Re: official backport pool (Was: [VAC] 25 july - 26 august)
- To: "John H. Robinson, IV" <firstname.lastname@example.org>
- Cc: email@example.com
- Subject: Re: official backport pool (Was: [VAC] 25 july - 26 august)
- From: Jonas Smedegaard <firstname.lastname@example.org>
- Date: Sat, 26 Jul 2003 20:42:28 +0200
- Message-id: <[🔎] email@example.com>
- In-reply-to: <20030726171104.GB4785@ucsd.edu>
- References: <20030723122738.GC27503@traveler.cistron.net> <20030724075056.GC9742@mykerinos> <20030724082100.GB4573@riva.ucam.org> <20030724084138.GA12479@iliana> <firstname.lastname@example.org> <20030724113650.GA3246@iliana> <email@example.com> <20030726064222.GA30539@snoopy.apana.org.au> <firstname.lastname@example.org> <20030726171104.GB4785@ucsd.edu>
On Sat, 26 Jul 2003 10:11:05 -0700
"John H. Robinson, IV" <email@example.com> wrote:
[referring to non-private discussion on debian-private about the
creation of an official "backport pool"]
> i'm not so much interested in in official backport pool as i am
> interested in an ``official'' central listing of all such backports.
> this helps to reduce duplication of effort. if i have already
> backported alsa (which i have), why should someone else go through the
> same efforts?
> some important information to know:
> package, version, backporter, location (apt deb line), variations from
> official package (eg: missing functionality, modified build-deps)
> a nice thing would be to take the version that was backported, and
> drop down a half-step, like adrian bunk did with his 2.4 kernel on
> potato backports.
> feel free to quote me entirely off-list; there is nothing private
> here. perhaps -project would be a good place for this.
Hmm - I dare quote you on-list, but another one than -private ;-)
I am not quite sure I understand what you mean by "half-step", but if
you mean bumping down the version number a bit then I certainly agree.
Here's the rules I have come up with for my own unofficial packages on
* First and second part of Debian version number is reserved for
official Debian use (first one for maintainer and second one for
* Third part is a hint of origin.
* Fourth part is actual revision number.
* Keep as tight to the version number of the original Debian package as
possible, while bumping up or down depending on the nature of the hack
(simple backports are bumped down, while added features are bumped up).
The above rules makes newer official Debian releases override the (then
older and possibly outdated, wether they brought new features or not)
It makes it sometimes a bit trickier to backport- dependencies(within
the packages in one source package or in other dependent packages) must
sometimes be similarly lowered as well.
I have heard complaints that it made the version number too long - but
as I see it the version number is a tool, and we ought to use it -
properly - when needed!
It would be quite useful if we defined rules like these and made them
somewhat official - so that we could ask blackdown.org, skolelinux, LTSP
and others to follow them as well.
ppp 2.4.1.uus-5.0.jones.1 (ppp 2.4.1.uus-5 with PPPoATM support)
php4 4.2.3-13.99.jones.1 (php4 4.2.3-14 backport)
ext2resize 1.1.17-3.0.jones.cvs20021227 (ext2resize 1.1.17-3 updated
libpng3 184.108.40.206-3.99.jones.2 (libpng3 220.127.116.11-4 backport, and in 2nd
attempt corrected shlibdeps)
gjukebox_4.1pre4.cvs20021129-0.0.jones.1 (no official Debian package)
* Jonas Smedegaard - idealist og Internet-arkitekt
* Tlf.: +45 40843136 Website: http://dr.jones.dk/
- Enden er nær: http://www.shibumi.org/eoti.htm