Jeremy C. Reed wrote:
> Does Debian really support old Debian releases? For example, Potato was
> supported for a long time. But as of June 30, it is now unsupported.
> 
> Debian also has had problems with upgrading from old versions to new
> versions. It is getting a lot better now. (I should rephrase that:
> Debian offers the most reliable and easiest package upgrade mechanism I
> have used.) But over the past 6+ years of using Debian, I have spent a lot
> of time manually updating versions of apt-get, dpkg (including having to
> revert to older versions), et cetera, so the upgrades would work
> correctly.
> 
> (Another example is OpenBSD: it has a consistent release every six months
> and then the two releases previous -- one year old -- version becomes
> dead.)
> 
> Having an end of life of old versions is definitely a good idea.
> Developers (volunteers) should spend their time on new or recent code, in
> my opinion.
> 
> Also, if the end-of-life'd Red Hat is really good, of course, someone
> could spend their time keeping it alive and up-to-date (and fork a new
> project/distro out of it). But it doesn't really seem worth it.

I am not saying I blame RedHat or anything.  Surely products have to have an 
EOL, but when RH9 came out its EOL was only tenish months away.  That week I was 
installing a web server and replacing a Win2k domain controller with a linux 
box.  I had used redhat for about 5 or 6 years by then but when I realized that 
the latest RedHat would end not even a year from when I was installing.  I 
decided I didn't want to be forced to reinstall in less than a year.

As for upgrading, its always something I had avoided, I used a redhat upgrade 
years ago and then decided that I would not attempt upgrades again.  So 
reinstall the newer version was really the only option.  I have been using 
dist-upgrade lately with much success (testing -> unstable), granted these boxes 
are just desktops.

Backporting security patches is not a valuable use of my time at the moment. 
Paying for security patches (well convenient ones at least) is not something I 
am fond of doing.  So I will be avoiding it somewhat.  But, since RH is the 
(well perhaps just one of the dominant) dominant commercial distribution I need 
to keep up to date on it.  I haven't figured out how though.

Actually this give me an idea for a presentation.  Building RPMs and DEBs.  I 
would like to see a presentation on that.  Or figure it out and give one.


cheers,
Austin