[ATrpms-devel] ATrpms stats for deciding on EOLing distros
Jarod Wilson
lists at wilsonet.com
Fri Jul 28 17:01:10 CEST 2006
On Monday 17 July 2006 06:40, Axel Thimm wrote:
> the EOLing of some distros comes up often, and ATrpms is currently
> swamped by supporting so many distros. So I created some stats out of
> the first days of July which can be used for deciding on what to do.
>
> ATrpms alone
> ============
>
> fc5-i386 36%
> fc4-i386 30%
> fc3-i386 11%
> fc2-i386 7.4%
> fc5-x86_64 5.2%
> el4-i386 3.8%
> rh9-i386 1.6%
> fc4-x86_64 1.5%
> fc1-i386 1.1%
>
> ATrpms + non-ATrpms
> ===================
>
> fc4-i386 38%
> fc5-i386 30%
> fc3-i386 12%
> fc2-i386 5.9%
> fc5-x86_64 5.1%
> rh9-i386 2.0%
> el4-i386 1.5%
> fc4-x86_64 1.4%
> fc1-i386 1.1%
>
> o x86_64: About 1 to every 7 FC5 users and 1 to every 20 FC4 users.
Interesting stat. (I'm happily almost nothing but x86_64 these days).
> o FC5 and FC4 almost on the same level still.
Heh, I've probably contributed to that by not updating a certain document for
FC5 still...
> o Rather few EL users.
How heavily do you think MythTV plays a role in that fact? RHEL5 should be a
better MythTV platform than RHEL4 (better DVB support possible), might even
shift to recommending it myself...
> FL plans to drop FC1 and FC2 by the end of July and RHL9/7.3 between
> October and December (although since there wasn't yet an announcement
> these dates may shift).
>
> Therefore I'd like to suggest to do the following:
>
> o keep as active distributions EL3-4 and FC3-6
> o demote the others in a staged way as
> 1) no builds for new (not updated) packages
> 2) updated packages failing to build under these old distros don't
> get fixed, the distro simply gets this package cut.
> 3) no updates at all
> 4) removal from main master site
>
> Comments?
Sounds good to me.
> Dates?
I'd more or less just mirror the FL schedule to start the distro demotion
process. Perhaps give people a few months (3?) of the distro in the demoted
state, then proceed with step 4.
--
Jarod Wilson
jarod at wilsonet.com
More information about the atrpms-devel
mailing list