[CentOS] EL 6 rollout strategies? (Scientific Linux)

Sun May 15 22:12:56 UTC 2011
Gordon Messmer <yinyang at eburg.com>

On 05/15/2011 02:23 PM, Ron Blizzard wrote:
> Obviously I missed the part where I (or someone) said (or claimed)
> that 6.1 could be done in a month.

Well, that is where this branch of the thread began.

http://lists.centos.org/pipermail/centos/2011-May/111443.html

Ljubomir Ljubojevic began the branch with the expectation that there 
would be not more than a month between C6.0 and C6.1.  I believe that he 
misspoke, and probably meant that there would be not more than one month 
between the upstream release of RHEL 6.1 and the release of C6.1.

http://lists.centos.org/pipermail/centos/2011-May/111473.html

Dag Wieers replied and asked if there was any reason to believe that 
this would actually happen.  I think this is a perfectly valid question. 
  If 5.6 could not be done in a month, why would we expect that 6.1 
would be?

> What does a month have to do with anything?

It was a time frame posed on the list.  I don't believe it has any 
significance beyond Ljubomir's suggestion that 6.1 could be completed in 
that amount of time.  It seems unlikely.

> There is a certain amount of time required to rebuild the "upstream"
> releases. Whatever that amount of time is, CentOS and SL seem to
> require about the same number. So I'm trying to figure out... why is
> CentOS attacked so much for taking too long? -- whereas SL is lauded
> as the "go to" distribution?

Well, CentOS is generally attacked for taking a long time because users 
have had no visibility into the process.  Most people make what I would 
think is a perfectly reasonable request: not that the distribution is 
available immediately and not that a specific release date is given and 
kept, but that information about the tasks to be completed is published. 
  The process around building CentOS has traditionally been very 
secretive, which makes the name "*Community* Enterprise OS" seem very inapt.

SL is not so much lauded, I think, as discussed right now because users 
who want a rebuild of RHEL 6.0 have no other option.  There is no beta 
of CentOS for them to install and review (which would make contribution 
significantly easier).  Users are trying to figure out whether or not it 
makes sense for them to wait for CentOS 6.0 or use something that's 
available now.  Because they have very little insight into the process 
or progress that has been made, they cannot easily evaluate that 
question.  It is reasonable for this to produce a great deal of anxiety.

> As I showed in the list of release dates, CentOS and SL have almost
> always been fairly close (CentOS usually a little quicker). So why the
> claim that CentOS is getting worse on release dates? (General claim,
> not specifically yours.)

Look at wikipedia's page describing CentOS.  They include a column for 
the delay between the upstream release and CentOS's.  For the 5 series, 
it looks like:

Release Delay
5 	28d
5.1 	25d
5.2 	34d
5.3 	69d
5.4 	49d
5.5 	44d
5.6 	85d

Almost every release in the 5 series took longer than the initial 
release for 5.0.  Even if you ignore the release of 5.6, there is a 
generally upward trend in the amount of time taken for each release. 
How could anyone reasonably claim that CentOS is NOT getting worse on 
release dates?

I can't even begin to comprehend the logical failure behind the idea 
that because SL and CentOS are keeping up with each other that CentOS is 
not getting worse.  Again, Dag interjected only to ask why any 
reasonable person would expect 6.1 to take only one month when 5.6 took 
three.  The fact that there is a general trend toward longer release 
delays supports that question.

> I see no pattern in the release dates to
> indicate CentOS is generally falling behind SL.

That's fine, but that's not what's being discussed.