Les Mikesell wrote:
Actually, I didn't expect it to be such a controversial issue. SoThe most reocurring reasons to not respin the isos (that I recall from previous discussions):
far no one arguing against it has come up with any advantage to
anyone in continuing to distribute known/fixed bugs. It is not that
big a deal to me because I always point new installs at a proxy server
and pull updates immediately, but most of the other people I know
who run Linux don't bother to do that. The only real problem for
me has been that for every version from RH9-FC2 I have one or more
kinds of machines that will not install due to hardware problems
(each machine will run one or more versions, but fails with at least
one - and oddly, most of them were purchased loaded with RH linux).
But, I'd rather switch than fight so I've been installing Centos 3.4
on all of those. I suspect though, that if the updates were backed into the isos the install problems would have been fixed.
1 - lots of work (both legal and technical) to prepare the isos and push them to mirrors (also , if the mirrors were meant to keep the original isos and the respins , there's the disk space issue)
2 - probability that the respins would be less tested (since many people who test the test releases like to live on the edge and after the final release , enable the development repository and not the updates-testing)
3 - problematic on the support point of view... with only 1 official iso set per arch per release , it's already hard to help some people sometimes. We end up with lots of questions without any mention to necessary details , like version of the package , which release , etc... With respins , we add another variable. If the same baseline is kept , at least everyone knows what version was in the release before updating (which is handy sometimes)
That's all that I recall now...
-- Pedro Macedo
