Debian Lenny frozen

Waiting a little bit for a major release isn't a good idea because there are lots of major
releases all the time for lots of different projects.

Each major release is almost always delayed. Each major release isn't going to work perfectly
when it hits the public so will need some maturing time.


For example:
> Does anything seriously depend on OpenOffice.org, for example?

_I_do_. 

Lets assume for a second that I have to deal with a office that depends on OO.org as part of
it's workflow.

This means I have documentation I have to maintain. This means that scripts and macros and
databases and all sorts of little things that I and dozens of other people have built up
around OO.org. Something like MS Office or OO.org isn't just for making spreadsheets or
editing word documents. They are programming environments in their own rights and are used for
all sorts of important office tasks and form the basis for important applications.

This, our personal time and effort, is more important then, say, improved Word compatibility.
If I need word compatibility it's not difficult to have a machine somewhere running Windows
with the latest version of OO.org so that I can test document conversions with MS Office.

As you can imagine delaying a release of a OS in order to shove some new version of OO.org
with a whole new sets of bugs and invalidating at least a portion of the documentation I've
made up as well as macros and other things that people have made is not a attractive concept.

------------------------------


If you want a newer version of OO.org stuck in Lenny it's not a big deal to do yourself.

I don't like apt-pinning because if you do that for something that lots of dependencies it's
easy to pull in a bunch of files from a newer release basically ruining the point of having a
stable release in the first place.

I like things like Backports.org and I like using deb-src and recompiling packages from newer
systems for older systems. 

This way the stability of the entire OS can be maintained by Debian while I can have my little
niggling task-specific 'must have' without disrupting it for everybody else.

--------------------


Of course it would be nice when the average Linux distribution person or programmer realizes
the significant advantages of binary distribution from upstream and put effort into making it
work better.

They do that already, libraries were frozen earlier.

The point of stable release is that all the software have been successfully tested together to
be working in a rather fluent way. That's why it's also risky to provide newer versions of
applications into a released stable release.

Maintaining both old and new version of some software has been done before, like with Octave,
but of course it requires a _lot_ of maintenance. There's eg. hardly enough resources to do
single OpenOffice.org version maintaining/testing at the moment, so it would definitely
require several committed volunteers to even think of eg. OOo 3.0 kind of cases provided as an
alternative version.

http://backports.org/ is there to provide some packages a bit more unofficially, ie. they do
not come with the usual kind of guarantee of quality.

What I often do is backport myself.  Add a deb-src line from testing or unstable to the stable

machine then do

apt-get build-dep package-foo
(any libraries or other tools complained about in this step need to be backported as well.
This 
process can get a bit recursive.  If it turns out I'm going to be building 4 or 5 support
libraries or 
worse then I won't bother.)

mkdir package-foo
cd package-foo
apt-get source package-foo
cd package-foo-<version #>
dpkg -rfakeroot -b
dpkg -i ../package-foo-<ver#).deb

Shouldn't building LSB compliant packages eliminate the need for pulling in other
dependencies?