CentOS-4.4 yumconf

"JP" == Jim Perrin <jperringmail.com>

Jim,

    JP> This was discussed before implementation, in the
-devel
    JP> channel on IRC. The full ramifications weren't
felt
    JP> initially because it worked fine for stock
centos. If
    JP> you'd like to participate in the discussions,
you're
    JP> welcome to join the irc channel.

Like, I suspect, most people using CentOS, I'm not paid to
work on
the OS, and I don't have time to hang out on IRC on the off
chance
that someone will bring up something that might completely
screw
up my systems down the road.

To be frank, saying, ``Oh, we talked about that on IRC,'' is
about
as useful as saying, ``Jennie, Bob, and I talked about it
over a
drink at the Boo Bar.''

IRC is a great place to work out implementation details and
maybe
even do some thought experiments to imagine what the impact
might
be, but mailing lists are a much better place to have real
discussions, especially when your changes might impact
people
outside the tiny circle of people who frequent the IRC
channel.

   Claire

*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-
*-*-*
  Claire Connelly                              cmcmath.hmc.edu 
  Systems Administrator                          (909)
621-8754   
  Department of Mathematics                 Harvey Mudd
College
*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-
*-*-*
_______________________________________________
CentOS-devel mailing list
CentOS-develcentos.org
http://lists.centos.org/mailman/listinfo/centos-devel

On Wed, 2006-10-11 at 18:22 -0500, C.M. Connelly wrote:
> Like, I suspect, most people using CentOS, I'm not paid
to work on
> the OS, and I don't have time to hang out on IRC on the
off chance
> that someone will bring up something that might
completely screw
> up my systems down the road.
> 
> To be frank, saying, ``Oh, we talked about that on
IRC,'' is about
> as useful as saying, ``Jennie, Bob, and I talked about
it over a
> drink at the Boo Bar.''
> 
> IRC is a great place to work out implementation details
and maybe
> even do some thought experiments to imagine what the
impact might
> be, but mailing lists are a much better place to have
real
> discussions, especially when your changes might impact
people
> outside the tiny circle of people who frequent the IRC
channel.

Will come out of my "lurker" closet to chime in
agreement with Claire.
This would seem to be the appropriate list to discuss the
philosophy and
implications of development decisions.  I subscribed
originally because
of testing packages and wanting to provide feedback through
the
appropriate channel.  I have stayed because of interest in
where the
development is headed and the opportunity to see what the
developers are
thinking in a high signal-to-noise environment.

I also don't get paid to develop CentOS, and don't have time
to hang out
on IRC, but can give a bit back to the community by
providing feedback,
bug reports, and occasional help on the users list. 
[Although I find my
opinions are sometimes not shared by the developers, hay,
that's what
community dialog is all about, and I tend to defer to their
opinions and
avoid pointless debates and flame wars.  But I digress...]

Please do consider more discussions on issues such as the
yumconf
approach on this list.

On Thu, 2006-10-12 at 09:09 +1000, John Newbigin wrote:
> I still would like (at a minimum) to see the files put
back into the 
> yumconf package.  Even better would be yumconf-centos
which provides 
> yumconf.

Agree.

Thanks for listening, and for the great OS.

Phil



_______________________________________________
CentOS-devel mailing list
CentOS-develcentos.org
http://lists.centos.org/mailman/listinfo/centos-devel

Claire,

C.M. Connelly wrote:
> Like, I suspect, most people using CentOS, I'm not paid
to work on
> the OS, 

Let me surprise you by saying that noone is paid to work on
CentOS, its 
a complete volunteer only setup. hell, it even costs me
major $funds and 
lots-a-time, to be involved with the CentOS Project, as I
suspect is the 
case with every one of the developers.

We'd like to grow from here, into an organisation that can
throw some 
finances behind some of the work, as well as further expand
on the work 
we are doing. And your continued donations help to that aim.

> and I don't have time to hang out on IRC on the off
chance
> that someone will bring up something that might
completely screw
> up my systems down the road.

Well, in this case - had things been done properly on your
machine, 
nothing would have broken. We've been over this a few times
now.

> 
> To be frank, saying, ``Oh, we talked about that on
IRC,'' is about
> as useful as saying, ``Jennie, Bob, and I talked about
it over a
> drink at the Boo Bar.''

For matters that only Jennie, Bob and I need to talk about,
that is 
indeed a good way to get this done. Just talk about them


> IRC is a great place to work out implementation details
and maybe
> even do some thought experiments to imagine what the
impact might
> be, but mailing lists are a much better place to have
real
> discussions, especially when your changes might impact
people
> outside the tiny circle of people who frequent the IRC
channel.

its not just a tiny circle, we *require* all developers to
be on IRC.

- K

PS: why does your MUA break threads ?

-- 
Karanbir Singh : http://www.karan.org/ :
2522219icq
_______________________________________________
CentOS-devel mailing list
CentOS-develcentos.org
http://lists.centos.org/mailman/listinfo/centos-devel

Heya Phil,

Phil Schaffner wrote:
> Will come out of my "lurker" closet to chime
in agreement with Claire.
> This would seem to be the appropriate list to discuss
the philosophy and
> implications of development decisions. 

agreed.

> I subscribed originally because
> of testing packages and wanting to provide feedback
through the
> appropriate channel.  I have stayed because of interest
in where the
> development is headed and the opportunity to see what
the developers are
> thinking in a high signal-to-noise environment.
> 
> I also don't get paid to develop CentOS, and don't have
time to hang out
> on IRC, but can give a bit back to the community by
providing feedback,
> bug reports, and occasional help on the users list. 
[Although I find my
> opinions are sometimes not shared by the developers,
hay, that's what
> community dialog is all about, and I tend to defer to
their opinions and
> avoid pointless debates and flame wars.  But I
digress...]

you know, I actually appreciate people who make noises when
we do stuff 
that they do and dont agree with. Isnt that what open source
is all 
about ? The ability to bring in lots of eyes on simple
issues ? We all 
gain, all around.

so please dont stop and/or hold back. Flame wars only happen
when one 
side refuses to accept the others point of view. I think
everyone is 
entitled to an opinion, and besides a flamefest every now
and then is 
also a good thing 

> 
> Please do consider more discussions on issues such as
the yumconf
> approach on this list.

yes, we know that we should have made a bit of noise about
this yumconf 
issue before it went out into the wild. I think we've all
learn't a 
lesson here.


- KB

-- 
Karanbir Singh : http://www.karan.org/ :
2522219icq
_______________________________________________
CentOS-devel mailing list
CentOS-develcentos.org
http://lists.centos.org/mailman/listinfo/centos-devel

Hay, Karanbir:

On Fri, 2006-10-13 at 08:40 +0100, Karanbir Singh wrote:
> Heya Phil,
> 
> Phil Schaffner wrote:
> > Will come out of my "lurker" closet to
chime in agreement with Claire.
... 
> > I also don't get paid to develop CentOS, and don't
have time to hang out
> > on IRC, but can give a bit back to the community
by providing feedback,
> > bug reports, and occasional help on the users
list.  [Although I find my
> > opinions are sometimes not shared by the
developers, hay, that's what
> > community dialog is all about, and I tend to defer
to their opinions and
> > avoid pointless debates and flame wars.  But I
digress...]
> 
> you know, I actually appreciate people who make noises
when we do stuff 
> that they do and dont agree with. Isnt that what open
source is all 
> about ? The ability to bring in lots of eyes on simple
issues ? We all 
> gain, all around.
> 
> so please dont stop and/or hold back. Flame wars only
happen when one 
> side refuses to accept the others point of view.

Don't worry about that - just ask anybody who knows me
off-line.  I'm
not the shrinking violet type.     When I
decline to reply, it's
generally either a "religious" issue (e.g. an
emacs vs. vi class of
debate) or because I see the point and can't disagree.

I also encourage anyone else to engage in serious dialog.

> I think everyone is 
> entitled to an opinion, and besides a flamefest every
now and then is 
> also a good thing 

More fun to read than to participate in IMHO, but they can
be
entertaining if the content is original and not carried on
to the point
of tedium.

Regards,
Phil


_______________________________________________
CentOS-devel mailing list
CentOS-develcentos.org
http://lists.centos.org/mailman/listinfo/centos-devel

On Fri, Oct 13, 2006 at 08:40:57AM +0100, Karanbir Singh
enlightened us:
> Heya Phil,
> 
> Phil Schaffner wrote:
> >Will come out of my "lurker" closet to
chime in agreement with Claire.
> >This would seem to be the appropriate list to
discuss the philosophy and
> >implications of development decisions. 
> 
> agreed.
> 
> >I subscribed originally because
> >of testing packages and wanting to provide feedback
through the
> >appropriate channel.  I have stayed because of
interest in where the
> >development is headed and the opportunity to see
what the developers are
> >thinking in a high signal-to-noise environment.
> >
> >I also don't get paid to develop CentOS, and don't
have time to hang out
> >on IRC, but can give a bit back to the community by
providing feedback,
> >bug reports, and occasional help on the users list.
 [Although I find my
> >opinions are sometimes not shared by the
developers, hay, that's what
> >community dialog is all about, and I tend to defer
to their opinions and
> >avoid pointless debates and flame wars.  But I
digress...]
> 
> you know, I actually appreciate people who make noises
when we do stuff 
> that they do and dont agree with. Isnt that what open
source is all 
> about ? The ability to bring in lots of eyes on simple
issues ? We all 
> gain, all around.
> 
> so please dont stop and/or hold back. Flame wars only
happen when one 
> side refuses to accept the others point of view. I
think everyone is 
> entitled to an opinion, and besides a flamefest every
now and then is 
> also a good thing 
> 
> >
> >Please do consider more discussions on issues such
as the yumconf
> >approach on this list.
> 
> yes, we know that we should have made a bit of noise
about this yumconf 
> issue before it went out into the wild. I think we've
all learn't a 
> lesson here.
> 
> 

Just to sort of close out the thread, what has been
determined as The Right
Way (TM) to manage this? Rebuild centos-release?

Matt

-- 
Matt Hyclak
Department of Mathematics 
Department of Social Work
Ohio University
(740) 593-1263
_______________________________________________
CentOS-devel mailing list
CentOS-develcentos.org
http://lists.centos.org/mailman/listinfo/centos-devel

Matt Hyclak wrote:
> 
> Just to sort of close out the thread, what has been
determined as The Right
> Way (TM) to manage this? Rebuild centos-release?


The right-way is whatever way works for you. if you dont
want the 
centos-base.repo to be installed and used, just remove it
and touch it 
back in, so rpm wont overwrite it.


-- 
Karanbir Singh : http://www.karan.org/ :
2522219icq
_______________________________________________
CentOS-devel mailing list
CentOS-develcentos.org
http://lists.centos.org/mailman/listinfo/centos-devel

> Just to sort of close out the thread, what has been
determined as The Right
> Way (TM) to manage this? Rebuild centos-release?

I'd say either rm Centos-Base.repo && touch
Centos-Base.repo, or build
a custom yumconf rpm  for RHEL boxen. I ended up going for
the latter
because I wanted access to centosplus and yum from the
centos repos on
my rhel boxen. Other solutions may work for you.


During times of universal deceit, telling the truth becomes
a revolutionary act.
George Orwell
_______________________________________________
CentOS-devel mailing list
CentOS-develcentos.org
http://lists.centos.org/mailman/listinfo/centos-devel

On Wed, 2006-10-11 at 16:22 -0700, C.M. Connelly wrote:
> "JP" == Jim Perrin <jperringmail.com>
> 
> Jim,
> 
>     JP> This was discussed before implementation, in
the -devel
>     JP> channel on IRC. The full ramifications
weren't felt
>     JP> initially because it worked fine for stock
centos. If
>     JP> you'd like to participate in the
discussions, you're
>     JP> welcome to join the irc channel.
> 
> Like, I suspect, most people using CentOS, I'm not paid
to work on
> the OS, and I don't have time to hang out on IRC on the
off chance
> that someone will bring up something that might
completely screw
> up my systems down the road.

That comment is utterly ridiculous.

You seem to be under the misconception that someone is
getting paid to
work on CentOS.

No one has EVER been paid one thin dime to work on CentOS
... nor has
anyone ever been charged one thin dime to download or use
it.

All the developers of CentOS donate their time and their own
boxes.  We
do get hardware and mirror donations, but either the
hardware is donated
by others OR the developers buy it themselves.  I have 4
main build
boxes to build 3 different arches, 1 of which was donated
the others I
bought to build CentOS. These boxes can not be used for
anytihng else,
as they have to have the build environmnet strictly
controlled to
produce good RPMS for CentOS. The story is much the same for
the other
developers.  Not only do we not get paid, it costs us
something to build
CentOS.

What makes anyone here think that I have 30-50 hours a week
to GIVE to
the CentOS project for free.  How about my $500.00 a month
electric bill
because I am running 4 build servers running 24x7, or my
$99.00 a month
internet bill so I can quickly upload ISOs.

The developers pay their own expenses, they donate their
weekends,
vacations, and off work hours to make CentOS happen, and
most of them
have done it for more than 2 years.

They spend their time "hanging out on IRC on the off
chance that"
someone wants to talk about the way the upstream provider
does their
updates.  They spend time helping people stand up a dhcp
server, set up
TLS with post fix, doing DDNS.  They do this on many CentOS
IRC
channels.

> 
> To be frank, saying, ``Oh, we talked about that on
IRC,'' is about
> as useful as saying, ``Jennie, Bob, and I talked about
it over a
> drink at the Boo Bar.''

If that is not the kind of OS you want, then you can pay
$2500.00 a pop
for a much more professional one.  Oh, but every update
cycle, they
still make major changes.  Look at the new things added this
update
cycle.

> 
> IRC is a great place to work out implementation details
and maybe
> even do some thought experiments to imagine what the
impact might
> be, but mailing lists are a much better place to have
real
> discussions, especially when your changes might impact
people
> outside the tiny circle of people who frequent the IRC
channel.
> 

That might be true and MANY things are worked out devel
list.  MANY,
MANY things are discussed here.  However, the yum
configuration files,
as well as the ones for up2date and any other update
mechanism need to
be in centos-release ... and that is where they are staying.
 Read below
for the facts.

---------------------------------------------------

Facts as to why centos-yumconf was discontinued and why the
configuration files are in centos-relase for CentOS-4:

1.  The upstream package up2date-4.4.69-25.src.rpm (in RHEL
4) has this
in the package:
--------------------------
# in rhel4/fc this file moved to redhat-releases
%if %
%config(noreplace) /etc/sysconfig/rhn/sources
%endif
--------------------------

The upstream package redhat-release-4AS-5.5.src.rpm (the
latest RHEL 4
AS package) has this in the package:

--------------------------
mkdir -p -m 755 $RPM_BUILD_ROOT/etc/sysconfig/rhn
install -m 644 sources $RPM_BUILD_ROOT/etc/sysconfig/rhn
--------------------------

So redhat-release is where the configuration file for
up2date lives.
This was changed upstream at the release of RHEL-4 and is
still the way
it is being done.

2.  The file fedora-release-3-9.src.rpm contains these from
fedora core
3 contains these lines:

-------------------------
mkdir -p -m 755 $RPM_BUILD_ROOT/etc/sysconfig/rhn
mkdir -p -m 755 $RPM_BUILD_ROOT/etc/yum.repos.d
install -m 644 sources
$RPM_BUILD_ROOT/etc/sysconfig/rhn/sources
for file in fedora*repo ; do
  install -m 644 $file $RPM_BUILD_ROOT/etc/yum.repos.d
done
-------------------------

So fedora-release is where the configuration files for
up2date and yum
live.  This is true for every Fedora release after FC3 as
well. (Though
up2date as gone away, the update systems have their
configuration files
in the fedora-release file).

3.  The file redhat-release-4.91Server-1.src.rpm does not
have any .repo
files as there was none produced with the beta ... however
it does have
this code:

-------------------------
mkdir -p -m 755 $RPM_BUILD_ROOT/etc/sysconfig/rhn
install -m 644 sources $RPM_BUILD_ROOT/etc/sysconfig/rhn

#mkdir -p -m 755 $RPM_BUILD_ROOT/etc/yum.repos.d
#for file in redhat*repo ; do
#  install -m 644 $file $RPM_BUILD_ROOT/etc/yum.repos.d
#done
-------------------------

So in the upcoming RHEL5, the sources file still lives in
redhat-release
and I was told that when repo files are distributed, they
will be
distributed via redhat-release.

4.  There is an apt included in CentOS Extras and the
configuration file
was included in that package.  If a newer apt is provided
somewhere else
and if it replaces the old apt, it also erases the
configuration file
for CentOS updates.  That happened to several CentOS users. 
Moving the
configuration file for apt to redhat-release solves that
problem too.

------------------------------------------------------------
---
For all the above reasons, all update configurations belong
in the
centos-release file.  That is how it is consistently being
done and it
is what we are doing.

I apologize for not discussing this issue on this list
before it was
done, however it is still where those files belong.  One of
the things
that is in our goals is to be more like upstream.  It just
makes sense
to do it the same way where ever possible ... I am sorry
that this
affected some power users in an unexpected way, but it is
still where
the files belong.

Thanks,
Johnny Hughes
_______________________________________________
CentOS-devel mailing list
CentOS-develcentos.org
http://lists.centos.org/mailman/listinfo/centos-devel

On Sat, 2006-10-14 at 06:59 -0500, Johnny Hughes wrote:
> On Wed, 2006-10-11 at 16:22 -0700, C.M. Connelly wrote:
> > "JP" == Jim Perrin <jperringmail.com>
> > <snip>

> > Like, I suspect, most people using CentOS, I'm not
paid to work on
> > the OS, and I don't have time to hang out on IRC
on the off chance
> > that someone will bring up something that might
completely screw
> > up my systems down the road.
> 
> That comment is utterly ridiculous.
<snip the response about how we all work for free:
irrelevant>

Your the second (3rd?) responder to miss/ignore the point? I
*guess* you
missed it since you launch into a (emotionally inspired?)
response that
appears defensive, as well as counter-productive, IMO.

If I may be permitted a guess, based on an impartial
reading? Her point
was that her particular situation does not allow her the
luxury of
monitoring the IRC stuff as well as the lists. I have that
luxury and
choose *not* to monitor multiple sources.

Everybody's situation is different. A *consistency* of
process benefits
all. If the discussion occurs on IRC and final decision
occurs there, it
does take some effort to make sure that at least a
notification appears
on the appropriate list(s). If there is not a formal
procedure in place
to assure this occurs, individuals are easily forgiven for
overlooking
that step.

But once that sort of failure is identified, there is no
forgiveness for
the "organization" if they don't expend at least
as much effort in
trying to ensure that the blunder is not repeated as they
did telling
the complainants they were wrong because they could have
done a or b or
c...

Since there are multiple "victims" (based on
posts), it should not be
viewed as an easily-dismissed aberration by one lax user.

As to the movement of the fails, the complaints I recall are
not
entirely that they were moved, but that they were moved
without any
notification apparent to those particular users. Of course
it is not
reasonable for the project to try to anticipate all possible
field-
implemented variations that may be affected by any change.
But that does
not mean, and you do *not* say/imply, that you should just
ignore the
possible damaging effects of changes
*Of_which_you_are_aware*.

As one possible solution that rational (I guess that means
"non-
engineering" types!   beings
might consider is to require a
notification of change to whatever list is deemed most
extensive in
coverage, or even multiple lists,
*when_a_change_has_been_identified_as_having_possible_effect
* in the
user community.

If this is beyond the scope of what CentOS hopes to achieve,
then a
simple say so would also be considered sufficient by all...
I would
*hope*.

MHO
--
Bill


_______________________________________________
CentOS-devel mailing list
CentOS-develcentos.org
http://lists.centos.org/mailman/listinfo/centos-devel