which means when reading these discussion logs as a non-expert one must always translate these nonsense strings into something that can at least be ordered. A simple scalar (version numbers) as an identifier for a software release is not so great, but unordered strings is worse.

what is the attraction of confusing code names?

Posted Sep 4, 2008 6:50 UTC (Thu) by amacater (subscriber, #790) [Link]

I helped test the Debian CDs for the last release. One day before the
release of Etch as "stable", Debian made a wrap-up release of Sarge to
incorporate all the last security fixes as an active distribution.
Two "Debian stable" releases in 24 hours :) How do you, as a user, know
what's happened in that situation when somebody says "Debian's made a
stable release"?

If you, as a user, know that you're tracking Etch and that today it's still
in testing, when it becomes a stable release you need to do nothing - and
the FTP masters need only change one link. [Two, actually, since then Sarge
becomes oldstable].

what is the attraction of confusing code names?

Posted Sep 4, 2008 10:37 UTC (Thu) by cortana (subscriber, #24596) [Link]

"Debian 3.1r7 and 4.0 were released". There are probably many other ways to phrase the statement.

IMHO the practice has continued because there is no policy for deciding what the version number of a release will be. The archive has to name its directories/symlinks something, so the code names have stuck around.

If it were up to me, the version numbers would be major.minor and major would always be incremented when a new stable release is made; the current weird-looking 'rX' prefix would become the minor release number.

what is the attraction of confusing code names?

Posted Sep 11, 2008 21:10 UTC (Thu) by anton (subscriber, #25547) [Link]

if the cute name is ever used, it should follow the version number, not replace it. For example: "Debian 3.5 (potato)"

I totally agree with that, but coming from another direction. Looking at some server, I want to know what it's running, so I cat /etc/issue:

Debian GNU/Linux 4.0 \n \l

WTF? Nobody ever uses these numbers, how should I remember if this is sarge or etch? Even worse, for a machine that was on a testing distribution some time ago, I got something like

Debian GNU/Linux testing/unstable \n \l

Which is not very helpful, because what used to be testing might now be stable. Fortunately they fixed this, and now I see:

Debian GNU/Linux lenny/sid \n \l