Hello.

>> The only added value of SL (comparing to CentOS) as
>> I see it now is:
>>
>> 1). Adding to anaconda the 'sites' functionality
>> 2). Adding AFS client (plus few more packages).
>>
>> (well, there are few more customizations but these are
>> minor I would say)
>>
> 
> The one other added value I see is our "tweaks" or SL rpm's.  They might
> be considered minor, but those minor things save alot of work.

Right: but this can be easily solved via an add-on repo.

> Connie and I have tested out a couple repository variations.  While we
> haven't tried what you said, the new anaconda has shown itself to be
> very flexible and I believe what you said it quite possible.

Well, my knowledge comes from looking at Fedora Core 6
plus some anaconda sources: I believe that we could use it
'as-is' .. but this is to be tested of course (hence again:
*now* is the right time..)


> 
>> Adopting the above would permit us to spend more
>> time on something which was supposed to be one
>> of our main goals ... and what is not really
>> achieved: adding 'scientific' packages to the
>> distribution...
>>
> 
> As one of the people pushing for this, I agree with this statement. That
> was one of the reasons I had mentioned this to Connie.  CentOS already
> has someone taking Fedora Extra's and packaging it for CentOS. Many
> scientific programs are already in this extra's repository, and it would
> be better for me to throw my efforts in with that repositories
> maintainer that do it on my own.

I see we agree on that !

> 
>> What is you opinion about re-basing on CentOS ?
>>
>> Pros ? Cons ?
>>
[..]

> 
> - They also have a much better bugzilla.

Indeed. What we have is not really used, is it ?

> 
> - Not having as large a control over the base OS.  But that is really a
> Pro also, because that also means not having as large a responsibility
> over the base OS.

Well: Since the goal for base CentOS is just recompilation of RHEL
(which is also the main goal of SL) ... I do not see as a big problem..

As for the updates coming at 'random times': I believe labs update
their users from their own (mirrored) distribution servers rather than
from central ones: so the control over that would still be there ...
> 
> - CentOS doesn't like to let users "sit on a release".  Now this might
> be changing since RedHat is finally letting users do that.  But that
> might be a concern that several scientists have.  You know, as well as
> I, that despite all explanations and reasoning, they don't want
> *anything* to change (and at the same time must have all the latest
> bleeding edge stuff).

No they don't, but we at CERN do not give them the choice:
Once SLC X.Y+1 is released, yum repositories change from X.Y to new
one.. and everyone is updated immediately (ie: once per week).
(exception being computer center batch farms where we target to have
an update once every 3-4 weeks only in the future)

.. surprisingly enough: We haven;t heard any complaints about that in
last 2.5 years ..


> This is one point I think we'll have a hard time with when talking to
> some of the user communities.  We'll have to come up with some
> reasonable solution.

Question: Is it *really* necessary ? ~ 9000 systems updating for
SLC3/4 from linuxsoft.cern.ch (half of these are outside CERN) ... -
and no complaints/requests about it ...

- If the answer is yes for a given user community: this community
could just pick what they want from update stream for their own
update repository ...

Thanks for replying

Cheers

Jarek

__
-------------------------------------------------------
_ Jaroslaw_Polok ___________________ CERN - IT/FIO/LA _
_ http://home.cern.ch/~jpolok ___ tel_+41_22_767_1834 _
_____________________________________ +41_78_792_0795 _