Hi,

Some short remarks and questions from my side:

I really think that, the amount of work I have to do is independent
whether we at PSI choose SL5 or CentOS5.

We made really good experience with SL in the past. So why should we
move to CentOS?

Will we have the same good contact to CentOS people as we have to SL
people? Who is behind CentOS?

For us it's critical that the support of SL3 and SL4 will continue. So
don't give up SL3 and SL4!

At PSI we have to distribute SL/CentOS 5 quite soon, because some of our
big user groups will jump directly from SL3 to SL5 and they need at
least one year of overlap. Therefore we already had to extend the
support of SL3 beyond the end of the official SL3 support. In this
respect we are a little more under pressure than Cern.

I really like the possibility to stay on, let's say SL 305, and not to
go every time to the latest release. So, will CentOS keep independent
releases in CentOS 5? Like RHEL will do.

We can definitely not start with SL 5.0 and then switch to CentOS 5.1.
In other words, if SL cannot ensure that they will also release SL51,
SL52, etc. We will (and have to) move to CentOS 5 even if we don't want to.

We would like to stay with SL, since we don't really see any big
advantage for us going with CentOS. However, CentOS could be our Plan B,
if SL cannot afford to release and support SL5.

Cheers,

    Urs




Jan Iven wrote:
> (Joining late..)
>
> On 28/02/07 23:34, Troy Dawson wrote:
>> Hi,
>> This is my last e-mail for the day.  I'm going to go home and think, but
>> here are some thoughts.
>>
>> First - I think that for at least S.L. 5.0 we should do it completely
>> independent of CentOS.  That is for both CentOS's and SL's good.  RHEL5
>> has so many different changes in the way they are doing things, that SL
>> and CentOS are probably going to do very different things.  I think this
>> is a good thing, so that we can look at both ways of making a
>> distribution, and determining which, if any, is best.
>> So, I'm planning on forging ahead with S.L. 5.0 on our own, no matter
>> what we decide for S.L. 5.1.
>
> At least at CERN SL(C)5 is currently not being pushed by any of our user
> communities (even if we have sometimes trouble getting laptop/desktop
> hardware that is working with SLC4), this is due to a gradual lockdown
> before the LHC startup.
>
> Which means we have time to prepare SLC5 as we see fit, and are not
> driven by "need to release <X weeks after TUV is out". Like Jarek, I
> believe the best time for a change would be between major releases, not
> during the "stable" lifetime.
>
> But I agree that looking at the "5" era from two different angles is
> going to make the final_and_eventually_unified SL5.x a better product. I
> hope that the initial divergence (SL "sites" vs addon "repos") is small
> enough to be overcome eventually.
>
>> Cons:
>> CentOS does *not* have an open development cycle.  There are no beta's.
>>  For the average user, it just comes.
>
> This might be true, but we would still have the freedom to look at beta
> releases from TUV ourselves - much as we do now (the rebuilding
> infrastructure wouldn't go away, since it might always be required for
> "urgent" security updates anyway). And somebody inside CentOS must be
> looking at betas, I would guess that if we contribute to CentOS in a
> noticeable way, it would not be too hard to join that circle.
>
> Regards
> Jan